Download - EVALUATION REPORT: Early Clinical Evaluation of a Robot ... · include chin or head motion, eye tracking motion, and voice. in early examination of control alternatives at the APb,

37

Journal of Rehabilitation Research and Development Vol . 22 No . 1 1985

EVALUATION REPORT

38

SEAMONE & SOHMEISSER : Evaluation Reports

°

Early Clinical Evaluationof a Robot Arm/Worktable Systemfor Spinal-Cord-Injured Persons a

WOODROW SEAMONE, B .A .E.GERHARD SCHMEISSER, M .D.

The Johns Hopkins UniversityApplied Physics Laboratory

andThe Johns Hopkins University School of Medicine

Abstract — In Part !' "Developmental Case Study", thisevaluation traces anecdotally the increasing application ofmicrocomputer technology in a research pnogramexamin-ing the uses of a robot arm in the rehabilitation of thehigh-spinal-cord-injured person . This program, supportedby the Veterans Administration since 1974, built upon stillearlier VA-supported work at Johns Hopkins on poweredupper-limb prostheses . The Johns Hopkins Universitypowered shoulder prosthesis served as the basis for theRobotic Arm/Worktable Symtem, whose evolution is de-scribed in appiications ranging from handling simplereading materials to self-feeding and the operation of apersonal computer, all under the user's control through thesame chin control interface employed for the control of theuser's power wheelchair.

Part II of this report focuses on the results of clinicalevaluation in the development of the system . In the earlystages of deve\opment, evaluation was provided by foursubjects who used the system regulariy for periods rangingfrom 4 months to a year . A later stage of evaluation featureda total of 16 subjects located in two VA Medical Centers,using the system for periods of from a few days to 4 monthsin duration.

The VA's new Rehab R&D Evaluation Unit has begun anintensive evaluation process based upon 25 of the RoboticArm/Worktable Systems currently on order.

aThis project is sponsored by the Veterans Administration Reha-bilitation Research and Development Service.

NOTE : Mr . Seamone is a member of the Principal Staff, Fleet Sys-tems Department . Applied Physics Laboratory, Johns Hopkins Rd .,Laurel, Maryland 20707 . Dr . Schmeisser is Professor of OrthopedicSurgery, The Johns Hopkins University School of Medicine,Baltimore, Maryland 21205 .

PART 1 : DEVELOPMENTAL CASE STUDY

INTRODUCTION -- Fast moving mioropruoeoaorteohnol-

ogy leading to low-cost, highly capable microcomputersystems has been a key technical element in a researchprogram aimed at examining the practicality of using arobot arm to assist the physically handicapped . Aparticu-larly critical need exists to provide the high-spinal-cord-injured person with devices that increase self-care andindependence for long periods of the day . Such individuals,with total loss of upper and lower limb function, representone of the areas of greatest need in rehab\!ita1ionengi-ne*ring.

A research program at the Applied Physics Laboratoryof The Johns Hopkins University has been under way since1974 . examining the applicability of robotics to the solu-tion of this problem. The program, sponsored by theVeterans Administration, has now reached the stage ofcomprehensive clinical testing at Spinal Cord Injury Centersat VA K6edico! Centers at Richmond, Virginia, and Cleve-land, Ohio.

Early in this program, visits were made to patients inSpinal Cord Injury Centers to determine the priority ofneeds which must be met by this equipment . Some of theimportant needs identified included self-feeding, manage-ment of a variety of reading materials, use of a telephone,use of a typewriter including insertion and removal of asheet of paper, various hygiene and personal needs, andvocational capability such as the use of a personalcomputer.

Two approaches offer possible solutions to these need&One approach, that of designing specific devices to addresseach of these needs, leads to an independent mechanicaldevice for each of the requirements . The cost and mainten-ance ofsuohduviceuarenonaidereb\e. Operation of morethan one such device in a given work area is usually difficultor impossible because of lack of integration among themechanical designs of the subsystems . For example, atypical electromechanical magazine-reading machine isso large and bulky that it precludes placing other deviceswithin reach of the SC1 patient . The alternative, a singlemultipurpose device such as a robot, offers the promise ofaccomplishing many of the required tasks within a reason-able work area, and its motion flexibility should enable itto handle a variety of needed tasks.

Robot Functional Requirements for the SC1 Patient

In the more severe spinal-cord-injured cases, individualsare left with little or no functional capability of the handsand arms. The basic function of the robot, therefore, is toprovide a substitute for lost manipulative capabilities.Since the motions, speed, force capability, and controlla-bility of the human arm clearly span a very wide dynamicrange, compromises must be made in the selection ofparameters for the robot in this human interface applica-ation. Among the compromises are some related to safety.Robots in induot/ial epp!ica1iuna, for example, clearlyneed to perform at high force ievels and with brisk re-

39

Journal of Rehabilitation Research and Development Vol . 22No . 1 1985

NEW VA EVALUATION UNIT HAS PROTOTYPES ON ORDER

You have probably already noticed, elsewherein this issue of the Journal, a list of promisingdevices under evaluation by the VA's new RehabR&D Evaluation Unit . Third on that list is theRobotic Arm/Worktable System for quadripleg-ics, An order has already been placed for25 of these in the form of production proto-types -- versions of the device that have beencarefully constructed so as to be suitable forcommercial production in volume.

They will get intensive testing, and it is verylikely that at least 100 spinal-cord-injured sub-jects will have the opportunity to live and workwith those 25 prototypes . Obviously they will befinding some things that can or must be im-proved about the device . But because theseprototypes will have been designed tobemanu-factured, it should be easier to plan improve-ments without the risk of creating manufactur-ing difficulties and delays later on.

The Robotic Arm/Worktable System has, withVA support, been under development at the

Johns Hopkins University Applied Physics Lab-oratory ainoe1874 . It was in that calendar yearthat Woodrow Seamone and Gerhard Schmeis-ser and their JHUIAPL colleagues first settledon the design rationale for the robotic aid thatthey called a "manipulator." It had grown,rather naturally, out of VA-sponsored work bythe same team on externally powered upperlimb prostheses, with special attention to theabove-elbow amputee.

In the paper starting on these pages, principalinvestigators Seamone and Schmeisser provideauondenood recapitulation ofthodevice's de-velopment with emphasis on the more recent,computer-dominated, period : they conclude witha description of a year-and-a-half of clinicalfield testing (starting in 1983) which precededacceptance of the Robotic Arm/Worktable Sys-tem as one of the first devices to receive theattention of the VA's new Rehab R&D EvaluationUnit.

sponsiveness, but r human beings are not normally permittedwithin range during robot activation . In contrast, a medicalapplication such as that considered here requires therobot to perform in intimate proximity with the user andct!!l not present any threat to the person -- even in theevent of a computer fault or other system failure. Forcelevels of 1 or 2 pounds and maximum velocities of lessthan 10 cm per second are typical of safe values for adevice operating in close proximity to a human being . Andeven at these safe levels, it remains important that theuser or attendant be able to stop the robot easily andreliably at all times.

The range of motions should, ideally, be greater thanone meter, with control precise to within a few millimeters.

A major design consideration is whether the robot shouldbe mounted on the user's wheelchair, or be mounted on aworktable which can be moved only by an attendant, or bemade independently mobile, Early models of robots exam-ined in the mid-70's included both wheelchair-mounted(1, 2, 5) and worktable models (3, 4, 6) . At JHU/APL thedecision was made to go to the worktable arrangement (6)because the size, bulk, and weight of a wheelchair-mountedsystem was judged unacceptable to the user.

The JHU/APL project was initiated in 1974 as a follow-upon more conventional prosthetics/orthotics research . Theultimate robot system cost was targeted to be less -1

$10 ' 000 . At that time, no commercial robot SyStE

to carry out the desired tasks, thE

c : sidered safe fornpom1ioninc|oaepnzximitytoa~.ace!cappedpenson .and

that was within the allowable cost range . However, duringthe 1809-1S73time period, JHUA4PLhed developed a low-cost 3-degrees-of-freedom powered prosthesis for shoulderdisarticulation amputees that had the desired limitedforce and velocity characteristics . Since the initial studyof the applicability of a robot arm was aimed at studyingrequirements and interface methods to control suchdevices (as opposed to detailed design and e!ectro-maohaniuul implementation methods) a decision wasmade 1ouoethmexisting JHUIARLpowered shoulder pros-thesis andmodifyitappnopriote!yfortheeadyvobotoiud\em.

The basic robot arm was initially implemented with fivedegrees of freedom:

1. Hand grasping;2. Wrist pronation/supination;3. Elbow flexion/extension;4. Shoulder motion flexion/extension (coordinated with

elbow motion) ; and5. Turntable motion (comparable to shoulder internal/

external rotation).A sketch showing these degrees of freedom and the

range of motions is shown in Figure 1 . Initial testing withthat model soon revealed it to be unsatisfactory, primarilybecause radial and vertical motions from a turntable couldnot carry out many of the desired functions satisfactorily.The result was an early decision to add a sixth degree offreedom, implemented by mounting the robot assembly ona servocontrolled 70'cm track . The addiUonal degree offreedom improved the geometric capability sufficiently toallow implementation of the planned list of tasks .

40

FIGURE 1Geometry of the JHU/APL. Robotic Arm.

input Modes to the Robot

in the more severe spinal-cord-injured cases, individualsare left with little or no functional capability of the handsor upper limbs . Many of these individuals retain full rangeof motion of head and neck and also retain the ability tospeak. Possible control inputs for mechanical devicesinclude chin or head motion, eye tracking motion, and voice.

in early examination of control alternatives at the APb,oontnml by voice commands was considered but was re-jected because it lacked high reliability . It also proved tobe difficult to control the end-point motion of the robot armby voice . (Considering the significant progress in voice-recognition technology within the past few years, it maybe appropriate to reconsider the applicability of voicecontrol to the robot problem .)

Some of the input interfaces for conventional robotsconsidered during the early phase of this project includekeyboard entry, joystick or model controls, and punched orprogrammed magnetic tapes . For applications to theneeds of quadrip!mAioa, most of these 'techniques areunsatisfactory because disabled persons lack the capabil-ity toprovide inputs to keyboard and joysticks for asus-tained period of time.

Chin motion input was selected, after much study andevaluation of alternatives, because of its positive controland good resolution capability . The individual user of arobot arm must be capable of watching the end-pointmotion of the robotic arm at all times while inputting thesystem. A chin motion up/down sensor with an extradegree of freedom (to permit the necessary !ateral headmotion) appeared to be a reasonable approach and hasbeen implemented for all clinical evaluation in the APLrobotic arm/worktable system .

Integrated Robotic Arm Worktable Concept

The MC/API_ robotic arm/worktable system has beendesigned with the specific goal of enabling the quadriplegicto execute total tasks with little or no attendant assistance.To accomplish this goal, a structured worktable concept isutilized ; i .e ., components are located on the worktable infixed locations that allow the robot to use manual step-by-step mo\ionaorpnyotonsdcompuhs pmont/o!!md motion tra-jectories to carry out a desired function . This conceptmakes manageable, for example, even the difficult task ofputting a single sheet of paper into the typewriter.

To control the robot, the user calls up the desired pro-gram with an appropriate motion to the chin controller . Thesystem is designed for the highly disabled person who isunable to use his arms or hands but has nearly normalrange of motion of the head and neck ; these motions con-trol either the robot system or the mobility of a conventionalelectric wheelchair.

Steering control of the electric wheelchair is achievedby lateral motion of the chin controller (7) . Torque of themotors, hence wheelchair motion, is controlled by how farthe chin control lever is depressed . Reverse mode iose-lected by means of a microswitch button located on thechin control lever . Very little of the apparatus is in front ofthe user . The quadriplegic may electively push it out of theway when it is not being used.

The user drives his wheelchair with the dual-purposechin controller for his normal daily activities . When heapproaches the docking position at the worktable, a probeon the wheelchair makes an optical (infrared) connectionto the worktable . Theuser lifts thechin controller momen-tarily and this motion serves to transfer control from thewheelchair motors to an optical link located near one armof the wheelchair . The chin controller then modulatesan infrared signal which has a pulse and a proportionalmode that enable it to control the robot arm . This transfer-ability and duality of control gives the quadriplegic much-needed mobility (via power wheelchair) as well as the abilityto drive up to his worktable and begin to manipulate theitems arrayed there by means of the robot arm — and tomake this transition from travel to manipulation withoutany need to transfer physically from the wheelchair or todeal with a change in control modality.

A block diagram of the total system is shown in Figure 2.The robot arm is a 6-dmgroee'of'freedom, computer-

controlled anthropomorphic limb . The individual degreesof freedom of this arm may (if the user so elects) be directlycontrolled by selection of the desired joint to be moved.Motion power is provided by two pancake-style DC torquemotors and one geared servomotor used on the horizontaltrack.

One motor activates elbow flexion, wrist pronation/supination, or shoulder flexion/extension with parallelo-gram motions of the forearm. The second motor, locatedinthevertical column, activates column rotation ortnnnina!'device grasping . Solenoid locks under software controllock all axes not in active motion . The controller for thearm is a microprocessor controller located within thevertical section of the arm,

41


FIGURE 2Block diagram of microprocessor-controlled robotic arm with wheelchair interface.

EXTERNAL DEVICESCONTROL & SENSE

ELECTRICWHEELCHAIR

CHIN MOTIONCONTROLLER

DISPLAYINPUT/OUTPUT

PULSE LINEARMOTION I

FUNCTIONALKEYBOARD

I

MICROPROCESSORCONTROLLER

OPTICALTRANS

OPTICALRCVR

6 DEGREE OFFREEDOM

ROBOTICARM

---+-GRASP POSITION

DATA

CMD,

CONTROLMOTION

FIGURE 3This robotic arm and worktable system is arranged for the use of a personal computer . Morse Code inputdevice is at far left . Note telephone at far right .

42

FIGURE 4This view shows part of the worktable when configured for use of an electronic typewriter, a low-cost machinewith self-correcting capability . The telephone is out of sight at the right, but within reach of the robotic arm'sterminal device . Vertical post at left of typewriter carries control display at top ; the loop alongside of the columnis actually the chin-activated input controller for the arm . Supported by a shorter column and brackets are theMorse Code keyer and a mouthstick stored in its socket.

The worktable is configured with the robot arm on its70-cm track, with components which are to be manipulatedplaced at strategic locations on the table . An example ofone configuration for the worktable which was evaluatedby quadriplegic users is shown in Figure 3. Componentssuch as a personal computer, telephone, and book readingrack are shown in that arrangement.

Prestored task trajectories — As an alternative to directcwntrol of any single axis-of-motion of the arm, the usermay call up prestored trajectories to accomplish a specifictask. The following tasks, which are only a partial list, useprestored application programs in the current model of therobot system:

1. Move mouthstick or Morse Code keyer into position;2. Pick up telephone and place it into position for use;3. Hang up telephone;4. Pick up Kleenex tissue ;

5. Move typewriter forward;6. Put sheet of paper into typewriter;7. Remove sheet from typewriter;8. Place either of 2 diskettes into a computer disk drive;9. Return diskettes to storage rack;

10. Pick up any of five magazines in storage rack andplace on reading stand;

11. Return book to storage location;12. Eat sandwich from plate;13. Eat with spoon in plate; and14. Eat from a bowl.

An important toml provided to the quadriplegic is amouthstick which is moved into a prominent position nearthe front of the table by the robot . Manipulative functions,such as putting a magazine in place for reading, areaccomplished by the robot, while page turning is ac-complished by the occupant mouthstick . Likewise, the

43Journal of Rehabilitation Research and Development Vol .22 No. 1 1985

robot can bring the phone to the user's ear, but dialingtelephone numbers is done with the mouthstick . The work-table may be configured in many ways with the specificvocational or personal tools desired or required by theuser; another arrangement of components on the worktableis shown in Figure 4. In addition to components such as atelephone handset, a book storage and reading rack, thetable includes an electronic typewriter . This typewriter ismoved into position by the robot arm, and typing is ac-complished with the mouthstick . The na!f'uorv*o1ingfea-tunea of this low-cost electronic typewriter (cost is lessthan $200) make it possible to produce error-free typed work.

The Morse Code computer interface - Figure 3 showedthe table configured to include a personal computer . Theinput to this computer is via Morse Code in lieu of thekeyboard . A standard Morse Code keyer operated by aminute motion of the chin gives the user full control of allkeyboard characters, including the Control and Shift keys.(Since the computer keyboard is not utilized directly, itcan be moved to the rear of the worktable .) A series ofaudible dots or dashes generated by minute chin motionsactivates a single-chip microprocessor which translatesthe code into the appropriate ASCII characters and sendsthem to the computer. An alternate input method is to usesip-and-puff in a tube to create more characters.

These approaches were verified bytests inwhich quadri-plegic vo!unteemopemted an Apple computer. Speeds of 60characters per minute are achievable, and simultaneouskey entries such as those required with Control, Shift, orRepeat keys are accommodated . In demonstrations atAPL, word processor programs such as Applewriter orMagic Window, and spreadsheet programs such as Visicalc,were shown to be practical software choices.

The demand for this Morse Code interface with theApple computer was great enough to convince a manufac-turer toeign a license agreement with APL to manufactureand ae!l the computer interface device . Integrating thecomputer with 1hnrobot a/mgives theindividual thecapa-bility to use reading materials in combination with thecomputer, and supports such tasks as changing disks forcomputer programs without the need to call an attendant.

Improvements in the self-feeding system — Self-feedingby the quadriplegic is one of the more important tasks tobe evaluated with this system . During the course of theresearch program, the devices and the prestored programsused for self-feeding went through many design changesbased on user feedback . Changes in the platelbowl arrange-ment andmdeoignoftheopoonimp/oved1heuoer'eabi!ityto eat sandwiches, hot dogs, French fries, and salads, aswell as bite-size food . This spoon arrangement with thewire clamp is shown in Figure 5.

Two bowls and a plate may be used for eating . Whenthe user approaches the desired portion of food or sand-iich in the dinner plate with the spoon, the wire clamp isut of the way. Clamp motion is activated to grasp the foodt the appropriate part of the cycle . After the sandwich or

other food is brought to the mouth, the user may commandthe clamp to open so that he may take the last bite off thespoon.

In thepnootored bowl eating program, the spoon is pre-programmed to go into the bowl (which automaticallybegins rotating), pick up a bite-size portion of food, scrapethe bottom of the spoon to remove drippings, and proceedto the user's mouth. When eating, the user may switchamong the several eating modes, e .g ., from bowl to plateor from plate to bowl . The self-feeding mode was tested atAPL and at the VA Medical Center in Richmond, Virginia,and shows great promise as a practical eating arrangement.

For use upon completing a meal, a prestored program isprovided which directs the arm to drop the utensil andthen go over to pick up a facial tissue . The user may usethis tissue to clean his mouth and then drop the tissue intothe wastebasket.

MICROPROCESSOR CONTROLLER

The current microprocessor-based controller uses theMostek F8 microprocessor as the control means for therobot . (A new system utilizing a 6809 microprocessor iscurrently being evaluated at the Laboratory .)

A block diagram of the F8 controller and electromechan-ical subsystems of the arm are shown in Figure 6 . Thecontroller resides on an 8 cm x 36 cm wire wrap card onthe back of the robot-arm vertical assembly . A total of 6144words of ROM exists for controller software and up to 48preprogrammed motion sequences : 4,017 words ofbattery-backup memory are available to store user-changeablemotion sequences . Totol power consumption is approxi-mately three watts for the computer system.

The major tasks performed by the software are systeminitialization, selection and operation of arm motion undermanual or preprogrammed control, and keyboard program-ming new motion sequence. Under manual control, theuser commands the robot arm to its desired position byissuing velocity commands by means of chin motion . Armmotion is started when the transducer is depressed half-way, the zero velocity position . The selected joint is broughtto a torque balance condition and the solenoid lock as-sociated with that joint is released . Motion of the chinlever up or down from this position produces bipolar veloc-ity proportional to displacement from the zerovelocity position . Viauel feedback of joint motion (or ob-serving the numerical position indicator on the display)provides the means to position the arm to a desired endpoint . Motion is terminated with a discrete command (apulse on the chin controller), and a lockup of the jointoccurs after velocity has been reduced by the computer toa low level.

An important mode of operation of this robotic arm isthe use of preprogrammed motion sequences . Selectingoperation under automatic control requires the user toselect one of aeveral tasks (phon e, eat, pick up books,etc .) by calling up an appropriate p age on the displaypanel . Page and name selection of the desired program(i .e ., eat a sandwich) is achieved with propodional corn-

44

SEAMONE & SCHMEISSER : Evaluation Reports

FIGURE 5Redesigned spoon with its clamping device in retract mode .

45


mend scanning the programs on each page and then sel-ecting the desired program by a pulse command.

An interpreter program in the software interprets theselected program and then sequences the arm through thecorresponding motion sequences . Motion during this phaseis automatic, is acceleration-limited, and is commanded tolock up smoothly.

Should motion ota!l due to an obstruction, the motionbeing executed will time out and the next command in thesequence will be executed . The user may terminate any ofthe preprogrammed sequences at any time by a singlepulse to the chin controller . This provides a safe stop inthe event the program has gone astray .

Keyboard programming — An important feature of thissystem is the programming keyboard for the applicationprograms. This keyboard is designed to be used by eitherthe therapist or an experienced quadriplegic user.

The robot arm motions are specified by functional keyson the programming keyboard, which is shown in Figure 7at a temporary location near the front of the table . (Thiskeyboard is plugged into the robot arm to make programchanges but is removed during normal operation .) Editingkeys allow the sequential numbered commands to beeasily inserted and deleted . Motion sequence files (robotarm trajectory motions) are identified by page name (col-lection of files) and program name (specific file) . These

EXTERNAL DEVICECONTROL & SENSE

t

F8 MICROCOMPUTER

8192 VVDS EPROM4096 mDS CMOSRAM SERIAL

INTERFACE9 PROGRAM6 POSITION

/mrmw/

DISPLAY-110

FUNCTIONALKEYBOARDINTERFACE

RS-232TERMINAL

INTERFACE

nonomCONVERTERS

comm.INTERFACE

no

oBIr ^mCONVERTER

MOVEMENTSELECTIONSOLENOIDS

SHOULDER EXTENSION

ELBOW FLEXION

WRIST ROTATION

TURNTABLE

HOOK

T

PULSE WIDTHMODULATION

DRIVERS

DISCRETE CO MAND

LINEAR COMMAND/n

RECEIVER

POSITIONPOTENTIOMETERS

/nTRANSMITTER

MANIPULATOR

WHEELCHAIR CHIN COMMANDER

FIGURE 6

Block diagram of present microprocessor-based control system and electromechanical subsystems .

46

SEAMONs&SCHMBS@ER : Evaivation Reports

names are specified with alphanumeric keys . Ali filesreside in battery-backup CMOS (complementary metal-oxide semiconductor) read-write memory, and all filehandling is transparent to the user.

The functional keys define BASIC-like language elements

for easy specification of motion sequences . Commandsexist for motion to a point, conditional and unconditionalbranching, stimulation and sensing of external devices,jumps to other motions sequences, pauses, and more.

FIGURE 7Partial view of worktable arranged for

uuoo,aconventional portable type-writer . Small keyboard temporarilyplaced to right of typewriter is used

to make changes in the robotic arm'sapplication programs; it is removed

during normal operation.

PART 2: CLINICAL EVALUATION

PHILOSOPHY—In undertaking the clinical evaluation itwas presumed that some sort of Robot Arm Worktable(RAWT) system might be worthwhile for some quadri-plegics to help them carry out a variety of ueeful manipula-tive tasks, as previously mentioned . It was also assumedthat the APIs Robot Arm Worktable (RAWT) system, asinitially fabricated, would require additional developmentand refinement to become worthwhile for these purposes.By seiecting and training quadriplegic volunteers to usethis equipment and make recommendations, featuresneeding improvement could be identified . In 1982 it wasproposed to the Veterans Administration's RehabilitationEngineering Research and Development Service b (VARERADS) that a 2-year clinical evaluation program becarried out at two VA Spinal Cord Injury Services . Alternate4-month testing periods at the Spinal Cord Injury Serviceswere to be followed by periods of approximately 2 monthsto upgrade the RAWT system at the JHU AK.. The upgrad-

b Now incorporated in the VA's Rehabilitation Research and Devel-opment 8*mioe

ing was to be performed in response to the recommenda-tions obtained from the quadriplegics during their periodsof evaluation.

The ultimate objective of the clinical evaluation was toobtain information in order to bring the equipment to a!evol of performance that might justify commercial fabrica-tion

Resources and Methodology

Clinical evaluation of the API_ RAWT and associatedequipment has been conducted in three locations, andunder somewhat different circumstances and with some-what different protocols . The locations have been theBaltimore-Washington area, the Richmond VA MedicalCenter (RVAMC), and the Cleveland VA Medical Center(CVAMdC).

Baltimore-Washington area tests—The first attempts atoperation of the API_ Robot Arm by quadriplegics wereconducted in the Baltimore-Washington (BW) area . Clinicalevaluation at this early stage of the research program wasdirected specifically at determining whether the RobotArm should be attached to the user's wheelchair, afvee'

47

Journal of Rehabilitation Research and Development Vol . 22 No . 11385

standing pedestal, a bedside table, an over-the-bed table,or a separate desk or worktable. Information was alsosought on whether additional freedoms of motion wererequired, what sort of useful activities might the deviceperform, how might it be controlled most easily by theuaer, and what sort of ancillary work tools and theirarrangement would be most useful . Informal protocolswere developed for each quadriplegic as uses of the equip-ment were explored . The staff consisted of the authors andseveral interested quadriplegics selected by the authorsfrom among those referred by miscellaneous personsfamiliar with the project . Two of the first four quadriplegicsto use the RAVVTvx*ne brought separately to the'APL forinitial orientation to the equipment . This was performed bydemonstration and by enabling them to operate it in thelaboratory for a few hours at a time under the immediatesupervision of the investigators . The unit was then set upin the quadriplegic's place of residence.

For the first evaluator, BW-1, the tests were conductedat the Maryland Vocational Rehabilitation Center, a state-owned facility operated for rehabilitation of disabled per-sons (Table 1) . During his period of evaluation he was livingin this institution and working with the RWAT in a well-equipped occupational therapy facility . An occupationaltherapist was assigned to assist him . This person had alsobeen oriented to the equipment.

The second evaluator, BW-2, resided in a local nursinghome. The RAWT was set up in his private bedroom andmounted on an over-the-bed table.

The third evaluator, BW-3, lived in his own family'ssuburban household . The RAWT was set up in his bedroomon a desk-type worktable where he could operate it fromhis wheelchair (Fig . 8).

The fourth evaluator, BW-4, resided in a small statehospital for the chronically ill . The RAWT was set up on avery small table with the intention that it be used exclusive-ly foraa>f-feeding . For security, the unit was located in theinstitution's OTIPT facility. The latter three evaluatorswere not assigned attendants to help in the evaluation ofthe equipment . One of the investigators visited or com-municated by phone with the evaluator approximatelyweekly, or as often as seemed justified.

Richmond VAMC evaluation project—Evaluation at a VASpinal Cord Injury Service was first conducted at theRVAMC early in 1983 . This evaluation followed a formelresearch protocol developed by the Chief of that Serviceand an Occupational Therapy student interested in thisas a thesis project. The latter person was thoroughlyoriented in the design, intended use, operation, andmaintenance of the equipment in the course of a several-day visit to the JHU Applied Physics Laboratory . Thisperson has since received her degree and has conductedthe project as a full-time activity during the periods ofclinical evaluation at RVAMC.

The protocol presented the aims of the project withinthat institution, the study design, instructional procedures,

informa-

TABLE 1Summary of patient characteristics : Baltimore-Washington area

Quad.CodeNo .

Age atEva.(Yrs .)

Time fromfrom InjuryEval .

Level ofInjury Cause Occupation

Residencefrom injuryto Eval . Physical condition

BW-1 36 5 yr . C4 Waterskiing

Marines VAMC Good trunk stabilityGood h*adnecKcontm!Weak nhou!derehmgNo upper limb functionNo respiratory problems

BW-2 35 2 yr. C4 Surfing Ph . D.physicist

Nursinghome

Fair trunk stabilityGood head-neck controlFair shoulder shrugTrace of one bicepsNo respiratory problems

BW-3 27 3 yr . C5 Fall Salesman Hissuburban

ho!d

Good trunk stabilityGood head-neck controlWeak biceps one sideNo respiratory problems

Customary usageof assistivedevices

Electric wheelchairwith VAPC chin-controller ; Head-stick user

Attendant propelledwheelchair ; goodmouthstick user

Attendan propelledwxeo!oxi r . lurgi'na

Y PCcontra

electricwheele .iLir withmobile armsupports

BW-4

31

7 yr.

C5

!nfe

Studenttion

State hospi-

Good

ability

Attendant propelledtal for

Good tcontro

wheelchairchronic

No upper

inctiondiseases

No reepeo problems

48

!nstax

s "

evaluaBulledao4o

g ,ip!*g|uwn1981 h

is ,um!aheua robo

.nn/wo,mamapplication ,»m*con-

and he liked the worktaoievaluate the self-feeding c

amiwo,kt ^it of the JHI.

m.ePR 1n-3c

nificantnmp!eton . For at

his but

`virnnnvuxt»si,

in the homes and home offices of volunteer)rjnam. The photo shown here appeal 3 1.1 the

a such volunteer : 35-years old at the ti e .he!seof

drandg. )n the

controllerme in agivon wa. no did notthe need for it.

for evaluatiM. developmall 1981 . If sh

cry mvsu/swhich n

o,uham3s . This user was

/of adu .

'pee c,i mx,

`envsto

me, cn~

on,

mgh! ' nnt

ast!oabovt txawhsnhich provided multiple furle lived at home and did not

Uonhorme, consent forms ei

ationquestionnaires, etciThe project was oriented

a!uate each of the com-ponent subsystems as vl

performance of theentire RAWT system when o! E

combination withthe APL Chin Controller m 'an E&J electricwheelchair . The quadriplegic Ix*

avaivators were tobe oriented in the purposes

3 opera-tion by the Principal In as

:tor . Atter learning

operateit to the extent allowed Jooignand their ovphysicaloonUibonm, they were to x!uate it in accord with theprotocol . Insofar as possible, each individual was to bescheduled to work with the equipment for at 1---t a fewhours daily over a 1 to 3 month period . This w(

permittwo individuals to work with the equipment eve

e samemonth.

Nine quadriplegics, designated R-1 through R-9 in Table2 . were involved in the evaluation at RVAMC. All wereinpatients and worked under the immediate supervision of

the Principal investigator of the pr

VAMC . Theirwork on the RAVVTwa000nduotec

!y in a roomadjacent to the OT facility . This r

convenientlyd for wheelchair access by

aantsi It wasjus enough not only to pr e ii ir

dip— approaches fora wheelchair operator to dock a

AWT but for evalua-tion of wt

' iroontro! mw~ ~ the RAWT and withminimum

to the equipment, tne evaluator, or otherpersons.

Cleveland VAMC evaluations—Evaluation at the Spinalr-rd Injury Service at CVAMC was started 6 months after

a start at RVAMC. The protocol followed was similar toat at RVAMC in aims and design, but different in staff

selection and evaluation procedures . For a

andalportion of the evaluation pe'md, the Senior C .

tionalTherapist at CVAMC was t

/ the Principal Inv m igatorwho developed and obtained approval of the protocol and

49


TABLE 2Summary of patient characteristics: Richmond VAMC

Quad.CodeNo .

Age atEval.(Yrs .)

Time fromfrom InjuryEval .


Residencefrom Injuryto Eval . Physical condition

Customary usageof assistivedevices

R-1 25 8 mo . C3 Car Marines;Mechanic

VAMC Good trunk stabilityFair head-neck controlGood shoulder shrugWeak biceps one sideMinimal fatigue problem

Electric wheelchairwith VAPC handcontroller

R-2 32 2 yr . C3 Gunshot Musician;Horse-breeder

Personalresidencewith wife

Poor trunk stabilityFair head-neck controlWeak shoulder shrugNo upper-limb functionRespirator dependentFatigue problem

Attendant propelledpower reclinerwheelchair

R-3 40 8 mo. C3 Waterskiing

Air Force;Standardiza-tion Officer

VAMC Poor trunk stabilityFair head-neck controlWeak shoulder shrugBeing weaned from

respiratorFatigue problem

Electric wheelchairwith VAPC chin-controller

R-4 30 1 yr. C4 Diving Securityguard;Law schoolstudent

VAMC Poor trunk stabilityPoor head-neck controlWeak shoulder shrugRespirator dependent

Usually on litter.Occasionallyelectric wheelchair& VAPC chin-controller

R-5 21 5 mo . C4 Car Marines;Carpenter;Laborer;Singer

VAMC Good trunk stabilityGood shoulder shrugWeak biceps one sideMinimal fatigue problem

Electric wheelchair& VAPC hand-controller

R-6 45 10 yr. C4-5incom-plete

Fall Inventorycontrolanalyst

Personalresidencewith wife

No hand functionSome limited arm

functionSome limited ability

to walk


R-7 40 7 yr . C4-5 Motor-cycle

Seniorcomputersystemsengineer

Personalresidencewith family

Good trunk stabilitySome use of deltoidsand left biceps


R-8 23 18 mo . C4 Diving Laborer;Pipe fitter

VAMC Fair head-neck controlPoor shoulder shrugNo upper-limb functionGood fatigue tolerance

Attendant propelledwheelchair

R-9 49 9 mo . C3-4 Car Insurancesalesman;Textile plantmanager

VAMC Good trunk stabilityWeak deltoid and bicepsRecently weaned from

respiratorNo upper-limb functionPoor exercise tolerance

Attendant propelledwheelchair

50

TABLE 3Summary of patient characteristics : Cleveland VAMC

Quad.CodeNo .

Age atEva.(Yrs .)

Time fromInjury toEve .


Residencefrom Injuryto Eval . Physical condition

C* 50 14 yr . C5 Tumor Dairy farmer;Mathteacher

Privatehouseholdwith ownfamily

No upper-limb functionPoor trunk stabilityHead-neck control good,

motion poorFair shoulder shrugWears bifocalsSevere respiratory

dysfunction

C-2 56 26 yr . C4 Diving Militaryserviceman


No upper-limb functionGood trunk stabilityHead-neck control good,

motion poorFair shoulder shrugWears bifocalsNo fatigue problem

C-3 24 3 yr. C4 Injurycalf-roping

Militaryserviceman

Privatehouseholdwith ownattendant

No upper-limb functionPoor trunk stabilityHead-neck control good,

motion poorGood shoulder shrugAll day fatigue tolerance

C-4 25 8yc C4 Surfing Militaryserviceman


No upper-limb functionPoor trunk stabilityHead-neck control and

range good.Good shoulder shrugAll day fatigue tolerance

C-5 25 2yc C4 Diving Militaryserviceman


No upper-limb functionFair trunk stabilityHead-neck control and

range goodGood shoulder shrugAll day fatigue tolerance

C-6 60 3 yr . C2 Fall Salesman Privatehouseholdwith ownfamily

No upper-limb functionPoor trunk stabilityHead-Neck control and

range very poorNo shoulder shrugFatigue tolerance very

limited

Central cord injury

Electric wheelchairCan operate hand

&Vx pChandcontrolled electric

controllerwheelchair

Cannot bring hand to faceNeeds help with weight shiftsGood trunk stabilityGood shoulder shrugSitting tolerance S hours per day.

Customaryof assistivedevices

Electric wheelchair&V4PCchincontroller . Manualrecliner adjusted tosemi-recline at 60°

Electric wheelchair&w^Pochincontroller . Norecline . Prefersheadstick tomouthstick

Electric wheelchair& VAPC chincontroller . Cyclespower recliner fuliyfrequently

Electric wheelchair& VAPC chincontroller . Cyclespower recliner fullyfrequently

Electric wheelchair&MEDchin-controller . Cyclespower recliner fullyfrequently

Attendant propelledmanual recliningwheelchair alwayspartially reclined

C-7

56

8 mo.

C5

Fall

Tree

VAMCincom-plete

51

Journal of Rehabilitation Research and Development Vol . 22 No11985

also the person who was personally selecting, orienting,and providing immediate supervision of the quadriplegicsworking with the equipment . Because of competing obliga-tions, afu!!-timocommitmenttothispn4eotdurinOmuohof the evaluation period was not possible.

Seven quadriplegics, designated C-1 through C-7 inTable 3, were involved in the evaluation at CVAMC . As canbe seen by the table, patients at CVAMC spent much lesstime in evaluation compared with Richmond quadriplegics.All were inpatients and worked under the supervision ofthe Co-principal Investigators at CVAMC . Their work on theRAWT was conducted in a room of a size that did notprovide optimal approaches for docking a wheelchair atRAWT, but was otherwise adequate.

Summary of patient characteristics

Significant characteristics of the four quadriplegics whoevaluated the equipment in the Baltimore-Washingtonarea are summarized in Table 1 . Prior to becoming quadri-plegics, ail were healthy adult males . They were between27 and 31 years old at the time of evaluation and hadbecome quadriplegics 2 to 7 years earlier . All had differentoccupations . Between injury and evaluation they had livedin different types of places . At the time of evaluation nonehad any respiratory problems, and all could sit full uprightalmost all day without significant fatigue . Other than theirquadriplegia, all were entirely well during their evaluations.All had good head-and-neck control and range of motion.Only one, B-3, had any uppeNimbfunction . it was signifi-cant to the extent that it enabled him to have marginalnontrol of an electric wheelchair with a manual VAPCcontroller and mobile arm support, and to get some food tohis mouth with the help of a mobile arm support, a specialutensil secured to his braced hand, and a backstop on hisplate.

Significant characteristics of the nine quadriplegicswho evaluated the equipment at RVAMC are summarizedin Table 2 . All are male. They were between 21 and 49 yearsold at the time of evaluation and had become quadriplegicsbetween 5 months and 10 years earlier, Most lived in aVAMC from injury to evaluation . These were the ones whohad become quadriplegics most recently . R-2, R-3, R-4, andR-9 had respiratory fatigue problems adequate to handicaptheir participation in the program . Several had head andneck control rated as only fair or poor . R-1, R-5, R-6, and R-7had adequate hand function good enough so that they hadbeen provided with VAPC hand controllers on their electricwheelchairs.

Significant characteristics of the seven quadriplegicswho evaluated the equipment at CVAMC are summarizedin Table 3. All are male . They were between 24 and 60years old at the time of evaluation . One had become aquadriplegic 8 months earlier and had not yet left CVAMC.All others lived at home and had been injured much earlier.Although only C-1 and C-6 were rated as having respiratorydysfunction or a fatigue problem, C-3 . C4, and C-5, hadbeen provided with power recliner wheelchairs which theychose to cycle frequently . C'1 . C'2, C-3, and C-6 had poorhead-and-neck control or range of motion or both . Only C-7

had been provided with a hand controller on his electricwheelchair.

Site evaluation procedures and results

The results of the evaluation In the Baltimore-Washingtonarea are summarized in Table 4 . Since these tests wereconducted during the early engineering developmentphase, the system being evaluated did not include many ofthe advanced features found in the later models . SW-1 wasthe first quadriplegic to evaluate the Robot Arm outside ofthe laboratory and for more than a few hours . Initially, thearm was mounted on a small pedestal placed immediatelyadjacent to his wheelchair to simulate its operation as ifmounted on the wheelchair itself. it became evident thataccessibility to work objects and their functional usecould be achieved best if these objects were stored withinreach of the arm and if the operator and these objects werealways located at specific places . Robot Arm motionscould then be standardized with resulting reduction of taskexecution time.

BW-2 was accustomed to working in his bed, dictatingand telephoning technical reports based on literaturearranged on a large over-the-bed reading stand . He used asimple mouthstick for turning pages in preference to aheadstick or commercial page-turner . He demonstratedthe feasibility of performing a variety of useful tasksincluding self-feeding, management of a hand-set phonefor private conversations, typewriter paper loading, andmoving lightweight journals individually between a readingstand and a storage file. He confirmed the advantages ofmounting the Robot Arm on a motorized track on a specialworktable with specific locations for the operator and thework objects . He also confirmed the need to expedite taskexecution by programming joint motions for performanceof entire tasks . He confirmed that the table-mounted chincontroller was a prac1ioal device for control of the RobotArm. After this device was modified for electric wheelchairnontrnl and mounted on one, he tested it repeuiedly, bothindoors for maneuverability at slow speeds on smoothsurfaces and outdoors for control at higher speeds onmore irregular surfaces, He confirmed that it was a practi-cal device for wheelchair control and was less obstructivein front of his face than a VAPC controller . (He was pro-vided with a VAPC controller for comparison .)

BW-3 was the first quadriplegic to evaluate the RobotArm worktable system from an electric wheelchair using awheelchair-mounted controller with the features of inde-pendent operator docking and transmission of the controlsignals by infrared beam through a probe-and-socket inter-face. He used the system in his bedroom in his family'shouse. He found that the docking and control system andall opeoial task subsystems worked reliably in this environ-ment ; hovvaver ' ainc*the RAWT system occupied a relativelarge portion of his bedroom and since he had adequateresidual function in his upper limbs to carry out many ofthe same tasks with simpler assistive devices, he con-cluded that the system was not sufficiently useful to justifyit for him . He recommended mounting the Robot Arm on asmaller worktable with the options of either using it ex-clusively for self-feeding or connecting it to an additional

52

SEAMVNE& SOW/1E158ER : Evaluation Reports

table in order to use the other activity devices mounted on it.BW-4 used the small RAWT exclusively for self-feeding

almost daily for several months . He ate over 100 mealswith it . He stated that he enjoyed using it for this purpose,and made various recommendations for improvements inthe eating utensils . He was able to compare the functionof an APL chin controller with a VAPC chin controller on anE&J electric wheelchair. His testing environments werethe chronic hoapital in which he resided and a local com-munity college he attended . When operating the wheel-chair he preferred the APL controller to the VAPC controllerdue to its reduced obstructiveness in front of his face.

The results of the evaluation at RVAMC are summarized inTable 5 . At the conclusion of their evaluations four of thesenine quadriplegics, R,1, R-6, R-7, and R-8, rated the RAWT,overall, very favorably . They ate 91 meals with it . They usedterms such as, "Eating with it is very satisfying," "I feel!ibno4ed,^^ExceHent '^and^(amextrnmo!yeuUefiad .^R-7particularly expressed the desire to have one of these unitsfor his use at home . The responses of three other quadri-plegics, R-3, R-5, and R-9, are rated as fair to good . They ate86 meals with it . Each felt that the system was fulfilling tohim in the performance of one or more tasks, but felt thatthe performance of other tasks should be improved . Theemperienceoot the remaining two quadriplegics, R-2 andR-4, were too short for inclusion in an overall evaluation.

The two subsystems most valued by the evaluators werethose for assistance in self-feeding and personal computeruse. Recommendations for utensil modification and platewarming were made for improvement in self-feeding . Anon-line printer was felt to be essential for fulfillment in useof the personal computer . The Robot Arm was able to insertfloppy discs in the disc drive, but its performance in de-ploying literature for visual reference while the computerwas in use was felt to be unsatisfactory . The designershad originally introduced telephone management into therepertoire of RAWT tasks in order to offer the quadriplegicthe option of a more private conversation and one withbetter transmission fidelity than iapossible with aspeaker-phone. The early evaluators at RVAMC did not think thatthis option was worthwhile, due to the disadvantageseither of waiting for the Robot Arm to respond or of beingunable to respond because of being involved in executionof another task . Modifications to expedite response werecompleted in time for R-6 to evaluate the equipment . Thesucceeding evaluators at RVAMC were either pleased orsatisfied with the modified system as an improvement overaspeaker-phune.

Six of the nine quadriplegics at RVAMC evaluated theAPL chin controller for electric wheelchair control (Tables2 and 5) . R'1 and R-5 were VAPC hand control users . R-3and R-4 were accustomed to VAPC chin controllers . R-2and R-9 were accustomed to attendant-propelled wheel-chairs . All six concluded that the APL chin controller wasunsatisfactory due to the extent of head-and-neck controland range paquirad, steering difficu!1y, starting with ajo!tpoor abuse-resistance, and problems in adjustment .

The results of the evaluation at CVAMC are summarized inTable 6 . At the conclusion of their evaluations none ofthese seven quadriplegics rated the RAWT as useful . Onlytwo, C-1 and C-7, were willing to work with the RAWT forsignificant periods . C-1 worked with it for 78 hours over a3-month period and ate 25 meals with it . C-7 worked with itfor 25 hours over a 5-week period and ate 12 meals with it.He restricted his involvement with the RAWT exclusively toevaluating the self-feeding subsystem . Of the remainingava!uatoro . C-2 and C-3 participated for only 6 and 4 deys,respectively . C4. C'5, and C-6 participated for only 1 or 2days each . Experience at Richmond has indicated thatadequate test time must be provided and the quadriplegicvolunteer must get totally involved in order to sort out themerit of using this equipment to carry out needed tasks.This is believed to be a factor in the results from CVAMC.Problems noted included the large space required for theRAWT, poor performance re!iabi!ity, poor compatibilitywith bifocals, food spill, poor compatibility when used witha headstick rather than umuuthatiok . Recommendationsby the CVAMC evaluators included increasing Robot Armspeed and lift force capability, expediting phone answer-ing, refining eating utensils, and enabling more simultane-ous use of subsystems, e .g ., integrating telephone man-agement with literature handling or self-feeding.

Only three of the CVAMC quadriplegics, C-1 . C-2, andC-3, evaluated the APL chin controlier for control of anelectric wheelchair . All of these judged it as unsatisfactory.Problems cited were poor adjustability, lack of abusereoiatanoo, starting with a jolt, excessive sensitivity,excessive head-and-neck motion required, interferencewith Hover Lift transfer equipment, lack of a power reclineron the wheelchair provided, and lack of compatibility witha reclined position.

The most frequent and significant problems for theCVAMC evaluators in using RAWT and the APL chin con-troller were incompatibility of the system with a reclininguser, inadequacies of the APL chin controller as awheelchair-control device, and, alternatively, lack of acommercially available wheelchair controller suitable forRobot Arm control as a substitute . These reasons werealso perceived to be the principal causes for the limitedparticipation of this group of quadriplegics in this project.All CVAMC evaluators except C-2 and C-7 were recliningusers. C-7 was able to use a VAPC hand controller anddeclined to use the APL chin controller (Tables 3 and 6).Most of the problems identified at CVAMC have beenaddressed in the latest chin controller model now in test atAPL.

Collective experience and discussion

Involved in this clinical evaluation in three geographicalareas were 20 male quadriplegics between 21 and 60 yearsof age at evaluation . They ranged from 5 months to 26years between time of injury and evaluation . Their levels ofinjury ranged from C-2 to C-5 . individual accumulations oftime actually working with the equipment ranged from 1hour to over 100 hours ; 316 meals have been eaten by theseindividuals using the Robot Arm .

a3

Journal of Rehabilitation Research and Development Vol . 22 No . 11985

TABLE 4Summary of results : Baltimore-Washington area

Quad.Code Duration of Meals EnvironmentNo. Evaluation Eaten of Evaluation

BW-1 7 mo. -0— State Voc.100 hr. Rehabii.

Center

BW-2

1 yr .

4-5

Nursing3-5 hr.

demo home/day

meals3-4 days

EquipmentEvaluated

Negative Aspects Overall Impression Recommendations

Robot arm

Operation too slowwith worktable and tedious . Workkeyboard con-

objects nmaooes-tmxed by

sible to robot armheadstick

Potentially usefulif control effortreduced and acces-sibility

Reduce controleffort . (Program-mab!emogaovoperation).accessibility byadded freedom ofmotion . Workobjects on table top.

Robot Arm on

Operation too slowover-bed table .

and tedious . NotTabletop chin-

compatible withcontroller

operation fromwheelchair

_

_Elec . WC &

Unable to controlAPL chin-

Robot Arm fromcontroller

wheelchairfor WC

Potentially usefulif control effortreduce and opera-tion fmmwxeemha!renable_

Okay for wheelchaircontrol

Improvemable mode ofoperation . Enableinteraction fromwheelchair

___Enable this control-ler to operate RobotArm from wheelchairby telemetry

BW-3 6 mo. 4-5 Family'sdemo suburbanmeals house

RAWT with ail

Operation too slowwm»000l sub-

and tedious . RAWTsystems

requires large space

_____Elec . VVC&

Large space requireAPL chin-

ments for electriccontroller for

WC indoorsWC & RobotArm

Useful especially tohelp with PC &phone . Prefers self-feeding uyorthouia& mobile arm sup-port tnRAVvTifresidual functionpermits

____No problems withAPL chin-controllerfor RAWT controlfrom WC. Enjoyselectric WC withAPL chin-controlleroutdoors in summer

Speed up RAWTmovements . Reducespace requirements

None

Improve eatingutensils . Enableeating

BW-4

4 mo .

100

Chronicmeals disease

statehospital

Smu!RAvvT

Slow . Food fre-for self-

quently falls fromfeeding only

utensil . Food isavailable only frombowls.

Even with existinginadequacies it en-ables iself-feeding . A verypleasing experiencefor this individual.

Elec . WC &

Poor maneuverabilityAPL chin-

in restricted spacescontroller forWC & RobotArm

The electric wheel-chair with APL chin-controller is a greaimprovement over anattendant-propelledwheelchair and aVAPC chin-controllerwheelchair. mnex-perience witxotxe,wheelchair control-ler . Chin controllerworks well withRAWT.

54SEAMONE & SOHMEISSER: Evaluation Reports

TABLE 5Summary of results : Richmond VAMC

Quad.Code

Duration of Meals Environment EquipmentNo .

Evaluation

Eaten

of Evaluation Evaluated

Negative Aspects

25

RVAMC0T

RAWT with all Clutter of inputfacility

worktool sub-

control equipment,systems

slowness in hand-ling

_

_Elec . WC &

APL CC difficult toAPL CC for

position properly &WC & Robot

requires too muchArm

neck motion for WCcontrol

R-2

1 wk .

RVAMC OT

RAWT with all RAVTT slow, cumber-facility

mmrktom!oub~

some, impersonalsystems

_ _____

Elec . WC &

(Same as R-1)API.. CC forWC & RobotArm

RecommendationsOverall Impression

"Very favorable -Eating aspect verypositive due to free-dom fmmaido . per-sonal control overspeed and order ofeating ."

_APL CC is not satis-factory for WC con-trol but low profileis advantageous

Make small RAWTportable for selffeeding with friends.Reduce clutter ofinput controls.Speed up hxtele-phoning functiono.

Correct problems ofCC

Improvements (Notidentified)

Discard CC

"Good possibilities—RAWT makes self-feeding an optionbut prefers beingfed by his wife ."__

(Same as R-1)

R-3

4 wk . 13

RVAMC OT

RAWT with all

Re self-feeding;facility

worktool sub-

spoon fails to emptysystems

plate or bowls &spills, no mouthwipe during meas,bowl rotator slips,food cools . Doesnot re-index for usersagging in chair .

_

lot oftial—enjoyed eatingwith it . Learning touse not difficult"

(Same as R-1)

Improve

En-ableduring meal . Stopslipping of bowl

sim-ultaneous use ofsystems, e .g,tele-phone andreadingmaterial .

__(Same as R-1)

_Elec . WC &

(Same as R-1)APL CC forWC & RobotArm

Improvement ofexisting CC orselection of bettercontroller.

-0— nv4MC0T

RAWT with all Unable to use AR.facility

worktool sub-

CC for either controlsystems

of RAWT or electric WC due to range ofElec . WC &

neck motion re-APL

quired . WithdrawnWC & Robot

from project afterArm

first trial .

Range of motionquired for APL CCexcludes personswith very limitedneck motion fromcontrol of RAWT orElec. WC by thismeans.

R-5

2 mo . Improve spoon.Install food warmer.

16

RVAMC OT

RAWT with all (Same as R-3)facility

wmrkxool sub-systems

hrs1day ontypewritersystem)

_____

Elec . WC &

Difficult to steerAR_ CC for

accurately . ManyVvC& Robot

collisions.Arm

Satisfied with re-vised phone systemand typewriting sys-tem, appreciate theopportunity to havea private phone con-versation without anassistant . Operationof RAWT learnedeasily.

Wheel'- tir controlsystem easilylearned but lacksfinesse .

Improvements inWC controller

55


EquipmentEvaluated

RAWT with allmmrkum!oubsystems

TABLE 5 (continued)Summary of results: Richmond VAMC

Quad.Code Duration of Meals EnvironmentNo . Evaluation Eaten of Evaluation

R-6 5 wk14 RVxwCOTvw40 facilitymin/meal

Negative Aspects

Excessively largeportions of food onspoon, spill, adher-ence of some foodsto plate guard

Overall Impression

Recommendations

Expressed enthusi-

Refinements in foodasm and satisfaction

pick-up featureswith all RAWT func-tions especially forself-feeding

Food spillR-7

2mo

39

RYAMCOT

RAWT with allAv 40

facility

worktool sub~min/

systems (2-5meal

hrs/day on PC)

Described use of

Refinements in foodRAVVTaa^Ubmra'

pick-up features.Uun^ self-feeding

MekeRAvvTuval-^exoe!len^ . Enjoyed able for his use atall features . Asked

home.to take RAWT home.

R-8

/mo

13

RVAMC OT

RAWT with allfacility

womuon!sub-systems

Clutter of inputtrol equipment . Lowtelephone receivervolume

Extremely satisfiedwith RAWT including

sub-systems

(Same as R-7)Refine input controlequipment reduceclutter in front offace

(Same as R-7)

(Same as R-7)Especially pleased

Bite controller inwith self-feeding and

preference to chinenhancement of PC

controllerusage . Bite control-ler wasuueuisiv*improvement in re-ducing input con-troller

WC control by user

Improvements insatisfactory but ad-

chin controller ad-iuotabi!i/yamg abuae

justability and abuseresistance are un-

resistancesatisfactory

R-9

4mu

57

HvAwCOT

RAWT with allfacility worktoo!sub-

systems . Inputby API_ chin

re-placed by APLBite controller.

Elec . VVC&APL. CC forWC & RobotArm

Unacceptable clutterof control equipment

con-troller resolved withBite controller butBite controller re-quires attendantattachment to WC &hook up to RAWT

_Chin-controller ad-justments difficultfor attendants

The evaluations by the four quadriplegics in the BaltimoreWashington area indicated that the Robot Arm would bemost uaeful if mounted on a motorized track on a work-table with various work objects and with the operatorspecifically located with reference to a specific location ofthe arm. They confirmed that by using a manual mode ofoperation new tasks could be executed independently, andby using a programmed mode repetitive tasks could beexecuted both independently and with less effort . Theyconfirmed that the involvement of an attendant to interfacea quadriplegic in an electric wheelchair with the Robot Armcontrols could be avoided by using thewheelchair con-troller foroon1ro! input and a\n!eme1er!ng link to send thecontrol signals from the chair to the arm.

Among the nine quadriplegics who tested the equipmentat RVAMC, seven indicated that they found the equipmentgratifying to use, especially for self-feeding. Generally,these were ones able to sit upright most of the day withiittle fatigue and who had good head-and-neck control andange . Among the seven quadriplegics who tested theiyetem at CVAK8C, none found the system useful due to a

number of problems delineated in Table Si Since most ofthese individuals did not choose to use the API_ chincontroller, most had interface problems with their ownwheelchair controllers and recliner . Five of the CVAMCusers reclined in their wheelchairs.

The most frequent and significant problems were:

1. Incompatibility of the system with a reclining user;

2. Inadequacies of the APL Robot Arm chin controllerfor wheelchair control ; and

3. Inadequacies of any commercially available wheel-chair

There are certain considerations that need to be takeninto account in solving these problems . To allow the RobotArm to bring objects to useful locations near the user'shead or food to his mouth, programs have been writtenwith the assumption that, for the duration of the user'sactivity with the RAWT, his head would be vertically

56

SEAMONE & SCHMEISSER : Evaluation Reports

TABLE 6Summary of results : Cleveland VAMC

Quad.Code Duration of Meals Environment EquipmentNo . Evaluation Eaten of Evaluation Evaluated

Cw 3mo 25 CVAMC OT RAWTwith allfor facility wmm000!aub-78 hr . systems

___Elec . vvC&

Chin controller ad-APL CC for

justments difficultWC & Robot

for attendants.Arm

Abuse resistanceunsatisfactory.Starts with a jolt.

Negative Aspects

Overall Impression

Recommendations

re-quired . Performancereliability . Difficultto see monitor withbifocals

This equipment

Enable to lift heaviervery vsefvl to this

books . Enable oim,user .

ultaneous use ofsubsystems,telephone and read-ing material . Refine

__ eating utensils._

Pm«emxPLchin

(None)controller to VAPCchin controller asless obstructive

4

CVAMCQT

RAWT with all Food spill, APL CCfacility

worktool sub-

on table

'systems

ible with VAPC CCon WC. Difficult tosee monitor withbifocals . Nmcom-patible mith headstick.

_____Bec. vVC&

(Same as C-1)APL CC for

Interferes withWC & Robot

Hoyer-Lift transfersArm

C-3 0deymfor

CVAMC0facility

RAWT with allworktowl sub-

21 hr . systems

_One trialoutdoors

_____Elec . WC &APL CC forWC & RobotArm

C-4 2 daysfor4h~

0 CVAwC0facility

RAWT withonly computersubsystem

C-5 1 dayfor0xr .

0 CvAMCOTfacility

RAvVTwith allworktool sub-systems

Not Much use for

Make Robot Arm freerobot arm in present

standing, roller-form. mountod, height-

adjustable, voicerespondent .

_APL chin controller

Improve controlleris not satisfactory

mounting systemfor WC control

and smooth outresponse

C-2

4dayofor23 hr .

RAWT not compat-

(Not given)ib!owith semi-reclined positionRAWT requires anattendant to movecontrollers if userprefers another typeWC chin controller,e .g ., VAPC arm istoo slow

___

_

Not equipped with

Outdoor functionpower recliner

good . Too sensitiveindoors .

Al! programs shouldend in phone parkmode for quick re-sponse . Increasearm lift and speedcapabilities.

Morse code keyer

(None)too slow

(None . Declinedfurther participation)

RAWT requires an

(Not given)attendant to movecontrollers if userprefers another typeWC chin controller,eg,mED. RAWTnot compatible withcycling power re-diner WC .

Increase arm grouplift & speed capabil-ity . Expedite phoneanswering . (Declinedfurther participation)

57

Journal of Rehabilitation Research and Development Vol . 22 No. 1 1985

TABLE 6Summary of results : Cleveland VAMC

Quad.CodeNo .

Duration ofEvaluation

MealsEaten

Environmentof Evaluation

EquipmentEvaluated

C-6 1 dayfor6 hri

0 CVAMCOTfacility

RAWT with ailmmrkum!sub-systems

C-7 5 wk.forI hrper5 daysper wk .

12 CvAmCOTfacility

RAWT withonly s g f-feeding sub-systems

oriented, facing the table, located as close as possible toit, and within a few centimeters of a particular heightabove the level of the table . A recliner violates theseassumptions . Modification of existing programs and re-configuration of the docking ports can largely solve thesecurrent problems of RAWT incompatibility with areu!iner.

In an attempt to work with the reclining user, one choicewas to evaluate a table-mounted controller independent ofthe wheelchair controller . To test that approach a tablemounted controller was evaluated . It could be adjusted forany given initial recliner position . its negative aspects(adjustments, more equipment in front of the user, and theneed for the help of an attendant to initially dock, to adjustthe controller, and even to permit the user to leave theworktable) defeats the basic concept of independence thesystem is supposed to provide . Since no commerciallyavailable wheelchair chin controllers are compatible withthis system, the APL chin controller was redesigned topermit it to work at various recliner positions . It has alsobeen made substantially more abuse resistant, and moreadjustable by the quadriplegic . This new controller modelwill undergo clinical testing early in 1985.

Summary And Future Plans

The clinical evaluation of the RAWT system over the past 2years at the two participating VA Spinal Cord injury Serviceshas contributed materially to the enhancement of thesystem's capability . These tests have shown that thesystem holds promise of providing a measure of indepen-dence to certain high-spinal-cord-injured persons . Capabil-ity for ue!f'fauding, use of a computer, and use of a tele-phone appeared to be its most popular features . Furtherclinical testing will be conducted for at least the next 18months, to determine the extent to which modifications toaddress existing problems have been successful.

Some of the new engineering features include a newchin controller which addresses many of the complaints ofthe users, simultaneous multiaxis motions,

corn-

pensation for the patient's change in posture during theday, and development of self-grooming capability . Thesechanges, along with the basic design, are now goingthrough the transition into a production prototype modelfrom which a oommeruial manufacturer is expected tomake these systems available on a broader scale to theveteran who needs and can benefit from this system m

Acknowledgement -- Much credit must be given to Ms.Karen Gurtier, Clinical Evaluator for the project at VAMedical Center, Richmond, Virginia, and Mrs . Carol Harley,Clinical Evaluator for the project at VA KXedica! Center,Cleveland, Ohio, for organizing and conducting clinicalevaluation at their centers.

Active participation of the quadriplegic volunteers atthese centers has been a critical contribution.

And finally, much credit goes to Mr . J . H . Loveless andMr. W. Schneider of the Applied Physics Laboratory fortheir contribution to the design and implementation of therobot arm/worktable system.

References1. Peizer s and Mason O p A S»ve

for the Physically Handicapped, IRIA, Sept 4-6,1978,pp 305-318.2. Taylor HJ : The Development of Two Wheelchair Manipulator

Systems . Ibid pp 117-131.3. G dte\+u

al . : The Sparticus Telethesis : Manipulator Controland Experimentation . Ibid pp 79-100.

4. Roessler H et al . : The Medical Manipulator and Its AdaptedEnvironment : A System for the Rehabilitation of the SeverelyHandicapped . Ibid pp 63-77.

5. Bejczy AK : Jet Propulsion Laboratory Voice Command Systemfor Motion Control . lbid pp 103-116.

6. Schneider W, Seamone W, and Schmeisser G : A Microprocessor-Controlled Arm Allows Self-Feeding for a Quadriplegic . ProcIEEE Computer Society Workshop : Application of PersonalComputers for the Handicapped . Johns Hopkins UniversityApplied Physics Laboratory . April 2-3, 1980, pp 31-36.

7. Seamone Wand Schmeisser G : New Control Techniques forWheelchair Mobility . APL Technical Digest 2(3), 1981.

Negative Aspects

Overall impression

Recommendations

RAWT not compat-

(None . ParticipationInade-

position required by

quatefatigue ^o!epmiavoer

ance)

Food adhering to

' interest-plate guam . Spoon

ing to work with butaccepts too much

not practical for himfood and/or spills .

unuaeforuo!f-Cannot dock pro-

feeding.perly using his man-ual VAPC controller

Assortment ofspoon designs withRobot Arm pick-upcapability . Enableproper docking withmanuai VAPC con-troller and interfacewith RAWT table topcontroller withoutattendant help.

Download - EVALUATION REPORT: Early Clinical Evaluation of a Robot ... · include chin or head motion, eye tracking motion, and voice. in early examination of control alternatives at the APb,

Top Related