Championing Research ThroughDesign in HRI

Michal LuriaHuman-Computer InteractionInstituteCarnegie Mellon UniversityPittsburgh, PA, USAmluria@cs.cmu.edu

John ZimmermanHuman-Computer InteractionInstituteCarnegie Mellon UniversityPittsburgh, PA, USAjohnz@cs.cmu.edu

Jodi ForlizziHuman-Computer InteractionInstituteCarnegie Mellon UniversityPittsburgh, PA, USAforlizzi@cs.cmu.edu

AbstractOne of the challenges in conducting research on the inter-section of the CHI and Human-Robot Interaction (HRI) com-munities is in addressing the gap of acceptable design re-search methods between the two. While HRI is focused oninteraction with robots and includes design research in itsscope, the community is not as accustomed to exploratorydesign methods as the CHI community. This workshop pa-per argues for bringing exploratory design, and specificallyResearch through Design (RtD) methods that have beenestablished in CHI for the past decade to the foreground ofHRI. RtD can enable design researchers in the field of HRIto conduct exploratory design work that asks what is theright thing to design and share it within the community.

CCS Concepts•Human-centered computing → Interaction design pro-cess and methods; User centered design; Scenario-based design;

Author Keywordsresearch through design; design methods; human-robotinteraction; user enactments; CHI; HRI

IntroductionWe agree with the workshop organizers that there are manychallenges in working across the CHI and Human-Robot

Interaction (HRI) communities. This workshop paper ad-dresses the challenge of using exploratory design methodsthat are common in CHI to conduct design research on HRItopics. In particular, we discuss the method of Researchthrough Design (RtD).

RtD has been established in CHI after much effort from de-signers in the community. Their argument was that throughthe process of designing and making, designers can gener-ate new knowledge and contribute to the research commu-nity [15, 10]. Today, RtD has become a valid form of inquiryand has a critical role in the design of human-computer in-teractions. One of the benefits of RtD is that it looks into“making the right thing”, as opposed to “making the thingright” [15, 3]. In other words, it allows to research what todesign, and not what is the best way to design somethingthat might have little value.

In the field of Human-Robot Interaction (HRI), much of thedesign research is focused on how to make a thing right.With some notable exceptions, the majority of design pa-pers published in HRI create and test a particular robot taskor function. If the robot was successful in performing andcommunicating that particular task, then the design is suc-cessful. This approach is very similar to the approach thatwas common in the CHI community before designers estab-lished RtD methodology.

We argue that the HRI community could greatly benefitfrom adopting exploratory design research methods fromCHI. By doing so, the community could expand the bound-aries of HRI and robot design, and gain knowledge aboutwhen designing robots is the right thing to do.

Research Through DesignOne of the important aspects of Research through Design(RtD) and other methods that make use of design thinking

Figure 1: The Drift Table explores opportunities for “ludicinteraction” with technology in the home. The table presents aerialmaps that move according to the weight placed on the table. Theprototype looks at curiosity and playfulness in technology for thehome environment [7].

is the ability to reconsider underlying assumptions. Previ-ous work shows that in a community, new ideas are likelyto adhere to elements from previous solutions and causefixation to solutions that are not useful in new contexts [4].

For instance, early work in CHI presented Whisper, a wear-able device that allowed one to insert their fingertip intotheir ear canal to communicate in a phone call [8]. Althoughthe presented technology was very impressive and pro-vided a technical contribution, the design was somewhatfixated on the assumption that people would not want towalk empty handed and seem as if they were ”taking tothemselves”. In the perspective of time, we learn that thisassumption was incorrect, and that people do not mind talk-ing on the phone without a visible device. This realizationbrings a new range of designs and possibilities. Throughexploratory design methods that re-frame and challenge as-

sumptions, researchers can help avoid fixation and expandthe boundary of acceptable designs in a field.

An example of a RtD project from the CHI community isThe Drift Table [9]. The drift table is a coffee table that dis-plays aerial photography in a small window at its center,according to the distribution of weight on it (see Fig. 1).Through the process of designing the artifact, the authorsexplored how people perceive technology for the homeand what are the opportunities to design for “ludic expe-riences” [9]. Another example is Odom et al.’s work onteenagers’ virtual possessions. The authors conducted aseries of exploratory interactions with teenagers, observedtheir relationships with physical and virtual possessions,and examined the design space of designing virtual pos-sessions to be more meaningful [13, 12].

In the field of HRI, Auger used a Speculative Design ap-proach [7] to explore why robots are not becoming domesticproducts, and what might help them become such [1]. Onerobot presented in the paper observes clouds and recog-nizes human faces in them (see Fig. 2). Through the inter-action, the robot creates an emotional and playful connec-tion with the user. Another project created IdleBot, a robotthat moves, but is barely interactive. This design questionswhether and when robots should be engaging [14].

The Challenge of Using RtD in HRIWe recently conducted a RtD study that explored scenarioswith multiple robots. We were interested in the question ofwhen should robots’ social presence (their presented entityand personality) move from one body to another (we callthis action re-embodiment), and when should the user bepresented with a new personality altogether [11]. Our goalwas to gain knowledge about the design space of sequen-tial interaction with robots and agents across multiple loca-

Figure 2: Auger used speculative design to explore why robotsare not adopted as domestic objects, and suggested somealternatives. For example, a robot that looks at the clouds andnotifies the user when there is a human-like face passing by [1].

Figure 3: We used RtD to explore when robots’ social presences might move from body to body, share a body, or each have a body of theirown. Through Speed Dating, we tested a set of selected environments and contexts (a home, a car, a hospital and a public service).

tions and over time. While previous work looked at whetherpeople perceive the movement of a social presence fromone body to another [6], it did not examine in what contextsand uses is this behavior valuable.

As this is a complex space of exploration that needs toprobe interactions that do not yet exist, simply testing themin a lab setting will not do. Thus, we turned to the exploratoryRtD method of Speed Dating with User Enactments [5]. Us-ing this method we emerged participants into scenarios thatare “brought-to-life” using actors, prototypes and props toallow participants firsthand experiences of possible situa-tions in the near future. Participants were then interviewedabout their experiences. Just like romantic speed dating,participants got ’a sip’ of many different scenarios throughuser enactments. By the end of the experience, they mightnot have learned much about any single scenario, but theyare more likely to have a better sense of their own needs,desires and values on the topic.

The main challenge of introducing such work in the HRIcommunity is that the method is not structured or con-

trolled. It is subjective and requires the researchers to fre-quently make design judgments, which is quite typical forRtD. Reviewing this work from a “user study” perspective islikely to point out plenty of methodology gaps, yet the goalof this type of work is not to systematically cover the en-tire space or reach internal validity. RtD is unstructured bydefinition—the flexibility allows exploration of an unknowndesign space and complicated “wicked” problems, in whichthe contexts and design choices are unlimited [2].

We currently share this work in the CHI and DIS communi-ties, as they are more familiar with RtD approaches and thismethodology. However, the topic of research is more likelyto be of interest to the HRI audience. In the near future,we would like to examine RtD for HRI to better understandhow it would fit into the field, and how to encourage HRIresearchers to accept RtD as a valid design contribution.We hope that doing so will enable researchers to conductdesign work that questions underlying assumptions andexpands the boundaries of HRI.

ConclusionRtD allows designers and researchers to discuss what isthe right thing to design. This question is not frequentlyasked in the field of HRI, but might be a critical one for thedevelopment of its design perspective. As a field, we canuse RtD to understand what is a robot, and what are theareas of exploration for designing them.

As the CHI community has already made significant effortstowards accepting RtD as a valid design contribution, in-stead of starting the argument from scratch, HRI designerscan join forces with CHI designers to promote RtD in HRI.One of the ways to do this is to discuss what RtD methodsmight be more relevant for HRI, and what are the contribu-tions they could bring. This can also help designers in HRIunderstand if there are adaptations that need to be made toRtD methods for HRI. Finally, publishing RtD on HRI topicsin CHI is likely to create a growing group of researchers thatare on the intersection of the fields, and allow RtD to enterHRI more naturally.

REFERENCES1. James Auger. 2014. Living with robots: A speculative

design approach. Journal of Human-Robot Interaction3, 1 (2014), 20–42.

2. Richard Buchanan. 1992. Wicked problems in designthinking. Design issues 8, 2 (1992), 5–21.

3. Bill Buxton. 2010. Sketching user experiences: gettingthe design right and the right design. MorganKaufmann.

4. Evangelia G Chrysikou and Robert W Weisberg. 2005.Following the wrong footsteps: fixation effects ofpictorial examples in a design problem-solving task.Journal of Experimental Psychology: Learning,Memory, and Cognition 31, 5 (2005), 1134.

5. Scott Davidoff, Min Kyung Lee, Anind K Dey, and JohnZimmerman. 2007. Rapidly exploring applicationdesign through speed dating. In Int’l Conference onUbiquitous Computing. Springer, 429–446.

6. Brian R Duffy, Gregory MP O’Hare, Alan N Martin,John F Bradley, and Bianca Schon. 2003. Agentchameleons: Agent minds and bodies. In Proceedings11th IEEE International Workshop on ProgramComprehension. IEEE, 118–125.

7. Anthony Dunne and Fiona Raby. 2013. Speculativeeverything: design, fiction, and social dreaming. MITpress.

8. Masaaki Fukumoto and Yoshinobu Tonomura. 1999.Whisper: a wristwatch style wearable handset. InProceedings of the SIGCHI conference on HumanFactors in Computing Systems. ACM, 112–119.

9. William W Gaver, John Bowers, Andrew Boucher, HansGellerson, Sarah Pennington, Albrecht Schmidt,Anthony Steed, Nicholas Villars, and Brendan Walker.2004. The drift table: designing for ludic engagement.In CHI’04 extended abstracts on Human factors incomputing systems. ACM, 885–900.

10. Ilpo Koskinen, John Zimmerman, Thomas Binder,Johan Redstrom, and Stephan Wensveen. 2011.Design research through practice: From the lab, field,and showroom. Elsevier.

11. Michal Luria, Samantha Reig, Xiang Zhi Tan, AaronSteinfeld, Jodi Forlizzi, and John Zimmerman. 2019.Re-Embodiment and co-embodiment: exploration ofsocial presence for robots and conversational agents.In Proceedings of the Designing Interactive SystemsConference. ACM.

12. William Odom, John Zimmerman, Scott Davidoff, JodiForlizzi, Anind K Dey, and Min Kyung Lee. 2012. Afieldwork of the future with user enactments. InProceedings of the Designing Interactive SystemsConference. ACM, 338–347.

13. William Odom, John Zimmerman, and Jodi Forlizzi.2011. Teenagers and their virtual possessions: designopportunities and issues. In Proceedings of the SIGCHIconference on Human Factors in computing systems.ACM, 1491–1500.

14. Caroline Overgoor and Mathias Funk. 2018. IdleBot:exploring the design of serendipitous artifacts. InProceedings of the 19th International ACMSIGACCESS Conference on Computers andAccessibility. ACM, 105–110.

15. John Zimmerman, Jodi Forlizzi, and Shelley Evenson.2007. Research through design as a method forinteraction design research in HCI. In Proceedings ofthe SIGCHI conference on Human factors in computingsystems. ACM, 493–502.