R

B

ARa

b

c

Ud

e

f

a

ARR2AA

KRHPRC

C

1h

Biomedical Signal Processing and Control 10 (2014) 65–78

Contents lists available at ScienceDirect

Biomedical Signal Processing and Control

journa l homepage: www.e lsev ier .com/ locate /bspc

eview

ridging the gap between robotic technology and health care

driano O. Andradea,∗, Adriano A. Pereiraa, Steffen Walterb, Rosimary Almeidac,ui Loureirod, Diego Compagnae, Peter J. Kyberdf

Biomedical Engineering Laboratory, Faculty of Electrical Engineering, Federal University of Uberlândia, Uberlândia, BrazilMedical Faculty, University of Ulm, Ulm, GermanyBiomedical Engineering Program, Alberto Luiz Coimbra Institute – Graduate School and Research in Engineering (COPPE),niversidade Federal do Rio de Janeiro, Rio de Janeiro, BrazilSchool of Science and Technology, Middlesex University, London, UKFaculty of Social Sciences, University of Duisburg-Essen, Duisburg, GermanyInstitute of Biomedical Engineering, University New Brunswick, Fredericton, Canada

r t i c l e i n f o

rticle history:eceived 16 January 2013eceived in revised form2 November 2013ccepted 24 December 2013vailable online 1 February 2014

a b s t r a c t

Although technology and computation power have become more and more present in our daily lives, wehave yet to see the same tendency in robotics applied to health care. In this work we focused on the studyof four distinct applications of robotic technology to health care, named Robotic Assisted Surgery, Roboticsin Rehabilitation, Prosthetics and Companion Robotic Systems. We identified the main roadblocks thatare limiting the progress of such applications by an extensive examination of recent reports. Based onthe limitations of the practical use of current robotic technology for health care we proposed a general

eywords:oboticsealth carerostheticsehabilitationompanion robotic systems

modularization approach for the conception and implementation of specific robotic devices. The mainconclusions of this review are: (i) there is a clear need of the adaptation of robotic technology (closedloop) to the user, so that robotics can be widely accepted and used in the context of heath care; (ii) forall studied robotic technologies cost is still prohibitive and limits their wide use. The reduction of costsinfluences technology acceptability; thus innovation by using cheaper computer systems and sensors isrelevant and should be taken into account in the implementation of robotic systems.

© 2014 Elsevier Ltd. All rights reserved.

ontents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 662. Robotic Assisted Surgery (RAS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

2.1. Existing systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 662.2. Pros and cons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 672.3. Future trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 672.4. Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

3. Robotics in rehabilitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 673.1. Existing systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 683.2. Pros and cons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 683.3. Future trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 683.4. Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

4. Prosthetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 694.1. Existing systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 704.2. Pros and cons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

4.3. Future trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.3.1. Attachment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4.3.2. Appearance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

∗ Corresponding author. Tel.: +55 34 3239 4729.E-mail addresses: aoandrade@feelt.ufu.br, aoandrade@yahoo.com.br (A.O. Andrade).

746-8094/$ – see front matter © 2014 Elsevier Ltd. All rights reserved.ttp://dx.doi.org/10.1016/j.bspc.2013.12.009

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

66 A.O. Andrade et al. / Biomedical Signal Processing and Control 10 (2014) 65–78

4.3.3. Actuation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 714.3.4. Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

4.4. Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 725. Companion Robotic Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

5.1. Existing systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 725.2. Pros and cons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 725.3. Future trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

5.3.1. The role of Requirements Engineering in the conception of CRS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 735.3.2. The need for visionary modularization of CRS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

5.4. Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 746. Discussion and conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

1

C(td

ogtgenc

mstooab

mtApa

2

mti

avoas

2

ribw

. Introduction

In 1917 the term robot was for the first time used by thezech dramatist Karel Capek [1]. In his popular scientific play RURRossum’s Universal Robots), Capek describes intelligent machines,hat although created for serving humans, dominated the world andestroyed humanity [1].

From that time on we have seen a great advance in technol-gy for robot conception and implementation which seems to beoverned by the law of Accelerating Returns [1] that suggests thathe time interval between relevant events becomes shorter as timeoes by, and the return, i.e, the value of technology, increasesxponentially. The advance of technology is accompanied by theecessity of increasing computational resources, e.g., memory andomputational speed, which is governed by Moore’s law.

Although technology and computation power have becomeore and more present in our daily lives, we have yet to see the

ame tendency in robotics applied to health care. We argue thathis is happening because robotic devices that have been devel-ped are still far from obeying one of the most important principlesf Cybernetics, which states that intelligent machines should adaptnd react to their environment so that there is complete interactionetween human and robots.

The aim of this review is to provide the reader with broad infor-ation about the application of robotic technology to health care. In

his context we review four important applications, named Roboticssisted Surgery, Robotics in Rehabilitation, Prosthetics and Com-anion Robotic Systems. We identify the main challenges that thesereas are facing and discuss possible solutions for them.

. Robotic Assisted Surgery (RAS)

Any surgery is intrusive in nature, so there has been a strongotivation to the development of new technologies for reducing

he complications of trauma related to operation by means of min-mal invasive procedures through small orifices.

In this perspective, robotic assisted procedures can be seens important tools that provide flexibility, stability and enhancedision for professionals executing surgeries [2], however as anyther new technology, RAS should be judged on its performancend cost-effectiveness and not only on its technological persua-iveness.

.1. Existing systems

There are many types of robotic technologies being developed in

esearch laboratories and by companies all over the world. Camar-llo et al. [3] suggest a classification of the use of robots in surgeryased on the level of responsibility and involvement the robot hasith the patient during a procedure. In this sense robots that are

solely used as an automated positioning system, such as a patient-mounted robotic platform for CT-scan [4] are considered the mostpassive.

Some robotic positioning systems can be considered activebecause the way they interact with the patient. For instance, theCyberKnife Robotic Radiosurgery System (Accuray – Sunnyvale,CA) is composed of a radiation delivery device, called a linearaccelerator, which is mounted on a robotic arm. This system auto-matically registers the preoperative path by correlating real-timeradiographic images with the preoperative CT images to locate andeliminate the tumor in the patient [3].

In 1994, the Food and Drug Administration (FDA) approved theuse of AESOP (Automated Endoscopic System for Optimal Position-ing) – Computer Motion (Goleta, California). This system is a roboticendoscope holder, which is used to hold and position rigid endo-scopes during minimally invasive surgery. This robotic arm wasdesigned to offer direct control over the laparoscope by means of afoot pedal and later on by voice control [5].

The concept of a master–slave telemanipulation system wasdeveloped in the early 1990s to overcome the issue of dexter-ity in complex procedures, and also with the goal of developingtelesurgery to operate on patients from remote places. Initially,Computer Motion developed the telemanipulator system ZEUSspecifically for cardiac operations. Later on, ZEUS was also usedfor laparoscopic procedures in animal models to verify the fea-sibility and applicability of robotic systems in different surgicalareas, including general surgery, gynecology, urology, and pediatricsurgery [6,5].

The NeuroMate (Integrated Surgical Systems) is the firstUnited States Food and Drug Administration-approved, commer-cially available, image-guided, robotic-assisted system used forstereotactic procedures in neurosurgery [7]. The precision robot“Evolution 1” (U.R.S. Universal Robot Systems, Schwerin, Germany)is a neurosurgical tool that has 7 actuated axes; it is a universalinstrument interface, a mobile pre-positioning system, includ-ing the control computer rack, and the touch operated graphicaluser interface [8]. This system has been used for robot-assistednavigated endoscopic third ventriculostomies in patients withhydrocephalus related to aqueductal stenosis [8].

Hagn et al. [9] discuss the advantages and disadvantages of dedi-cated and versatile surgical robotic systems. The first type of systemis dedicated for specific applications or diseases, whereas the lat-ter can be adapted to a wide range of applications. In this context,the authors proposed the MiroSurge, which is a versatile roboticsystem that can be adapted to multiple surgical domains (e.g., vis-ceral surgery and neurosurgery). The specialization of the system is

obtained by the use of specialised instruments connected to driveunits of the robot.

Currently, the Da Vinci Surgical System (Intuitive SurgicalInc., Sunnyvale, CA, USA) is one of the most successful, general,

A.O. Andrade et al. / Biomedical Signal Processing and Control 10 (2014) 65–78 67

Table 1Pooled estimate from a meta-analysis comparing Robotic-Assisted Radical Prostatectomy (RARP) and Laparoscopic Radical Prostatectomy (LRP) for outcome measures favoursRARP.

Outcome measure Number of studies Sample size Statistical heterogeneity measures I2, P-value Pooled estimate [95% CI]

Operative time (min) 9 1415 89.8%, < 0.00001 WMD −22.79 [−44.36, −1.22Hospital stay (days) 7 1235 76.2%, 0.0003 WMD −0.80 [−1.33, −0.27]Blood loss (mL) 10 1655 90.0%, < 0.00001 WMD −89.52 [−157.54, −21.49]Incidence of transfusion 7 1820 0%, 0.83 RR 0.54 [0.31, 0.94]

Source: extracted from the report of the Canadian Agency for Drugs and Technologies in Health [12]Legend: CI, confidence interval; RR, risk ratio; WMD, weighted mean difference. Pooled estimates are reported as WMD for continuous measures and as RR for dichotomousm

cet(vdhmmpum

2

mte[

rilieta(

da1r

2

dmfSa

ichismhao

easures.

ommercial and studied surgical robot [10], which is used in clinicalnvironments such as a hospital. This is a robotic master–slave sys-em that consists of a remote console where the operating surgeonmaster) directs the robotic surgical arms (slave) via a teleroboticideoscopic link, with the aim of facilitating certain surgical proce-ures. This system allows for enhanced stereoscopic and enlargedigh definition imaging. It has the potential for tremor free preciseovements and it uses intracorporeal articulated instruments withultiple degrees of freedom allowing partially overcoming the

roblem of the fulcrum effect seen with conventional laparoscopysing rigid instruments [2,11]. It also allows gearing down of theotions to make them more precise.

.2. Pros and cons

Despite the growing of evidence for the successful use of RAS,ost studies reported are case series from large centres, therefore

here is lack of conclusive comparative studies. In the current lit-rature mainly short-term follow-up outcome data are available2,11,10,12–15].

Most of the available evidence on efficacy is for robotic assistedadical prostatectomy, which is also the largest current indicationn the world [16]. There is evidence that perioperative blood loss isower than with conventional techniques [2,11,10,12–15] but, tak-ng into account the most updated review [12], evidence for otherxpected advantages, such as reduced incontinence, reduced erec-ile dysfunction or shorter length of hospital stay, is less consistentnd highly dependent on skill and experience of the surgical teamTable 1).

The reviews [2,11,10,12–15] highlighted that the main cost-rivers of the robot-assistance in surgery are the capital acquisition,round D 1.7 million [12], and maintenance that is approximately0% of the acquisition cost, followed by the high costs of limitede-usable surgical instruments.

.3. Future trends

Several factors including establishing adequate access, twoimensional vision, decreased depth perception, restricted instru-ent maneuverability, decreased dexterity and dampened tactile

eedback are all unique limitations that make Robotic Assistedurgery challenging for surgeons trained in traditional openpproaches [17,18].

Camberlin et al. [2] discuss some future development neededn robotic-assisted surgical systems: (i) development of smaller,heaper, faster, and safer devices with improved features such asaptic feedback; (ii) improved instrumentation, such as smarter

nstruments with capabilities to do smart sensing, informing theurgeon about tissue oxygenation, blood flow, molecular infor-

ation and even tumour margin information by intraoperative

istology; (iii) provision of additional help to the surgeon withnatomic overlays incorporating information from other sources,r even offering optical biopsy capabilities; (iv) paradigm shift from

intracorporeal tools attached to an extracorporeal device to entirelyintracorporeal devices: intra-abdominal cameras and intracorpo-real self propelled mobile robots could be used for microsurgeryand other applications such as real-time intra-operative anatomyand histology, or for the delivery of new therapeutic techniquessuch as local phototherapy.

2.4. Recommendations

Over the past fifteen years we have had great advance inrobotic surgery but its use is still limited to a few research centresand hospitals in the world. Researchers and engineers responsi-ble for developing, implementing and assessing this technologyshould focus on critical issues that could bring it to the reality ofpatients. From the literature review it is possible to extract rel-evant messages that could guide researchers in the developmentand improvement of RAS:

• The evidence of the effectiveness, safety, costs and budget impactof robotic surgery should be evaluated in a broad range of proce-dures.

• New systems should minimize training costs and the learningcurve of the user.

• More effort should be dedicated to the implementation of mod-ular and versatile systems.

• Designs should prioritize the reusability of supplies.

3. Robotics in rehabilitation

Rehabilitation incurs considerable costs to health care systemsall over the world. Brain injuries, movement disorders and chronicpain affect hundreds of thousands of people worldwide and havea profound impact in their quality of life. Strokes are the thirdmost common cause of death worldwide after heart disease andcancer (The Stroke Association, UK, 2008), and the most commoncause of acquired physical disability. Tremor on the other hand, isone of the major causes of functional disability and the most com-mon disorder in neurological practice, affecting mostly the elderlypopulation [19–21]. Chronic neurological pain can worsen due toanxiety/stress and other factors. Not only does this affect the indi-vidual’s mental state but also their physical state and rehabilitationoutcome. This in turn poses diverse challenges to health servicesand rehabilitation centres as the individual experiences emotionaldiscomfort in addition to psychological trauma and reduced mobil-ity.

Although repetitive task-oriented movements are the singlemost important variable of motor skill acquisition [22,23] and havea therapeutic gain [24,25], the literature suggests that a patientrecovering from neurological trauma must play an active role in

the rehabilitation process if improvement is to occur. However,as motor learning is not a passive, imprinting process – requir-ing active problem-solving and experience – there needs to beenough capability for the patient to participate [26]. Consequently,

6 l Proc

pgar

3

shgccrdf

rla

dstt(od

3

irtlitntrsTtsscobFttrieTtipcibrmm

8 A.O. Andrade et al. / Biomedical Signa

roviding patients recovering from a brain injury with meaningfuloals (use of familiar objects with well-established sensory cuesnd known semantic properties) can potentially lead to betterecovery.

.1. Existing systems

Physical rehabilitation of brain injuries and strokes is a time con-uming and costly process. Over the past decade several studiesave emerged looking at the use of highly sophisticated technolo-ies, such as robotics and virtual reality to tap into the needs oflinicians and patients [27]. Although traditional physiotherapyoupled with machine mediated therapy can be beneficial in theecovery process, once patients are discharged from hospital it isifficult to exercise correctly the affected limb without the supportrom clinicians.

A number of robotic systems can be found in the literature. Limbehabilitation can be classified into two categories, upper limb andower limb rehabilitation, which can include sub-categories suchs hand, wrist, elbow and shoulder rehabilitation.

A recent review on lower limb robotic rehabilitation [28]escribes the most relevant available technologies. In general theseystems can be grouped according to the rehabilitation principlehey follow [28]: (i) treadmill gait trainers, (ii) foot-plate-based gaitrainers, (iii) overground gait trainers, (iv) stationary gait trainers,v) ankle rehabilitation systems (stationary systems and active footrthoses). The description of a number of upper limb rehabilitationevices can be found in [29,27,30].

.2. Pros and cons

While such technologies can be a valuable tool to facilitatentensive movement practice in a motivating and engaging envi-onment, success of therapy also depends on self-administeredherapy beyond hospital stay. Despite all the technological evo-ution observed over the last decade, one of the main challengess to how best use robotic technology to strengthen the physio-herapist’s skills [31]. Robotic technologies are advanced tools andot a physiotherapist replacement, thus unlikely to assemble allhe skills of a physiotherapist, but exceling at conducting simpleepetitive intensive manual therapies. Hence, once clinical deci-ions are made, these can be considered and executed on the robot.his observation has been clearly demonstrated by some roboticherapy studies that matched the level of assistance and inten-ity between the intervention and control groups [32,33]. In bothtudies comparable improvements between the intervention andontrol groups were observed [32,33]. So, what is the advantagef using robots, if intensive conventional physiotherapy deliveredy human therapists can have similar functional gains on patients?or starters, given the pressures imposed on health systems andhe lack of available therapists, it is unpractical during conventionalherapy to maintain high levels of intensity, as it is possible withobots. One advantage of robotic therapy over conventional therapys that robots allow therapists to take a step back from physicallyngaging in assisting the patient to perform repetitive movements.he robot can provide longer and more intensive repetitive assis-ance thus allowing therapists the opportunity to observe, makenformed decisions on best course of action and manage moreatients. This is perhaps the reason why most of the research to dateoncentrates on the principle of massed practice. Robotic therapys appealing because it can deliver complex therapies that would

e too difficult for therapists to do, for instance provision of preciseepeatable force and haptic feedback coupled with interesting andotivating visual feedback and/or the ability to augment move-ent errors to help correct a movement pattern [34].

essing and Control 10 (2014) 65–78

3.3. Future trends

A recent review [27,35] has concluded that despite mount-ing evidence suggesting robotic therapies are not more likely toimprove patients’ activities of the daily living than any other ther-apy, it has shown great capacity to improve patients’ motor functionand core strength [27,35–37]. The consensus exists that equalimportance should be placed by clinicians and researchers work-ing in the field in establishing guidelines on the study design andassessment and on pushing for more efficient, safer and inex-pensive technologies. To this effect, the International Consortiumon Rehabilitation Robotics and the COST European Network onNeurorehabilitation are undertaking a cross-disciplinary basic andapplied research coordinated efforts to aid the development of new,efficient and patient-tailored robot-assisted therapies. Through theprovision of structured overviews relating to current and emerg-ing robot-assisted therapies to clinicians and therapists, the groupexpects this will increase the availability of effective and standard-ised clinical practice across Europe. It is expected that this initiativewill pave the way for an international effort to increase the num-bers on robotic trials by considering nonclinical sensitive measureswith a common treatment and measurement protocol independentfrom the robot platform.

A hospital or clinical environment might use devices able toretrain a variety of movements over a large percentage of the nor-mal range of human movements. However, an emphasis shouldbe placed on allowing the implementation of therapies resem-bling activities of the daily living such as, picking up a book afterreading at a table top and placing it on a bookshelf. Loureiro andSmith [38] for example propose a multimodal robotic system thatencourages arm and hand movements in addition to stabilisationof the trunk while moving from a sitting to a standing posture ormaintaining a standing pose. The ROBIN system (see Fig. 1) wasdesigned to deliver therapies supporting activities of the daily livingwhilst combining retraining of simple reach and grasp movement(while seating/standing), which could have a meaningful transferto everyday functional gains and higher functional independence.The ROBIN system is based on previous work with the Gentle/Gsystem [39] and following recent results from a reach and graspstudy [40] it is being used to investigate the effects of dynamicgravity compensation in reach and grasp movements during stand-ing or sitting tasks. The authors anticipate that the ROBIN systemwill lead to the development of hybrid control methodologies thatestimate the internal state of the patient through a multimodalapproach combining several sensor information. Emerging evi-dence from functional neuro-imaging suggests that task-orientedsensorimotor participation through daily training of the arm andhand can positively influence stroke recovery [41–43]. Optimi-zing robotic rehabilitation on functional outcome should take intoaccount exposures of the nervous system to real-life activities dur-ing therapy.

There is a need for more effective tools so that neuro-therapiescan be moved away from the therapy gymnasium and into theperson’s home. These tools have the potential to make a largeimpact on the recovery of people following their stroke, as therapywill be available on demand at the convenience and in a familiarenvironment to the patient. As part of this therapy process wecan exploit the dual nature of robotic devices to both assistand measure movement. Rehabilitation robots that are alwaysavailable provide unbiased and consistent therapies coupled withintrinsic measurements can quantify and possibly model therapyprogress. Robotic assistance is particularly useful where patients

lack control of movement and less complex systems based onsensor feedback might be more useful in the home when dexteritycontrol still lacks and there is a need to motivate the patient toengage in the therapy. Although several groups are active in the

A.O. Andrade et al. / Biomedical Signal Processing and Control 10 (2014) 65–78 69

) UsedR

fiir

Naadermt

chossecsttbat

ntdsd[

3

foaups

Fig. 1. Illustration of the ROBIN system. (lefteprinted with permission: [38].

eld of rehabilitation recent developments have concentratedn implementing new devices and systems that will assist in theecovery of the upper limb robotics [44–47].

With the emergence of low-cost gaming consoles such as theintendo Wii and the Microsoft Xbox Kinect, new opportunitiesrise for home-therapy paradigms centred on social interactionsnd values, which could reduce the sense of isolation and otherepression related complications. New approaches combining tel-rehabilitation concepts with collaborative play [48], and simpleobotic sonic interaction [49] have the potential to increase engage-ent and participation of individuals in remote and localized group

herapy.New approaches combining telerehabilitation concepts with

ollaborative play [48], and simple robotic sonic interaction [49]ave the potential to increase engagement and participationf individuals in remote and localized group therapy. A recenttudy combining exploration of a painting aided by haptic andound cues, concluded that group interaction resulted in increasedngagement with the interactive installation and increased exe-ution of movements [40,50]. This work showed the potential foruch interventions for development of analytical skills, imagina-ion, promotion of spatial skills realisation and enhancement ofouch/hearing sensory channels. Although such approaches mighte of value to neurorehabilitation, such concepts of augmentedrtefact installations with technology need to be carefully designedo promote social integration and potential use in public spaces.

While clinical scales can help us to examine the impact in theeuro-recovery process, their coarse nature requires extensive andime-consuming trials, and on top of that they fail to show usetails important for optimizing therapy. Alternative, robot-basedcales offer the potential benefit of new finer measurements andeeper insight into the process of recovery from neurological injury51].

.4. Recommendations

One of the main roadblocks to progress in the field is the needor evidence on the effectiveness of rehabilitation robots. The costf rehabilitation robots is still high when compared to drug-based

nd human-based therapies, which in turn makes wide-scale eval-ation of such therapies difficult. Often studies have been limited toilot evaluations, typically of less than 50 participants, to demon-trate the device’s basic working principles [52].

while seating, (right) used while standing.

A systematic review on the clinical effect of robot-aided ther-apy [36] have concluded that although substantial improvementsin short-term outcomes have been reported, upper limb robotictherapy fails to transfer such gains into higher level functionalindependence. A more recent Cochrane report [37] reviewingelectromechanical-assisted gait training concluded that while evi-dence emerges showing recovery improvements of independentwalking in people following a stroke, the importance of this type ofdevices and how it should be used in clinical practice is still unclear.Suggesting, therefore, that further research should address whichfrequency or duration of walking training might be most effectiveand consequently how long the benefit can last.

The lack of conclusive evidence generated thus far has an impacton the acceptance of robotic therapies in clinical practice. Clinicaloutcomes reported in the literature vary [27] and might be ascribedto variations on patient characteristics, exposure to therapies andintensity. This in turn has implications to the study design andprompts for careful interpretation of the data.

The main recommendations to increase the use and acceptanceof rehabilitation robotics are:

• Effort should be placed on reducing the cost and developingdevices suitable for use after hospital discharge, in unsupervisedenvironments such as the patient’s home.

• Robotic devices need to be designed to be modular, adapt to thepatient individual needs, support therapy paradigms promotingfunctional independence and evaluate therapy progression.

• One of the challenges the field needs to face is how to makecurrent technology accepted and transparent to the user. Thediversity of available studies reporting different results in dif-ferent formats is one of the factors of slow uptake of robot-aidedtherapy. Thus, robotic driven clinical protocols and assessmentmethodologies are needed to harmonize the diverse clinical out-come results being reported in the literature.

4. Prosthetics

Prosthetics is the oldest of the technological rehabilitation sci-ences and for as long as people have been surviving the trauma of

lost limbs, there have been persons motivated to replace some ofthe functions of those limbs by building prostheses. It is not uncom-mon that these innovators have been the persons with amputationsthemselves.

7 l Proc

4

tBfn

badb

mpc

batg

lspt

4

aopu

ltptbtgapmpm[

oIdcw

citkttfSfot

0 A.O. Andrade et al. / Biomedical Signa

.1. Existing systems

A possible classification of artificial limbs may consider the wayhe prosthesis is powered for generating the required movements.ased on this criterion, prosthetic devices can be subdivided into

our classes: passive Prostheses, Body Powered Prostheses, Exter-ally Powered Prostheses and Hybrid Prostheses [53].

Passive Prostheses are completely passive, i.e., they can onlye moved and placed in desired positions through external forces,nd often are used for cosmetic reasons. The use of these types ofevices may have a positive psychological effect on the individualy increasing self-esteem [53].

Body Powered Prostheses are operated so that the force of move-ent of a body part reflects the execution of the movement of the

rosthesis. Movements of the shoulder and other parts of the bodyan be used for controlling the artificial limb [53].

Externally Powered Prostheses are those which are energizedy some external source of energy, e.g., batteries. These prosthesesre usually controlled by the movement of remaining natural struc-ures or by electrical muscle activity provided by a chosen muscleroup [53].

Hybrid Prostheses are those in which the control functions of theimb (e.g., electrical muscle activity) can be combined with a controlystem via harnesses for the functions of the elbow. This type ofrosthesis is known as hybrid prosthesis because it combines twoypes of control in the same prosthesis [53].

.2. Pros and cons

In this section the distinct types of existing prosthetic devicesre discussed in terms of their control, which can be considered onef the most interesting challenges related to prosthetics. Ideally, arosthetic limb should be controlled without any effort from theser, similar to the subconscious control of a natural limb.

There are only two practical sources of power for the prostheticimb: body powered, in which the motion is provided by harnessinghe motion of some other body part and mechanically linking thatart to the prosthesis; or electrically powered using rotary elec-ric motors (permanent magnet or brushless). Other sources haveeen suggested [54–56], but so far they have not been used beyondhe laboratory. The reasons are generally practical: to recharge aas cylinder for a pneumatic arm is possible, but complex to do indomestic environment, or to create hydraulic systems that canrovide enough force without being too heavy. However electricotors are ill suited to provide the stop–start motion required of a

rosthetic limb. Hence the idea from Schultz et al. to use an electricotor in a hand to provide the pressure for an hydraulic system

54].The first prostheses were generally passive devices that relied

n intact parts of the body for their positioning and controlling.n the late 19th century the first body-powered prostheses wereeveloped. This extremely successful design allowed the user toontrol the device such that the movement of a part of the bodyas reflected in the movements of parts of the prosthesis [53].

Despite some modifications, currently, this design remains basi-ally the same in current prothetic limbs. The control mechanisms also the most popular among users. There are many reasons forhis success, according to Doeringer and Hogan [57] some of theey factors are: it results in a relatively inexpensive prosthesis;he final prosthesis is not too heavy; after training, the user beginso use the prosthesis as a natural extension of his body, having,or example, the notion of weight and size of the prosthetic limb.

mit also commented on the much shorter feedback mechanismsor body powered devices [58]. One problem with the majorityf designs of body powered devices is the large inefficiencies inhe mechanisms. The group in Delft is also working towards faster

essing and Control 10 (2014) 65–78

body powered devices that require far lower actuation forces thanpreviously possible [59].

Despite some modifications, currently, this design remains basi-cally the same and the control mechanism is the most popularamong users. The reasons for this success are not well established,but according to Doeringer and Hogan [57] some of the key factorsare: it results in a relatively inexpensive prosthesis; the final pros-thesis is not too heavy; after training, the user begins to use theprosthesis as a natural extension of his body, having, for example,the notion of weight and size of the prosthetic limb.

Kruit and Cool [60] described the main drawbacks of the mech-anism described above: the mechanism of harnesses used topropagate the movements of the body is usually uncomfortable; themovement of the prosthesis requires significantly large forces; thenumber of control inputs is limited and thus the number of degreesof freedom of the prosthesis is also limited. An alternative to thebody-powered control is to employ the myoelectric control, whichuses the electrical activity of muscle contraction as a controllingsignal for prostheses.

While switches and levers are other means to control the limbs,it is generally myoelectric signals that are used to instruct and con-trol powered prosthetic arms [61]. It is also being investigated asthe means to control newer powered lower limb system.

Electromyographic signals are those generated by muscle asthey contract. The resulting signal is noisy and prone to interfer-ence, both electrical and mechanical. While much can be done toameliorate these extraneous signals, what is really being soughtis to measure the intent of the user, which would require connec-tions to the nerves or cortex of the brain. Progress in this directionis being made [62], but it is uncertain how popular such invasivesurgery would be, when it has been developed enough to be use-ful. For example, it is a salutary lesson that the take up for neuralstimulation of muscles for persons with paralysis is low, despitehaving some very compelling advantages over external stimula-tion. This group of patients has a great deal more to gain from theimplants that a prosthesis user would, and yet few are interested.It is probable that for surgical implants to become more popular toeither group, the results of the implant would need to be seen togive users a much greater advantage than they currently do. So far,the number of channels for connection in either direction is verylimited and so control would have to be limited also.

The idea of myoelectric control is not new and came about in1948 [63]. Electromyography has established as the most commonbiopotential employed for controlling artificial limbs, however overthe past 15 years there is ongoing research seeking other formsof control based on more natural controlling strategies, such asthose that employ brain or neuronal activity together with sensoryfeedback [64–66,56].

In the past, myoelectric prostheses employed a type of control-ling called “two-site two-states”, from which a pair of electrodesis placed on two distinct muscles. The contraction of one of thesemuscles produces the opening of the hand. The antagonist muscle isused in the same way to control the closing of the hand. As pointedout by Scott and Parker [63], this approach works in a manner anal-ogous to the human body, i.e., two antagonistic muscles (or groupof muscles) control the movement of a joint.

However, as patients must learn to generate independentcontractions of the muscles, which require a high degree of con-centration, the training can be lengthy, requiring a lot of mentaleffort. There are also some situations in which it is not possibleto find two available groups of muscles, and also the need of con-trolling more than one joint. For these situations other controlling

approaches have been developed. For instance, the “one-site three-states”, from which a little contraction of muscles produces theclosing of the hand, a strong contraction open it, and the lack muscleactivity stops the hand.

l Proc

ctdftaae[

4

adtTtronaaonmtdda

4

temsttbpthlotratl

amFOsms

toiit

A.O. Andrade et al. / Biomedical Signa

Currently there are a number of methods using proportionalontrol based on the electrical muscle activity to control the speed,orque and position of prosthetic joints. However, due to the ran-om nature of the myoelectric signal, errors and inaccuracies occuror various reasons [67]. Moreover, the learning process involved inhe generation of myoelectric patterns must be learned by the user,nd this is a task which requires concentration, regular training andgreat amount of physical effort, making the translation of myo-

lectric activity into commands for a prosthetic limb a challenge68].

.3. Future trends

The designs of the prostheses and the functions they replacere as varied as the persons who have the loss. While a genericesire to replace what is missing is often foremost in the minds ofhe person with a recent loss, work by McLaughlin and Desmond ofrinity College Dublin [69], have shown that the persons’ desires forheir prosthesis change with time. Anecdotally, there is a complexelationship between desire for a like-life replacement and the levelf loss. So that the person with a digital loss may wish something soatural at to be unobtainable, while those with a high level of losst the shoulder, may wish merely to manipulate objects effectivelynd careless of appearance. This creates a need for a wide rangef different solutions using knowledge from materials science toeurology. Since every potential user is an individual, the deviceust be customised to the user’s loss, needs and cognitive ability

o use the limb. Surveys of different patients therefore show theseifferent requirements [70–72]. For this paper future trends will beiscussed taking into account the following relevant requirements:ttachment, appearance, actuation and control.

.3.1. AttachmentConventional prosthetic attachment has been to make a socket

o fit around the residual limb. It needs to fit closely enough to allowasy transfer of forces to and from the prosthesis to the world foranipulation. It cannot be too tight to restrict blood flow and so for

ome persons, depending on the level and form of the loss, addi-ional straps may be needed to hold the limb in place. Historically,he socket was made of metal, and more recently plastics. Now rub-er roll-on sleeves spread the load evenly and have fewer pressureoints between the limb and the socket. However, all of these solu-ions create problems with cooling. Animals use their skin to radiateeat and so help to control their internal temperature. The loss of a

imb creates a loss of a significant percentage of the radiative areaf the body [73]. Adding other materials on the residuum reduceshe area still further, a rubber roll-on liner being a particular bar-ier to heat loss. There are now groups looking at new materials todd to the rubbers in liners to increase heat transfer [73]. Howeverhere has been little work looking at active heat transfer for theimb absent population.

One form of attachment that can circumvent the problem ofsocket which is rigid, hard and likely to impede joint range ofotion, is that of attaching the prosthesis directly to the bone.

irst demonstrated by Per-Ingvar Brannemark [74] attachment bysseointegration is a standard technique for teeth and there are

everal million teeth world wide attached this way. There are otherethods with variations on the theme, but so far no other team has

uch extensive experience [75].An advantage of Osseointegration is that the person can feel

hrough the bone. This so called “Osseoperception” has been

bserved to change the size and shape of the amputee’s phantommage of their residuum. The person with a loss in the leg no longers floating in mid air, but connects to the ground. Osseointegra-ion’s drawback is the same as all surgical interventions, there are

essing and Control 10 (2014) 65–78 71

many persons with a limb absence that will not welcome such aninvasion.

4.3.2. AppearanceThe natural, immediate, response to the loss of a bodily part is

often to want something that closely resembles the missing part.However as time goes on this tendency can change. When the per-son wants function they will often opt for less anthropomorphicand more functional, whether this is a carbon fibre c-shaped bladefor sprinting, or a hook for handling tools. A person with an ampu-tation may adopt different prostheses for different specific tasks(each user is unique with their requirements and needs), but gen-erally the interest is in a device with a pleasing appearance, notnecessarily a human like one. Masahiro Mori proposed the UncannyValley as a manifestation of the acceptability of a human-like object[76]. Based on Freud’s ideas of The Uncanny, this idea suggests thatthe more lifelike an object or robot is the more acceptable it is untilit reaches a point where it becomes “spooky”, when it becomesunacceptable. Beyond this, once it is very close to life-like (suchas a living being) it becomes acceptable again, creating the valleyin a plot of acceptability against appearance. Mori suggested thatprostheses are on the acceptable side of the rift. Anecdotal evidencesuggests this is untrue. Prosthetic users ask for greater and greaterrealism in their prosthesis, but at some point they will reject it as“too spooky”, hence the device has fallen in the valley. Anecdotally,it can be seen that the new high tech limbs attract some users whowish to show off their limbs and not hide them in cosmetic cover-ings. What is clear is that any covering for a prosthesis, designed toprotect the mechanics of the device from dirt and ingress of mois-ture, needs to be resilient, resistant to tearing, while also flexiblenot impeding the motion of the limb. So far, the rubbers developedfor this task fall far short of the target of fulfilling all of the aboveand being cheap enough to be practical. Previously, rigorous stud-ies of the Uncanny Valley has concentrated on the face, but recentwork has show that there is some evidence of an uncanny nature ofmechanical and prosthetic hands and how eerie they are perceivedto be [77].

4.3.3. ActuationTo ensure that the widest range of people can use a limb it must

retain all its drives within the envelope of the missing joint. Soa hand must have all the drives for the fingers within the hand[78,79]. To do otherwise would reduce a small market to an evensmaller one. As the amputation level moves up the arm the numbersof persons reduces considerably [80], hence to restrict the marketto those persons would be ineffective. This is a tighter requirementthan Natural Selection created for the limb, where the muscles forthe hand are in the forearm, and some of those for the shoulderare in the chest. Thus the design of the limb might well reduce thenumber of degrees of freedom to fewer than those lost. For exam-ple, even the most sophisticated prosthetic wrist used by humansubjects in the field does not incorporate radial/ulnar deviation[81].

4.3.4. ControlThis is tied up very closely with the means to actuate the device,

since a body powered limb uses the motions of the body to drivethe prosthesis and the feedback sense of how the device is movingallows for modification of its motions, hence its control.

One of the ways that EMG control of the limb is being enhancedis to train a computer to recognise the patterns of muscle signalsacross the residual limb [82,83] as an arm is moved. One flaw in the

ability to detect signals is that the amputation reduces the numberof possible sites to derive a control signal. Kuiken conceived of theidea of Targeted Muscle Reinnervation (TMR). This is based on theobservation that if a nerve end is placed in a muscle it will remake

7 l Proc

crasptts[

tnyTattdct

tttbtu[ao

4

tu

•

•

•

•

•

•

•

5

cp

likely within the nature of social settings [98]. Therefore the fol-lowing two paragraphs will focus on the pros and cons regardingthe changes in a typical field for the insertion of CRS, namely carefacilities for the elderly.

2 A.O. Andrade et al. / Biomedical Signa

onnections. While for reinnervation of intact limbs the low rate ofeattachment is a disappointment, Kuiken realised if he denervatedmuscle that was not being used (for example the pectoralis in

omeone with a loss at the shoulder), then divided the muscle andlaced a nerve from a muscle that was lower down the arm (such ashe radial nerve), then some control sites would be created. Effec-ively the muscle is amplifying the nervous signal [84,85]. TMR istill an invasive technique and requires much training to be useful86].

Pattern Recognition in general, is an appealing idea, but so farhere have been no genuine clinical demonstrations of the tech-ique reported, although this is likely to change within the nextear or so. An important advance in this field is Prosthesis Guidedraining (PGT) [87], which allows the user to retrain the systemt any time by having the device go through preset motions andhen the user copies those motions. This gets round any idea thathe system is fixed, it can be adapted to the changing of the userue to experience or fatigue. If Pattern Recognition has a practi-al future, it will only be in conjunction with some form of userriggered retraining.

One problem with most of the control formats in use is thathey have been pragmatically developed with little reference tohe way that the natural solutions have evolved. An example is thathe motion of fingers and thumb are different to each other [88],ut when the action of a user was measured using a hand wherehe fingers and thumb moved in the same way, the motions of theser were such to reproduce (or get closer to) the natural motions89]. The new frontier is to understand more fully how control ischieved in the natural hand or leg and adapt the training and usef the devices to better match the way we use the prostheses [90].

.4. Recommendations

The information and discussion presented in this section suggesthat for the advancement of prosthetics and its acceptability bysers a number of barriers should be overcome:

The mental effort required for controlling a prosthetic device isstill high, requiring training and concentration. The current avail-able technology may not be accessible to individuals sufferingfrom even mild cognitive disability.The use of information from nerves and also the brain for con-trolling an artificial limb is still a promise. Future research shouldtry to clearly show the clinical usefulness of these methods.The types of materials used for attachment of the prosthesis tothe body should promote natural heat exchange.The costs of an artificial limb are considerable high, thus the useof cheaper materials in its design is essential for its acceptanceand popularization.New actuators based on smart materials should be investigatedin order to overcome known problems related to electric motorsand hydraulic systems.Although a number of studies show the success of Pattern Recog-nition to the control of prostheses (specially upper limb devices)there is a lack of clinical evidence showing the actual success ofthe method.More understanding of how control is achieved in the naturallimbs is necessary so that customized training protocols can bedevised to the individual.

. Companion Robotic Systems

Companion Robotic Systems (CRS) are robotic systems thatreate close and effective interaction with a human user for theurpose of giving assistance and achieving measurable progress in

essing and Control 10 (2014) 65–78

a number of situations such as convalescence, rehabilitation andlearning [91].

In this context, CRS, which bind together science, engineering,society and health care are relevant and recent robotic technol-ogy capable of offering additional psychophysical aid to individuals.Currently it is possible to find a number of studies focused on thedevelopment of CRS. Some of them aim to develop methods andtechnologies for the construction of cognitive robots, able to evolveand grow their capacities in close interaction with humans in anopen-ended fashion. A practical application of a companion robotcould be the cognitive stimulation and therapy management for theelderly by means of a robotic companion working collaborativelywith a smart home environment [92].

5.1. Existing systems

Currently there are a number of companion robotic systemsbeing developed and studied all over the world. The European Con-sortium so-called Accompany (Acceptable robotiCs COMPanionsfor AgeiNg Years)1 is working towards the development and assess-ment of robotic companion technology facilitating independentliving at home for elderly users [93].

The CompanionAble2 is another initiative to the developmentof a low-cost home robot companion which aims at assisting olderpeople suffering from mild cognitive impairment (MCI) in livingindependently at home in their daily life [94].

Mobiserv3 is a robotic system to support independent living ofolder adults by the provision of Health, Nutrition and Well-Beingservices. Other similar projects are FLORENCE4 and SRS – Multi-Role Shadow Robotic System for Independent Living.5 A descriptionand comparison among many other CRS can be found in [95,96].

5.2. Pros and cons

Even if the most obvious and direct risk of any assistive technol-ogy is the potential of physical harm, several other critical aspectsshould be considered. On the one hand, the risk of physical harmremains of course crucial for the assessment of CRS. On the otherside, CRS could be subsumed mainly to the field of “Socially Assis-tive Robotics” (SAR). Therefore David Feil-Seifer and Maja J. Mataricstated in their paper Socially Assistive Robotics, that “while this[physical harm] is an important risk to examine, SAR is primar-ily concerned with robots that provide assistance through social,rather than physical, interaction” [97] (p. 25). However it is oftenthe case that even in the field of SAR social interaction is based onphysical interaction. Especially if the social interaction is directed toinhabitants of care facilities specialized on patience that are suffer-ing from (severe) dementia. A well known and comparatively widespread robot for these purposes is the seal-looking robot “Paro”.In the end physical harm must be put on a high degree of refer-ence regarding the possible cons of CRS. Despite this consideration,one should bear in mind that physical harm is a quite commonand well assessed risk within the research and development of allkind of robotics, including service robotics and that for this reasonthe main challenge in predicting undesired side effects lies most

1 http://accompanyproject.eu/.2 http://companionable.net/.3 http://www.mobiserv.info/.4 http://www.florence-project.eu/.5 http://www.srs-project.eu.

l Proc

stabbtisc(cWe[oo

sowaa[iattcpsowawvlm[

5

5C

as–

aiibo

vTt[atianam

A.O. Andrade et al. / Biomedical Signa

First the pros are quite easy to identify: a strictly positive con-equence of the integration of service robots in care facilities forhe elderly is the activation as well as improvement of the moodnd disposition of the inhabitants of care facilities: “response toeing held and petted. Experimental results suggest that Paro maye effective for reducing stress in nursing-home residents. In addi-ion, when placed in common areas of nursing homes, it producedncreased social activity among residents. This suggests that SARystems may be useful not just for their direct therapeutic appli-ations but more generally as catalysts for social interaction” [97]p. 25). Another main positive effect lies in the exoneration of theare workers by the assumption of routine tasks by the robots [99].

ada et al. show in their study that a properly designed CRS couldasily lead to an increased amount of human–human interaction100]. Feil-Seifer and Matari outline the positive outcome in termsf emphasizing in the enhancement rather than the replacementf therapeutically goals [97] (p. 28).

At the same time multifaceted cons could be identified andhould be considered. From a strictly philosophical–ethical pointf view as the “most prominent nonphysical risk posed by SAR [asell as CRS in general]” Feil-Seifer and Matari summed up “include

ttachment to the robot, deception about the abilities of the robot,nd influence on the human–human interaction of a robot’s user”97] (p. 27). One reason is the fact that a very similar functional-ty could lead to two divergent outcomes depending on the socialnd living condition of the person in need for care: “However, ifhe robot is used as a replacement or substitute for human care,hen the robot might serve to reduce the amount of human–humanontact. This is especially a concern if the robot is the only thera-eutic influence in a user’s life. For populations that are known touffer from isolation, including the elderly or children with devel-pmental disorders, robots might facilitate further isolation evenhile delivering a therapeutic benefit” [97] (p. 28). Otherwise fromsocial point of view the exploitation of the knowledge of the careorker to optimize the workflow purely from a financial point of

iew by increasing the efficiency of the work process would alsoead to a worse overall situation of a care facility for both the

ain target groups: the care workers as well as the inhabitants101].

.3. Future trends

.3.1. The role of Requirements Engineering in the conception ofRS

Due to the sensibility of the field as a morally charged contextnd its characterization as a highly human driven ‘relational’ workector, its development should be guided by participatory methodsi.e. the development should be driven by user’s needs.

Adopting the appropriate participatory methods is crucial tovoid the risk of first enhancing/improving the care-giving andts setting by strengthening key practices and then fragment-ng/reducing it to a series of interaction loops and tasks that coulde described as interaction algorithms and by that able to be carriedut by CRS.

Requirements Engineering is a necessary condition for an inno-ation process guided by the actual needs of the potential users.he idea is based on a number of works that emphasize the impor-ance of potential users in the context of innovative technologies102,103]. Users play a crucial role in the process of the diffusion ofn innovation [104]. Therefore they should participate already inhe early phase of development until a stable shape of a technolog-cal system is reached [105]. Requirements Engineering provides

functional contribution to the development of innovative tech-ologies. This implies that this method raises the acceptance ofn innovation as the addressed users take part in the develop-ent of the product [106]. Unspecific knowledge about real life

essing and Control 10 (2014) 65–78 73

problems is minimized even before first steps of the technologicalimplementation.

One of the activities of Requirements Engineering is therequirement analysis, which encompasses those tasks that go intodetermining the needs or conditions to meet for a new or alteredsystem. This can be done by either using a user-centred design(based on personas) or participatory design (based on the coop-eration of designers and users) [107].

As the field of Requirements Engineering for CRS is very young,open research methods serve best to sample the needs of therelevant actors. For that reasons qualitative methods offer the pos-sibility to get a nearly comprehensive impression of the field forfuture use of the technologies. Qualitative studies are an appropri-ate way when dealing with new technologies in an unexplored fieldor context. The use of the Grounded Theory approach has the advan-tage to be very open and flexible so that a wide range of contexts andfacts can be observed [108] and taken into account in the followingdevelopment process. In contrast, standardized methods requirea relatively wide knowledge about the subject of the engineering[109]. As this knowledge is not available in innovation processes[110,111] the use of quantitative, standardized methods tends toignore relevant information and contexts. Close to ethnographicmethods, observations and interviews are not only restricted to theformal descriptions of the work but show the real practices estab-lished among the staff as well. This leads to the assumption thatthere is a difference between the description of work processes bythe care management and the everyday work [112,113].

The use of scenarios is an effective way to describe concreteworking practices. In an iterative process users, engineers, anddesigners communicate through the use of scenarios. The aim isto balance the users’ needs and the technological feasibility thatis good enough to satisfy and sustain the users. The scenarioscan be interpreted as “boundary objects” [114]. They are used totranslate the interests and background of a heterogeneous groupof actors and serve the exchange of knowledge in spite of theheterogeneous interests [115]. As an instrument for participatorytechnology development, Scenario-Based Design offers significantpotential for early inclusion of future users [101]. Methods andinstruments aimed at achieving a potential user’s participationand the resulting cooperation of heterogeneous social groups canbe seen as translation tools. Their purpose is to act as translatorsbetween different social fields and the specific knowledge asso-ciated with them. These translation capabilities and participatorymethods should result in the best possible convergence of differentorientations and purposes [116].

The previously mentioned process of collecting data and gath-ering requirements is usually just the first step of an elaborateapplication design process. Beyer and Holtzblatt [117] for exampledeveloped a process called “Contextual Design” consisting of sixsteps starting with data collection by observing and questioningcustomers during work. Afterwards the key issues of each indi-vidual’s work are captured and consolidated over the customerpopulation. Subsequently a high-level story (the vision) about hownew technology will address their work practice is created and thedetails worked out in storyboards. Finally, mock ups of the newsystem are created allowing their testing and modification by theend user until they meet their requirements.

Another method is writing use cases: “The use case describes thesystem’s behavior under various conditions as the system respondsto a request from one of the stakeholders” [118]. Each use casefocuses on describing how to achieve a goal or a task and is basedon one functional requirement. The focus is on interactions and

the use case is often documented in a use case diagram (e.g. in theUnified Modeling Language, UML).

A quite promising method is the already mentioned Scenario-Based Design, a user centered design method which proceeds along

74 A.O. Andrade et al. / Biomedical Signal Proc

informa on from the environment

(e. emperature)iGDM iDPM

Medical Pr

Individual Adapta

iEM

iNSM

iCM

iPM

Medical c

Situa onal needs

tate

Pain iDM

Fig. 2. Suggested companion architecture with an integrated individual AdaptationModule (iAM): (1) the individual Disease Module (iDM), (2) the individual Gen-eral Diagnostic Module (iGDM), (3) the individual Need State Module (iNSM), (4)t((

t[wguwga

op“tibtacg

5

aaflrp

mvpaetc

1

2

he individual Cognitive Module (iCM), (5) the individual Dialog Preference ModuleiDPM), (6) the individual Emotion Module (iEM) and (7) the individual Pain ModuleiPM).

he user centered design process as described in the ISO 9241-210119]. Scenarios are narrative stories consisting of at least one actorith a social, motivational and emotional background and personal

oals as well as objects and tools the actors deal with. The actor issually called “persona” which is a description of a specific personho is a target user of a system being designed, providing demo-

raphic information, needs, preferences, biographical information,nd a photo or illustration [120].

Typically, multiple personas are developed in the early stagesf design that represent the spectrum of the target audience. Theoint of developing personas is to avoid the trap of designing for theaverage” user that does not actually exist, and instead to make surehat the system will work for somebody specific rather than no onen particular. Personas are one piece of a scenario, the other pieceeing a description of how this person would typically interact withhe new system being designed. A scenario describes a sequence ofctions and events which lead to a result. This very vivid descriptionan be used to gather valid and reliable feedback from the userroups [99].

.3.2. The need for visionary modularization of CRSThe main challenge – to fill part of the gap between robotic

nd health care – researchers face when implementing the featuresbove is that they should be personalized, i.e. the system should beexible enough to be adapted to specific characteristics of a wideange of users. In this context calibration should be necessary. Aossible scheme of individualization of CRS is proposed in Fig. 2.

The individual Adaptation Module (iAM) should be the core ele-ent of CRS. It is possible that the user’s individualization occurs

ia an avatar in a private setting, or rather that individualizationrocesses are carried out by a service company with the help of“companionlogue”, which would be considerably time-and cost-xpensive. The complete data set, which is structurally assessed byhe individualization or calibration, is located in the iAM. The iAMonsists of several submodules:

individual Disease Module (iDM): this module receives and pro-cesses information from the medical record of the user. There isa challenge to translate medical records to what a robot need todo for care as the medical record normally contains information

of different type, different quantity and quality, and often trans-lating that to actions for a robot is a very challenging problem;individual General Diagnostic Module (iGDM): general param-eters are processed and recorded in this module. Relevant

essing and Control 10 (2014) 65–78

parameters that should be assessed are age, gender, personality(neuroticism, extroversion), behavioral inhibition and activation,memory, attachment (sure, insecure, clingy). A number of avail-able tools could be used for the quantification of these parametes(see [121]);

3 individual Need State Module (iNSM): need states shall be indi-vidually assessed via a questionnaire or interview: technicalaffinity, taste for music, color, means of transportation, literature,and more should be evaluated;

4 individual Cognitive Module (iCM): cognitive preferences shouldbe assessed via a questionnaire or interview. In this respect,a dynamic adaptation could proceed via structural questionsrelated to priming, e.g., “Have you ever experienced somethinglike that before? Does this seem familiar to you?”; attribution,e.g., “What would you think if a robotic companion would reactin the following manner? To what extent do you see the cause ofthe reaction?”; projection, e.g., “Do you think a companion robotwill be able to solve a personal social problem? Please try to givea reason for your answer”;

5 individual Dialog Preference Module (iDPM): data collection canbe performed via questionnaires. Here the voice (i.e., male vs.female, young vs. old, pertaining to the user, user’s friend, mother,etc.) of the CS will be assessed. Furthermore, phrases and feed-back that are experienced as helpful by the user in certainsituations will be queried;

6 individual Emotion Module (iEM): in this respect, the aim ofcalibration would be to gain a maximum of information aboutthe individual features [122] from all signals (audio, video, psy-chobiological data) collected. This information would be usedto guarantee accurate and robust recognition rates of emotionsin real time. In consequence, processes should be described onhow an individual calibration should proceed. A distinction ismade between a laboratory calibration and field calibration. Itis conceivable that the decision would not be mutually exclu-sive (i.e., an “either-or” decision), but would rather allow thecombination of both approaches. The dimensional perspective ofvalence/arousal/dominance and basic emotions should be usedas a model of emotion;

7 individual Pain Module (iPM): in this module there should beincluded individual feature selection of audio, video and psy-chobiological signals for the quantification of pain.

While it is highly probable that several modules will be needed,the options presented here should only represent priorities foran individualization and calibration. In prospect, it can already beshown that the quality of CRS will be dependent on the complexityof the modules and their networks. For example, the applicationof companion feature to serve drinks at the bedside requires alower number of modules relative to its implementation (i.e., sup-port) than in the medical treatment of cancer. However, CRS shouldtend to include a maximum of user information. Relative to modulenetworking, it appears relevant to discern the weight/priority themodules will carry.

5.4. Recommendations

The following features are essential for the development of CRS[121,123,122]:

1 Sensitivity to the user’s emotions and pain. Because of the impor-tance of emotions in setting priorities, decision-making and

action control, contextual aspects, emotional processes and painshould be captured by the companion robot.

2 Sensitivity to the environment and situations. In other words,CRS should reliably capture contextual and environmental

A.O. Andrade et al. / Biomedical Signal Proc

gyolonhce

T ci t

obo R

Health

Care

LE

AR

NIN

G/T

RA

ININ

G

CU

ST

OM

IZA

TIO

N

CL

INIC

AL E

VID

EN

CE

MO

DU

LA

RIZ

AT

ION

TE

CH

NIC

AL B

AR

RIE

RS

CO

ST

AC

CE

PTA

BIL

ITY

Fb

3

4

5

6

cvaSRhta

movptcnlua

tmtbpit

ig. 3. Key identified variables that should be taken into account in the process ofridging the gap between robotic technology and health care.

parameters, such as time (year, month, day of week, time of day),space (location) and temperature.Verbal interaction in a user-adaptive manner. In this respect,robust speech recognition and speech synthesis is required. Theuser’s emotional state and contextual information should influ-ence the selection of the dialogue strategy and increase theusability of a companion robot.Support to the user in the decision making process. Simply put, acompanion system should be able to provide solutions for com-plex tasks, either completely independently or in cooperationwith the user, while proceeding in a goal-oriented and com-prehensible manner, as well as to provide decision support andrecommendations for action in dialogue with the user [124].Empathic interaction for companions. “The envisaged relation-ship of the user with the robot is that of a co-learner “robot anduser providing mutual assistance for the user not to be dominatedby the technology, but to be empowered, physically, cognitivelyand socially” [93].

. Discussion and conclusions

Bridging the gap between robotic technology and health carean be seen as a complex process governed by a number ofariables. The purpose of this paper was to identify these vari-bles in the context of specific areas (i.e., Robotic Assistedurgery, Robotics in Rehabilitation, Prosthetics and Companionobotic Systems) of great interest for the health sector. Fig. 3ighlights the identified variables: clinical evidence, customiza-ion, learning/training, modularization, technical barriers, cost andcceptability.

Clinical evidence has to do with the use of evidence basededicine, which is “the conscientious, explicit, and judicious use

f current best evidence in making decisions about the care of indi-idual patients” [125]. From our literature review [126–128] it isossible to generalize the conclusion that there is lack of prospec-ive, controlled and randomized clinical trials to provide a solidlinical evidence to the actual benefits of the use of robotic tech-ology to the patients. Future studies should be focused on the

ong-term outcomes of the technology. However, universal eval-ation criteria for various devices and control strategies should bedopted [127].

A relevant issue is that measuring the success of the use ofechnology is an essential part of any development, especially in

edicine and rehabilitation, and for all investigated applicationshis is still an open question. Any claims of real benefits can only

e substantiated by controlled comparative studies directly com-aring techniques to relevant conventional approaches. Gathering

nformation about the performance of technology in real life, byhe prospective registration of data on patient characteristics and

essing and Control 10 (2014) 65–78 75

follow-up on outcomes in centres and teams that perform a suffi-cient number of these interventions is needed to gather additionalmeaningful experience with the performance of technology in dailypractice.

Customization is related to the adaptation of the robotic tech-nology to clinicians and patients. For instance, the dynamics ofthe manipulator of a robotic-assisted surgery device can affect themovements of the operator, in comparison with free-space move-ments [129]. A possible tailored solution for such drawback couldbe the development of tools to enhance the visualization of the sur-gical field [130]. Our main conclusion, which is in the line with otherstudies [131], is that current robotic systems have limited cus-tomization capability and require the clinician to assess progressand adapt procedures accordingly. Based on the limitations of thepractical use of current robotic technology for health care we pro-posed a general modularization approach for the conception andimplementation of specific robotic devices. The system should havean individual adaptation module and many specific submodulesrelated for the following aspects: disease, diagnostics, needs, cog-nition, dialog preferences, emotion and pain.

For the investigated applications there is a common messagethat the technology needs to be adapted to the users more thanto get the users to adapt to it. Adaptation of the technology tothe actual needs of the user is a major roadblock that shouldbe overcome in theses areas. In this context, tools of Require-ments Engineering should be used for formalization and proposeof methodology for the transformation of requirements into formalmodels of robotic tasks [132].

Learning and training are two important factors that influenceon the use and disuse of a technology. Some studies point out thatlittle is known about the learning curve of robotic surgery for sur-geons in training [133]. Furthermore, there is evidence that there isa degradation of skills that occurs during periods of robotic surgicalinactivity in newly trained surgeons [134]. The relevance for train-ing patients to interact with upper-limb prosthetics is discussed in[135]. Essentially, training methods have to be developed in a waythat functionality and usability in everyday life will improve [135].

Modularization allows for the use of modules in the concep-tion of specific designs. It is a desirable feature in robotics appliedto health care because it can promote cost reduction, flexibility,augmentation and exclusion of unnecessary functionalities. Modu-larity has not been investigated in any significant way in thedevelopment of the investigated areas, although it is possible tofind some attempts in the literature [136,137].

Technical barriers are specific for each investigated area andmany of them have already been highlighted in other sectionsof this paper. However, the improvement of user interface is acommon requirement. For instance, lack of proper feedback andvisualization to clinicians and patients is a major concern [138].

For all studied robotic technologies cost is still prohibitive andlimits their wide use. The reduction of costs influences technologyacceptability, thus innovation by using cheaper computer systemsand sensors are relevant and should be taken into account inthe implementation of robotic systems. Furthermore, uncertaintyremains about the cost-effectiveness of robotic technology com-pared with traditional approaches [139].

The use of robots in healthcare is a relatively new conceptand the public’s perception and acceptance is not well understood[140]. In a recent study regarding the investigation of attitudesand reactions to a healthcare robot [140] the authors concludedthat participants saw many benefits and applications for healthcarerobots, including simple medical procedures and physical assis-

tance, but had some concerns about reliability, safety, and the lossof personal care. In this context patients should be informed beforethe operation about the possibility of converting their procedure tolaparoscopic or open due to robotic malfunction [141].

7 l Proc

rtufa

A

CBnlfFiB

R

6 A.O. Andrade et al. / Biomedical Signa

Despite implicit or even explicit claims of the superiority ofobotic systems for health care, when compared to more tradi-ional methods, the clear advantage of these systems are currentlynproven and are highly dependent on the skills of the users, there-ore, the success of such technologies are still heavily dependent ondequate training and experience.

cknowledgements

The authors would like to thank the Brazilian agencies CNPq,APES and FAPEMIG for supporting research carried out at theiomedical Engineering Laboratory of the Faculty of Electrical Engi-eering, Federal University of Uberlândia. The Transregional Col-

aborative Research Centre SFB/TRR 62 “Companion-Technologyor Cognitive Technical Systems” funded by the German Researchoundation (DFG). This review was also supported by the Pan Amer-can Health Organization (PAHO) (Project DECIT-MS Carta acordoR/LOA/1100001.001 – OPAS-BR).

eferences

[1] R. Kurzweil, The Age of Spiritual Machines: How We Will Live, Work andThink in the New Age of Intelligent Machines, Orion Business, London, 1999.

[2] C. Camberlin, A. Senn, M. Leys, C. De Laet, Robot-assisted surgery healthtechnology assessment Health Services Research (HSR), in: Tech. Rep. 104C(D/2009/10.273/09), Belgian, Health Care Knowledge Centre (KCE), BrusselsBelgium, 2009, February.

[3] D.B. Camarillo, T.M. Krummel, J.J.K. Salisbury, Robotic technology in surgerypast, present, and future, Am. J. Surg. 188 (4A Suppl.) (2004) 2S–15S.

[4] B. Maurin, B. Bayle, O. Piccin, J. Gangloff, M. de Mathelin, C. Doignon, P. Zanne,A. Gangi, A patient-mounted robotic platform for CT-scan guided procedures,IEEE Trans. Biomed. Eng. 55 (10) (2008) 2417–2425.

[5] J. Marescaux, F. Rubino, The ZEUS robotic system: experimental and clinicalapplications, Surg. Clin. North Am. 83 (6) (2003) 1305–1315.

[6] M. Eto, S. Naito, Robotic manipulators in cardiac surgery: the computer-assisted surgical system ZEUS, Mini. Invas. Ther. Allied Technol. 10 (6) (2001)275–281.

[7] N. Nathoo, M.C. Cavusoglu, M.A. Vogelbaum, G.H. Barnett, In touch withrobotics: neurosurgery for the future, Neurosurgery 56 (3) (2005) 421–431.

[8] M. Zimmermann, R. Krishnan, A. Raabe, V. Seifert, Robot-assisted navigatedendoscopic ventriculostomy: implementation of a new technology and firstclinical results, Acta Neurochir. (Wien) 146 (7) (2004) 697–704.

[9] U. Hagn, R. Konietschke, A. Tobergte, M. Nickl, S. Jorg, B. Kubler, G. Passig, M.Groger, F. Frohlich, U. Seibold, L. Le-Tien, A. Albu-Schaffer, A. Nothhelfer, F.Hacker, M. Grebenstein, G. Hirzinger, DLR MiroSurge: a versatile system forresearch in endoscopic telesurgery, Int. J. Comput. Assist. Radiol. Surg. 5 (2)(2010) 183–193.

[10] R. Tooher, C. Pham, The da Vinci surgical robotics system: technologyoverview, in: Tech. Re ASERNIP-S Report No. 45, Australian Safety and Effi-cacy Register of New Interventional Procedures-Surgical, Adelaide, SouthAustralia, 2004, July.

[11] P. Thavaneswaran, Robotic-assisted surgery for urological, cardiac and gynae-cological procedures, in: Tech. Rep. ASERNIP-S Report No. 75, AustralianSafety and Efficacy Register of New Interventional Procedures-Surgical, Ade-laide, South Australia, 2009, May.

[12] C. Ho, E. Tsakonas, K. Tran, K. Cimon, M. Severn, M. Mierzwinski-Urban, J.Corcos, S. Pautler, Robot – assisted surgery compared with open surgery andlaparoscopic surgery: clinical effectiveness and economic analyses, in: Tech.Rep. CADTH No. 137, Canadian Agency for Drugs and Technologies in Health,Ottawa, Canada, 2011, September.

[13] S.M. Martínez, M.R. Goyanes, M.G. Rodríguez, J.A.B. Amaro, Efectividad,seguridad e indicaciones del equipo quirúrgico Da Vinci, Plan de Calidad parael Sistema Nacional de Salud del Ministerio de Sanidad y Política Social Edi-tion, Informes de Evaluación de Tecnologías Sanitarias, Agencia Laín Entralgo,Madri, Spain, 2009.

[14] A.L. Méndez, R.V. Portero, Cirugía robótica mediante el sistema de telema-nipulación robótica da Vinci en la prostatectomía, in: Informes de Evaluaciónde Tecnologías Sanitarias, Agencia de Evaluación de Tecnologías Sanitarias deAndalucía, Sevilla, Spain, 2010.

[15] M.A. Secretariat, Robotic-assisted minimally invasive surgery for gynecologicand urologic oncology: an evidence-based analysis, Ontario Health Technol.Assess. Ser. (Internet) 10 (27) (2010) 1–118.

[16] C. Ho, E. Tsakonas, K. Tran, K. Cimon, M. Severn, M. Mierzwinski-Urban, J.

Corcos, Pautler, Robot-assisted surgery compared with open surgery andlaparoscopic surgery, CADTH Technol. Overviews 2 (2) (2012) e2203.

[17] S.H. Baik, Robot Surgery, InTech, Rijeka, Croatia, 2010.[18] D.M. Herron, M. Marohn, S.-M.R.S. Consensus, A consensus document on

robotic surgery, Surg. Endosc. Other Intervent. Tech. 22 (2) (2008) 313–325.

essing and Control 10 (2014) 65–78

[19] M. Manto, M. Topping, M. Soede, J. Sanchez-Lacuesta, W. Harwin, J. Pons,J. Williams, S. Skaarup, L. Normie, Dynamically responsive intervention fortremor suppression, IEEE Eng. Med. Biol. Mag. 22 (3) (2003) 120–132.

[20] M.F. Almeida, G.L. Cavalheiro, A.A. Pereira, A.O. Andrade, Investigation of age-related changes in physiological kinetic tremor, Ann. Biomed. Eng. 38 (11)(2010) 3423–3439.

[21] P.H. Mansur, L.K. Cury, A.O. Andrade, A.A. Pereira, G.A. Miotto, A.B. Soares, E.L.Naves, A review on techniques for tremor recording and quantification, Crit.Rev. Biomed. Eng. 35 (5) (2007) 343–362.

[22] R. Schmidt, Motor Learning Principles for Physical Therapy, Foundation forPhysical Therapy, Alexandria, VA, 1991, pp. 49–63.

[23] C. Winstein, Knowledge of results and motor learning – implications for phys-ical therapy, Phys. Ther. 71 (1991) 140–149.

[24] C. Butefisch, H. Hummelsheim, P. Denzler, K.H. Mauritz, Repetitive trainingof isolated movements improves the outcome of motor rehabilitation of thecentrally paretic hand, J. Neurol. Sci. 130 (1) (1995) 59–68.

[25] S. Hesse, C. Bertelt, M. Jahnke, A. Schaffrin, P. Baake, M. Malezic, K. Mauritz,Treadmill training with partial body weight support compared with physio-therapy in non-ambulatory hemiparetic patients, Arch. Phys. Med. Rehabil.75 (10) (1995) 1087–1093.

[26] R. Held, A. Hein, Movement-produced stimulation in the development ofvisually guided behavior, J. Comp. Physiol. Psychol. 56 (5) (1963) 872–876.

[27] R.C.V. Loureiro, W.S. Harwin, K. Nagai, M. Johnson, Advances in upper limbstroke rehabilitation: a technology push, Med. Biol. Eng. Comput. 49 (1) (2011)1103–1118.

[28] I. Díaz, J.J. Gil, E. Sánchez, Lower-limb robotic rehabilitation: literature reviewand challenges, J. Robot. 2011 (2011) 1–11.

[29] S. Masiero, M. Armani, G. Rosati, Upper-limb robot-assisted therapy inrehabilitation of acute stroke patients: focused review and results of newrandomized controlled trial, J. Rehabil. Res. Dev. 48 (4) (2011) 355–366.

[30] A.E.Q. van Delden, C.E. Peper, G. Kwakkel, P.J. Beek, A systematic review ofbilateral upper limb training devices for poststroke rehabilitation, Stroke Res.Treat. 2012 (2012) 1–17.

[31] W.S. Harwin, J.L. Patton, V.R. Edgerton, Challenges and opportunities forrobot-mediated neurorehabilitation, Proc. IEEE 94 (9) (2006) 1717–1726.

[32] L.E. Kahn, M.L. Zygman, W.Z. Rymer, D.J. Reinkensmeyer, Effect ofrobot-assisted and unassisted exercise on functional reaching in chronichemiparesis, in: Proceedings of the 23rd Annual International Conference ofthe IEEE, vol. 2, 2001, pp. 1344–1347.

[33] A.C. Lo, P.D. Guarino, L.G. Richards, J.K. Haselkorn, G.F. Wittenberg, D.G. Feder-man, R.J. Ringer, T.H. Wagner, H.I. Krebs, B.T. Volpe, J.C.T. Bever, D.M. Bravata,P.W. Duncan, B.H. Corn, A.D. Maffucci, S.E. Nadeau, S.S. Conroy, J.M. Pow-ell, G.D. Huang, P. Peduzzi, Robot-assisted therapy for long-term upper-limbimpairment after stroke, N. Engl. J. Med. 362 (19) (2010) 1772–1783.

[34] J.L. Patton, M.E. Stoykov, M. Kovic, F.A. Mussa-Ivaldi, Evaluation of robotictraining forces that either enhance or reduce error in chronic hemipareticstroke survivors, Exp. Brain Res. 168 (3) (2006) 368–383.

[35] S. Carda, M. Invernizzi, A. Baricich, C. Comi, A. Croquelois, C. Cisari, Robotic gaittraining is not superior to conventional treadmill training in Parkinson dis-ease: a single-blind randomized controlled trial, Neurorehabil. Neural Repair26 (9) (2012) 1027–1034.

[36] G.B. Prange, M.J. Jannink, C.G. Groothuis-Oudshoorn, H.J. Hermens, M.J. Ijzer-man, Systematic review of the effect of robot-aided therapy on recovery ofthe hemiparetic arm after stroke, J. Rehabil. Res. Dev. 43 (2) (2006) 171–184.

[37] J. Mehrholz, B. Elsner, C. Werner, J. Kugler, M. Pohl, Electromechanical-assisted training for walking after stroke, Cochrane Database Syst. Rev. 7(2013) CD006185.

[38] R. Loureiro, T. Smith, Design of the ROBIN system: whole-arm multi-modalsensorimotor environment for the rehabilitation of brain injuries while sit-ting or standing, in: Proceedings of the IEEE International Conference onRehabilitation Robotics (ICORR), IEEE, 2011, pp. 1–6.

[39] R.C.V. Loureiro, W.S. Harwin, Reach and grasp therapy: design and controlof a 9-DOF robotic neuro-rehabilitation system, in: IEEE 10th InternationalConference on Rehabilitation Robotics (ICORR 2007), 2007, pp. 757–763.

[40] R. Loureiro, W. Harwin, R. Lamperd, C. Collin, Evaluation of reach and grasprobot-assisted therapy suggests similar functional recovery patterns on prox-imal and distal arm segments in sub-acute Hemiplegia, IEEE Trans. NeuralSyst. Rehabil. Eng. (2013) 1.

[41] J. Liepert, I. Uhde, S. Graf, O. Leidner, C. Weiller, Motor cortex plasticity duringforced-use therapy in stroke patients: a preliminary study, J. Neurol. 248 (4)(2001) 315–321.

[42] S.H. Jang, Y.H. Kim, S.H. Cho, J.H. Lee, J.W. Park, Y.H. Kwon, Cortical reorganiza-tion induced by task-oriented training in chronic hemiplegic stroke patients,Neuroreport 14 (1) (2003) 137–141.

[43] G.F. Wittenberg, R. Chen, K. Ishii, K.O. Bushara, S. Eckloff, E. Croarkin, E. Taub,L.H. Gerber, M. Hallett, L.G. Cohen, Constraint-induced therapy in stroke:magnetic-stimulation motor maps and cerebral activation, Neurorehabil.Neural Repair 17 (1) (2003) 48–57.

[44] S.E. Fasoli, H.I. Krebs, J. Stein, W.R. Frontera, R. Hughes, N. Hogan, Robotictherapy for chronic motor impairments after stroke: follow-up results, Arch.Phys. Med. Rehabil. 85 (7) (2004) 1106–1111.

[45] P.S. Lum, C.G. Burgar, M. Van Der Loos, P.C. Shor, M. Majmundar, R. Yap,MIME robotic device for upper-limb neurorehabilitation in subacute strokesubjects: a follow-up study, J. Rehabil. Res. Dev. 43 (5) (2006) 631–642.

[46] D. Reinkensmeyer, S. Housman, If I can’t do it once, why do it a hundred times?Connecting volition to movement success in a virtual environment motivates

l Proc

A.O. Andrade et al. / Biomedical Signa

people to exercise the arm after stroke, in: Proceedings of the IEEE VirtualRehabilitation, Venice, Italy, 2007, pp. 44–48.

[47] S. Masiero, A. Celia, G. Rosati, M. Armani, Robotic-assisted rehabilitation ofthe upper limb after acute stroke, Arch. Phys. Med. Rehabil. 88 (2) (2007)142–149.

[48] M.J. Johnson, R.C.V. Loureiro, W.S. Harwin, Collaborative tele-rehabilitationand robot-mediated therapy for stroke rehabilitation at home or clinic, Intell.Serv. Robot. 1 (2) (2008) 109–121.

[49] H.H. Le, M.J. Loomes, R.C.V. Loureiro, Mapping arm movements to roboticsonic interaction promote group dynamics and increase engagement at a task,in: Proceedings of the International Conference on Neurorehabilitation, vol.1, Toledo, Spain, 2012, pp. 843–846.

[50] H. Le, R. Loureiro, F. Dussopt, N. Phillips, A. Zivanovic, M. Loomes, A hapticallyenhanced painting as a tool for neurorehabilitation, in: IEEE InternationalConference on Rehabilitation Robotics 2013, 2013, pp. 24–25.

[51] H.I. Krebs, B.T. Volpe, M.L. Aisen, N. Hogan, Increasing productivity and qualityof care: robot-aided neuro-rehabilitation, J. Rehabil. Res. Dev. 37 (6) (2000)639–652.

[52] S. Balasubramanian, J. Klein, E. Burdet, Robot-assisted rehabilitation of handfunction, Curr. Opin. Neurol. 23 (6) (2010) 661–670.

[53] A. Barbosa, E.A. Lamounier Junior, A.O. Andrade, A. Cardoso, Virtual and Aug-mented Reality: A New Approach to Aid Users of Myoelectric Prostheses,InTech, 2012, pp. 409–425 (Chapter 17).

[54] S. Schulz, C. Pylatiuk, M. Reischl, J. Martin, R. Mikut, G. Bretthauer, Ahydraulically driven multifunctional prosthetic hand, Robotica 23 (3) (2005)293–299.

[55] D. Simpson, G. Kenworthy, The design of a complete arm prosthesis, Biomed.Eng. (N.Y.) 8 (2) (1973) 56–59.

[56] B. Peerdeman, D. Boere, H. Witteveen, R. Huis in ‘tVeld, H. Hermens, S. Strami-gioli, H. Rietman, P. Veltink, S. Misra, Myoelectric forearm prostheses: stateof the art from a user-centered perspective, J. Rehabil. Res. Develop. 48 (6)(2011) 719–737.

[57] J.A. Doringer, N. Hogan, Performance of above elbow body-powered prosthe-ses in visually guided unconstrained motion tasks, IEEE Trans. Biomed. Eng.42 (6) (1995) 621–631.

[58] G. Smit, Natural Grasping Design and Evaluation of a Voluntary Closing Adap-tive Hand Prosthesis, Technical University of Delft, Delft, Netherlands, 2013(Ph.D. thesis).

[59] G. Smit, D.H. Plettenburg, Efficiency of voluntary closing hand and hook pros-theses, Prosthet. Orthot. Int. 34 (4) (2010) 411–427.

[60] J. Kruit, J.C. Cool, Body-powered hand prosthesis with low operating powerfor children, J. Med. Eng. Technol. 13 (1–2) (1989) 129–133.

[61] A. Muzumdar, Powered Upper Limb Prostheses: Control, Implementation andClinical Application, Springer, Berlin, Germany, 2003.

[62] P.M. Rossini, S. Micera, A. Benvenuto, J. Carpaneto, G. Cavallo, L. Citi, C. Cipri-ani, L. Denaro, V. Denaro, G. Di Pino, F. Ferreri, E. Guglielmelli, K.P. Hoffmann,S. Raspopovic, J. Rigosa, L. Rossini, M. Tombini, P. Dario, Double nerve intra-neural interface implant on a human amputee for robotic hand control, Clin.Neurophysiol. 121 (5) (2010) 777–783.

[63] P.A. Parker, R.N. Scott, Myoeletric prosthesis: state of the art, J. Med. Eng. Tech.12 (4) (1988) 143–151.

[64] J.E. Paciga, P.D. Richard, R.N. Scott, Error rate in five-state myoelectric controlsystems, Med. Biol. Eng. Comput. 18 (3) (1980) 287–290.

[65] P.E. Patterson, J.A. Katz, Design and evoluation of a sensory feedback systemsthat provides grasping pressure in a myoletric hand, J. Rehabil. Res. Dev. 29(1) (1992) 1–8.

[66] T.H. Geake, Advanced feedback system for myoelectrically controlled pros-theses, School of Computer Science and Electronic Systems, KingstonUniversity, UK, 1994 (Ph.D. thesis).

[67] L.A. Hoff, Errors in frequency parameters of EMG power spectra, IEEE Trans.Biomed. Eng. 38 (11) (1991) 1077–1088.

[68] P.A.O. SCOTT, E.L. Morin, R.N. Scott, Myoletric signal characteristics from mus-cles in residual upper limbs, IEEE Trans. Rehab. Eng. 2 (4) (1994) 266–270.

[69] P. Gallagher, D. Desmond, M. MacLachlan, Psychoprosthetics, Springer, Lon-don, 2008.

[70] D.J. Atkins, D.C.Y. Heard, W.H. Donovan, Epidemiologic overview of individ-uals with upper-limb loss and their reported research priorities, J. Prosthet.Orthot. 8 (1) (1996) 2–11.

[71] E. Biddiss, T. Chau, Upper limb prosthesis use and abandonment: a survey ofthe last 25 years, Prosthet. Orthot. Int. 31 (3) (2007) 236–257.

[72] P.J. Kyberd, C. Wartenberg, L. Sandsjö, S. Jönsson, D. Gow, J. Frid, C. Almström,L. Sperling, Survey of upper-extremity prosthesis users in Sweden and theUnited Kingdom, J. Prosthet. Orthot. 19 (2) (2007) 55–62.

[73] T. Bertels, T. Kettwig, Breathable liner for transradial prostheses, in:Proceedings of the Myoelectric Controls Symposium (MEC’11), University ofNew Brunswick, Fredericton, Canada, 2011.

[74] K. Hagberg, E. Häggström, S. Jönsson, B. Rydevik, R. Brånemark, Osseopercep-tion and Osseointegrated Prosthetic Limbs, Springer-Verlag London Limited,London, 2008, pp. 131–140, Ch. 10.

[75] K. Hagberg, R. Brånemark, One hundred patients treated with osseointegratedtransfemoral amputation prostheses: rehabilitation perspective, J. Rehabil.

Res. Develop. 46 (3) (2009) 331–344.

[76] M. Mori, The uncanny valley, IEEE Robot. Automat. Mag. 7 (2012) 98–100.[77] E. Poliakoff, N. Beach, R. Best, T. Howard, E. Gowen, Can looking at a hand

make your skin crawl? Peering into the uncanny valley for hands, Perception42 (9) (2013) 998–1000.

essing and Control 10 (2014) 65–78 77

[78] P.J. Kyberd, C. Light, P.H. Chappell, J.M. Nightingale, D. Whatley, M. Evans, Thedesign of anthropomorphic prosthetic hands: a study of the SouthamptonHand, Robotica 19 (06) (2001) 593–600.

[79] P.J. Kyberd, A. Clawson, B. Jones, The use of underactuation in prostheticgrasping, Mech. Sci. 2 (1) (2011) 27–32.

[80] P.J. Kyberd, W. Hill, Survey of upper limb prosthesis users in Sweden, theUnited Kingdom and Canada, Prosthet. Orthot. Int. 35 (2) (2011) 234–241.

[81] K. De Laurentis, S. Phillips, Design of a powered two-DOF prosthetic wrist,in: Proceedings of the 12th World Congress of the International Society ofProsthetics and Orthotics, 2007, p. 524.

[82] A. Fougner, E. Scheme, A.D. Chan, K. Englehart, O. Stavdahl, Resolving the limbposition effect in myoelectric pattern recognition, Trans. Neural Syst. Rehabil.Eng. 19 (6) (2011) 644–651.

[83] M. Ortiz-Catalan, R. Branemark, B. Hakansson, J. Delbeke, On the viability ofimplantable electrodes for the natural control of artificial limbs: review anddiscussion, Biomed. Eng. 11 (2012) 33.

[84] T.A. Kuiken, G.A. Dumanian, R.D. Lipschutz, L.A. Miller, K.A. Stubblefield, Theuse of targeted muscle reinnervation for improved myoelectric prosthesiscontrol in a bilateral shoulder disarticulation amputee, Prosthet. Orthot. Int.28 (3) (2004) 245–253.

[85] T.A. Kuiken, L.A. Miller, R.D. Lipschutz, B.A. Lock, K. Stubblefield, P.D. Marasco,P. Zhou, G.A. Dumanian, Targeted reinnervation for enhanced prosthetic armfunction in a woman with a proximal amputation: a case study, Lancet 369(9559) (2007) 371–380.

[86] K. Stubblefield, S.B. Finucane, L.A. Miller, B.A. Lock, Training individuals touse pattern recognition to control an upper limb prosthesis, in: MyoElec-tric Controls/Powered Prosthetics Symposium (MEC’11 Raising the Standard),2011.

[87] B. Lock, A.M. Simon, K. Stubblefield, L. Hargrove, Prosthesis-guided trainingfor practical use of pattern recognition control of prostheses, in: Proceedingsof the Myoelectric Controls Symposium (MEC’11), Fredericton, NB, Canada,2011.

[88] A.M. Wing, C. Fraser, The contribution of the thumb to reaching movements,Quart. J. Exp. Psychol. A 35 (2) (1983) 297–309.

[89] C. Fraser, A.W. Wing, A case study of reaching by a user of a manually-operatedartificial hand, Prosthet. Orthot. Int. 5 (3) (1981) 151–156.

[90] R.M. Bongers, P.J. Kyberd, H. Bouwsema, L.P. Kenney, D.H. Plettenburg, C.K.Van der Sluis, Bernstein’s levels of construction of movements applied toupper limb prosthetics, JPO: J. Prosthet. Orthot. 24 (2) (2012) 67–76.

[91] D.F. Seifer, M.J. Mataric, Defining socially assistive robotics, in: 2005 IEEE 9thInternational Conference on Rehabilitation Robotics, 2005, pp. 465–468.

[92] J. Broekens, M. Heerink, H. Rosendal, Assistive social robots in elderly care: areview, Gerontechnology 8 (2) (2009) 94–103.

[93] F. Amirabdollahian, R. op den Akker, S. Bedaf, R. Bormann, H. Draper, V.Evers, G.J. Gelderblom, C.G. Ruiz, D. Hewson, N. Hu, I. Iacono, K.L. Koay, B.Krose, P. Marti, H. Michel, H. Prevot-Huille, U. Reiser, J. Saunders, T. Sorell,K. Dautenhahn, Accompany: acceptable robotiCs COMPanions for AgeiNGYears – multidimensional aspects of human–system interactions, in: The6th International Conference on Human System Interaction (HSI), 2013,pp. 570–577.

[94] H.-M. Gross, C. Schroeter, S. Mueller, M. Volkhardt, E. Einhorn, A. Bley, T.Langner, M. Merten, C. Huijnen, H.v.d. Heuvel, A.v. Berlo, Further progresstowards a home robot companion for people with mild cognitive impairment,in: 2012 IEEE International Conference on Systems, Man, and Cybernetics,2012, pp. 637–644.

[95] D. Lee, T. Yamazaki, S. Helal, Robotic companions for smart space interactions,IEEE Perv. Comput. 8 (2) (2009) 78–84.

[96] A.V. Libin, E.V. Libin, Person–robot interactions from the robopsychologists’point of view: the robotic psychology and robotherapy approach, Proc. IEEE92 (11) (2004) 1789–1803.

[97] D. Feil-Seifer, M.J. Mataric, Socially assistive robotics, IEEE Robot. Automat.Mag. 18 (1) (2011) 24–31.

[98] M. Decker, R. Dillmann, T. Dreier, M. Fischer, M. Gutmann, I. Ott, I. Spieckergenannt Döhmann, Service robotics: do you know your new companion?Framing an interdisciplinary technology assessment, Poiesis Praxis 8 (1)(2011) 25–44.

[99] D. Compagna, S. Derpmann, K. Mauz, The operation of autonomous mobilerobot assistants in the environment of care facilities adopting a user-centereddevelopment design, Enterprise Work Innovat. Stud. 5 (5) (2009) 11–24.

[100] K. Wada, T. Shibata, T. Saito, S. Kayoko, K. Tanie, Psychological and socialeffects of one year robot assisted activity on elderly people at a healthservice facility for the aged, in: IEEE International Conference on Roboticsand Automation (ICRA), 2005, pp. 2785–2790.

[101] D. Compagna, Lost in translation? The dilemma of alignment within partici-patory technology developments, Poiesis Praxis 9 (1) (2012) 125–143.

[102] J.M. Carroll, Scenario-based Design. Envisioning Work and Technology in Sys-tems Development, 1st ed., Wiley, New York, NY, 1995.

[103] A. Tapus, M.J. Mataric, B. Scasselati, Socially assistive robotics [grand chal-lenges of robotics], IEEE Robot. Automat. Mag. 14 (1) (2007) 35–42.

[104] W.E. Bijker, T. Hughes, T. Pinch, The Social Construction of Technological Sys-tems. NEW Directions in the Sociology and History of Technology, 7th ed.,

MIT Press, Cambridge, MA, 1999.

[105] J. Weyer, U. Kirchner, L. Riedl, J. Schmidt, Technik, die Gesellschaft schafft,Soziale Netzwerke als Ort der Technikgenese, 1st ed., Sigma, Berlin, 1997.

[106] G. Abels, A. Bora, Demokratische Technikbewertung, Transcript-Verlag, Biele-feld, Germany, 2004.

7 l Proc

Donald, Attitudes and reactions to a healthcare robot, Telemed. J. e-Health 16

8 A.O. Andrade et al. / Biomedical Signa

[107] A. Chakraborty, M.K. Baowaly, A. Arefin, A.N. Bahar, The role of requirementengineering in software development life cycle, J. Emerg. Trends Comput.Inform. Sci. 3 (5) (2012) 723–729.

[108] A. Strauss, Grundlagen qualitativer Sozialforschung. Datenanalyse und The-oriebildung in der empirischen soziologischen Forschung, 1st ed., Fink,München, 1994.

[109] J. Friedrichs, Methoden empirischer Sozialforschung, Westdeutscher Verlag,Opladen, 1990.

[110] W. Rammert, Technik und Innovation, Technical University Technology Stud-ies Working Papers, Technische Universität Berlin, Berlin, 2008.

[111] H. Braun-Thürmann, Innovation, 1st ed., Transcript-Verlag, Bielefeld, 2005.[112] S. Frenkel, M. Korczynski, K. Shire, M. Tam, On the Front Line. Organization of

Work in the Information Economy, ILR Press, Ithaca, NY, 1999.[113] T. Herrmann, P. Mambrey, K. Shire, Wissensgenese, Wissensteilung und Wis-

sensorganisation in der Arbeitspraxis, 1st ed., Westdt. Verl, Wiesbaden, 2003.[114] S. Star, J. Griesemer, Institutional Ecology. Translations and Boundary Objects:

Amateurs and Professionals in Berkeley’s Museum of Vertebrate Zoology1907–1939, 1989, pp. 387–420.

[115] M.B. Rosson, J. Carroll, Scenario-based Design, 2nd ed., Erlbaum, Mahwah, NJ,2003, pp. 1032–1050.

[116] J. Strübing, Symbolischer Interaktionismus revisited., Konzepte für dieWissenschafts- und Technikforschung, Zeitschrift für Soziologie 26 (5) (1997)368–386.

[117] H. Beyer, K. Holtzblatt, Contextual Design. Defining Customer-Centered Sys-tems, Morgan Kaufmann Publishers Inc., San Francisco, CA, 1998.

[118] A. Cockburn, Writing Effective Use Cases, Cockburn, Highsmiths Series Edit-ors, Addison-Wesley, 2002.

[119] ISO, ISO 9241-210: 2010 Ergonomics of Human–System Interaction – Part210: Human-centred Design for Interactive Systems, 2010.

[120] A. Cooper, The Inmates Are Running The Asylum, Sams, 1999.[121] S. Walter, K.C.S. Limbrecht, V. Hrabal, H. Traue, The Individual Adaption Mod-

ule (iAM): A Framework for Individualization and Calibration of CompanionTechnologies, Taylor & Francis, CRC Press, 2012, pp. 614–623.

[122] S. Walter, K. Jonghwa, D. Hrabal, S.C. Crawcour, H. Kessler, H.C. Traue, Transsi-tuational individual-specific biopsychological classification of emotions, IEEETrans, Syst. Man Cybernet. Syst. 43 (4) (2013) 988–995.

[123] S. Walter, C. Wendt, J. Bohnke, S. Crawcour, J.W. Tan, A. Chan, K. Limbrecht, S.Gruss, H.C. Traue, Similarities and differences of emotions in human–machineand human–human interactions: what kind of emotions are relevant forfuture companion systems? Ergonomics (2013) (Epub ahead of print).

[124] S. Biundo, A. Wendemuth, Von kognitiven technischen Systemen zuCompanion-Systemen, Künstliche Intelligenz 24 (4) (2010) 335–339.

[125] D.L. Sackeit, W.C. Rosenberg, J.A.M. Gray, R.B. Haynes, W.S. Richardson, Evi-dence based medicine: what it is and what it isn’t, Br. Med. J. 312 (13) (1996)71–72.

[126] H.-y. Yu, D.F. Friedlander, S. Patel, J.C. Hu, The current status of robotic onco-logic surgery, CA: A Cancer J. Clin. 63 (1) (2013) 45–56.

essing and Control 10 (2014) 65–78

[127] M. Zhang, T.C. Davies, S. Xie, Effectiveness of robot-assisted therapy on anklerehabilitation – a systematic review, J. Neuroeng. Rehabil. 10 (2013) 30.

[128] M.R. Banks, L.M. Willoughby, W.A. Banks, Animal-assisted therapy and lone-liness in nursing homes: use of robotic versus living dogs, J. Am. Med. Direct.Assoc. 9 (3) (2008) 173–177.

[129] I. Nisky, S. Patil, M.H. Hsieh, A.M. Okamura, Kinematic analysis of motorperformance in robot-assisted surgery: a preliminary study, Stud. HealthTechnol. Inform. 184 (2013) 302–308.

[130] F. Volonte, F. Pugin, N.C. Buchs, J. Spaltenstein, M. Hagen, O. Ratib, P. Morel,Console-integrated stereoscopic OsiriX 3D volume-rendered images for daVinci colorectal robotic surgery, Surg. Innovat. 20 (2) (2012) 158–163.

[131] A. Panarese, R. Colombo, I. Sterpi, F. Pisano, S. Micera, Tracking motor improve-ment at the subtask level during robot-aided neurorehabilitation of strokepatients, Neurorehabil. Neural Repair 26 (7) (2012) 822–833.

[132] M. Bonfe, F. Boriero, R. Dodi, P. Fiorini, A. Morandi, R. Muradore, L. Pasquale, A.Sanna, C. Secchi, Towards automated surgical robotics: a requirements engi-neering approach, in: Proceedings of the 4th IEEE RAS and EMBS InternationalConference on Biomedical Robotics and Biomechatronics (BioRob), 2012, pp.56–61.

[133] G.E. Tasian, D.J. Wiebe, P. Casale, Learning curve of robotic assisted pyeloplastyfor pediatric urology fellows, J. Urol. 190 (4) (2013) 1622–1627.

[134] E.L. Jenison, K.M. Gil, T.S. Lendvay, M.S. Guy, Robotic surgical skills: acquisi-tion, maintenance, and degradation, JSLS 16 (2) (2012) 218–228.

[135] H. Bouwsema, C.K. van der Sluis, R.M. Bongers, The role of order of practicein learning to handle an upper-limb prosthesis, Arch. Phys. Med. Rehabil. 89(9) (2008) 1759–1764.

[136] D.J. Gow, W. Douglas, C. Geggie, E. Monteith, D. Stewart, The development ofthe Edinburgh modular arm system, Proc. Inst. Mech. Eng. H 215 (3) (2001)291–298.

[137] A.N. Spring, J. Kofman, E.D. Lemaire, Design and evaluation of an orthoticknee-extension assist, IEEE Trans. Neural Syst. Rehabil. Eng. 20 (5) (2012)678–687.

[138] A. Simorov, R.S. Otte, C.M. Kopietz, D. Oleynikov, Review of surgical roboticsuser interface: what is the best way to control robotic surgery? Surg. Endosc.26 (8) (2012) 2117–2125.

[139] T.H. Wagner, A.C. Lo, P. Peduzzi, D.M. Bravata, G.D. Huang, H.I. Krebs, R.J.Ringer, D.G. Federman, L.G. Richards, J.K. Haselkorn, G.F. Wittenberg, B.T.Volpe, C.T. Bever, P.W. Duncan, A. Siroka, P.D. Guarino, An economic analysisof Robot-assisted therapy for long-term upper-limb impairment after stroke,Stroke 42 (9) (2011) 2630–2632.

[140] E. Broadbent, I.H. Kuo, Y.I. Lee, J. Rabindran, N. Kerse, R. Stafford, B.A. Mac-

(5) (2010) 608–613.[141] Z. Akbulut, A.E. Canda, A.F. Atmaca, E. Asil, E. Isgoren, M.D. Balbay, What if the

hand piece spring disassembles during robotic radical prostatectomy? J. Soc.Laparoendosc. Surg. 15 (2) (2011) 275–278.