Haptics in Human-Robot Interaction
Current Researchers: Naomi Fitter
As robotics applications shift from repetitive factory labor to everyday environments, robots must possess more perception and action capabilities that enable them to communicate with humans in a natural manner. Since the dawn of robotics, researchers have mandated that robots pose no risk of physical harm to nearby people. A small human-robot interaction (HRI) subfield called physical human-robot interaction (PHRI) centers on situations that involve direct physical contact, such as human-robot handshakes and extending human mechanical power using robot strength . More recently, socially assistive robotics, the use of robots to aid or motivate humans, has become another prominent HRI subfield. Social robots have a vast potential to affect the feelings and behaviors of humans. Although social interaction, rather than physical contact with humans, is the main focus of socially assistive robotics, PHRI may also come into play. Cognitive science studies indicate that touch, especially with the hands, is a dominant part of the mental model of a human user, and that manipulation of physical objects, as well as gestural interaction, enhances the learning and communication of humans . Accordingly, in the future, the most effective socially assistive robots will likely be those that also employ PHRI.
Because movement and physical contact are known to play important roles in human-human interactions, I have become very interested in the nascent overlap between socially assistive robotics and PHRI. Through direct interaction with humans, social-physical robots may be able to help serve the needs of the world’s growing population, for example by assisting human teachers in the class- room or caring for the elderly at home. The small amount of existing research in this area elucidates that robots must exhibit ethical and emotional behaviors that are acceptable to the human(s) with whom they are interacting, linking cognitive science and robotics in the field of HRI . Studies that use social robots to teach the norms of play to children with developmental delays  or help socially normal students practice English skills exemplify this connection . Further examples in the literature describe successful robot use in the classroom to teach special topics such as recycling and health education. Robots also teach or help teach English as a Second Language in a small number of classrooms in Asia . Studies like these give support to the idea that children could benefit from interacting with robots in educational settings.
Figure 1: Social-physical interaction between the PR2 and a human.
High-fives and hand-clapping games are a viable entry point into simple teaching applications in the emergent area of social-physical robots. These activities are especially interesting because they are simple HRI activities with emotional implications, like the triumph of a high five or the joy of a hand-clapping game. They are also low-risk and easier to control than some of my target applications for social-physical robots, such as use in the classroom. Another benefit of hand-clapping is that it involves the hands, an essential body part especially sensitive to touch and pivotal to the field of haptics. Our lab has substantial experience interpreting tactile feedback, and we are eager to apply these techniques to making social-physical HRI “feel right.” To begin exploring social-physical robotics, I have done experiments and observations focused on un- covering if humans enjoy interacting with clapping robots and how humans interact with each other in key hand-clapping activities.
Figure 2: An example pair of experiment subjects.
Preliminary experiments and observations involved Willow Garage’s PR2, a humanoid robot that was readily available and shares many traits with my vision of classroom assistant robots. This two-armed, human-scale robot is too expensive for applications in public schools, but its preexisting wrist accelerometers and fingertip pressure sensor arrays make it a viable first platform. I programmed the PR2 to play simple hand-clapping games such as “Double Double This This” with a human and then exposed different volunteers and tour groups to this social-physical robot, observing their reactions and interest, as depicted in Figure 1. A key technical challenge is for the robot to react quickly to hand contact; while computer vision in this application would be slow and unreliable, we were able to reliably and quickly detect hand-gripper impacts using transient accelerometer signals. The initial robot programming was enough to capture the humans’ attention, although the robot joint controller that the initial program used was not fast enough to accomplish convincing robot movement. This aside, the results of these initial observations were promising, and my next step is to develop a faster and more convincing controller with the PR2 (or other more appropriate robot) to hold the attention of the human with which it is interacting.
Figure 3: A viable future application of social-physical HRI.
To help improve the quality of robot’s movements, I am now studying the movement of humans in hand-clapping activities. I recruited eleven pairs of people between 22 and 48 years of age and measured their high-five hand positions and accelerations using one Ascension trakSTAR 3D magnetic tracking sensor and one Sparkfun MMA7361 triple axis accelerometer breakout board affixed to the back center of one of each subject’s hands, as shown in Figure 2. Signals were measured throughout trials involving different high-five tempos and leadership of the tempo by a given subject. This experiment provided me with abundant information about how a high-fiving robot should move, as well as survey information on how interested people would be in doing this type of activity with a robot. As described in my HRI 2014 Late-Breaking Report, the overwhelming response from subjects was that the activity was enjoyable and they wanted to do similar activities with a robot. I am currently working through the collected information to design human-inspired robot movement that will make my target HRI activities more enjoyable and captivating for human subjects by con- trolling the PR2 gripper position, velocity, and collision timing to match patterns that human hands displayed throughout the high- fiving activities. This analysis could help to form a cross-platform framework to describe social-physical HRI using dynamics.
 R. C. Arkin, M. Fujita, T. Takagi, and R. Hasegawa. An ethological and emotional basis for human–robot interaction. Robotics and Autonomous Systems, 42(3):191–201, 2003.
 J. Han and D. Kim. r-learning services for elementary school students with a teaching assistant robot. In Proc. ACM/IEEE Human-Robot In- teraction (HRI), pages 255–256, 2009.
 A. M. Howard, H. W. Park, and C. C. Kemp. Extracting play primitives for a robot playmate by sequencing low-level motion behaviors. In Proc. IEEE International Symposium on Robot and Human Interactive Communication, pages 360–365, 2008.
 S. Ikemoto, H. B. Amor, T. Minato, B. Jung, and H. Ishiguro. Physical human–robot interaction. IEEE Robotics and Automation Magazine, 2012.
 T. Kanda, T. Hirano, D. Eaton, and H. Ishiguro. Interactive robots as social partners and peer tutors for children: A field trial. Human- Computer Interaction, 19(1):61–84, 2004.
 S. R. Klemmer, B. Hartmann, and L. Takayama. How bodies matter: five themes for interaction design. In Proc. ACM Conference on De- signing Interactive Systems, pages 140–149, 2006.
Naomi T. Fitter and Katherine J. Kuchenbecker. Analyzing Human High-Fives to Create an Effective High-Fiving Robot. Late-breaking report in Proc. 9th ACM/IEEE International Conference on Human-Robot Interaction, Bielefeld, Germany, March 2014.