Kinesthetic Interaction Kyle B. Reed, Michael Peshkin, Mitra J. Hartmann, J. Edward Colgate, and James Patton Abstract— In physical and occupational therapy two people interact through force and motion. Other common examples of this interaction include lifting and moving a bulky object, teaching manual skills, dancing, and handing off a baton or a drinking glass. These tasks involve kinesthetic interac- tion, a communication channel distinct from spoken language and gestures. Understanding kinesthetic interaction should be important in designing robots to assist with physical and occupational therapy. In this paper we describe our experiments on kinesthetic interaction between two people cooperating on a 1 degree of freedom task. We characterize the interaction forces between the two people, dividing them into a productive “net force” and an orthogonal “difference force.” Our results suggest three effects (1) an emergent specialization of the two participants into different roles, (2) an oscillation of forces at about 8 Hz, and (3) a steady force in opposition to one another that could be analogous to co-contraction in an individual. I. INTRODUCTION A basic form of human interaction is the physical co- operation necessary to perform a manual task with others. Physical cooperation represents a communication channel distinct from facial expression, gesture, and spoken language, yet one that has been much less studied. We expect that a significant channel of dyadic communication should be through forces and motions, applied either directly to one another’s limbs or via a mutually grasped object. Such mechanical cooperation is involved in a wide variety of tasks including moving objects too large or heavy for an individual, tasks requiring two hands for object stability while other hands perform manipulative movements, and di- rect physical guidance of another person’s limbs for teaching and learning. Whether by direct limb-to-limb coupling or coupling through a mutually grasped object, these activities create a kinesthetic interaction between two individuals in which one person feels and reacts to the forces and mo- tions of the other person. Cooperative mechanical control is also important for tasks whose complexity may exceed the cognitive or motor capacity of an individual, as in piloting of remote vehicles and shared control teleoperation for minimally invasive surgery. Manuscript received February 11, 2005. This work is supported by NSF grant ECS-0433948. K. B. Reed is with the Laboratory for Intelligent Mechanical Systems, Northwestern University, Evanston, IL, 60208, USA (847-467-1070; e-mail: reedkb@northwestern.edu). M. Peshkin is with the Laboratory for Intelligent Mechanical Systems, Northwestern University, Evanston, IL, 60208. M. J. Hartmann is on a joint appointment with Mechanical and Biomed- ical Engineering, Northwestern University, Evanston, IL, 60208. J. E. Colgate is with the Laboratory for Intelligent Mechanical Systems, Northwestern University, Evanston, IL, 60208. J. Patton is with the Rehabilitation Institute of Chicago, IL, 60611. Our area of interest, lower limb rehabilitation, also in- volves haptic interaction between two people: patient and therapist. Several previous efforts have tried to create a robot that itself performs aspects of physical therapy, such as the Lokomat [1]. In contrast to such efforts, our research strives to create a device that assists the physical therapist in physically demanding tasks. Hands-on interaction with patients is highly valued by physical therapists; we aim to enhance it, not replace it. The hands-on interaction between physical therapist and patient involves the communication of muscle tone, force, and motion to the therapist, and the selective delivery of force and motion by the therapist. In this and further studies, we will explore the interaction between subjects to explore how two people physically cooperate, compromise, and guide one another, and how machine-generated forces and motions can enter into the human-human physical conversation. II. RELATED WORK Shared manual control occurs in learning to operate some kinds of equipment, most critically in the teaching/learning of helicopter control. Wegner & Zeaman [2] were interested in applications of the dyadic effects they were studying, and discussed automobile and aircraft dual controls. They also identified some dual control tasks in everyday life, including the two-handled saw, the see-saw, and the balancing of a tandem bicycle. Shaw [3] adds couples dancing. Other ev- eryday activities exhibiting dual control are maneuvering and positioning a board or other object and placing a bed sheet on a bed. In some of these examples, the participants establish a working relationship in which they must compromise (e.g. positioning the bed sheet symmetrically) or in which they must divide authority according to task phase. Sallnas and Zhai [4] performed a study with a haptic system simulating the handoff of an object on a computer screen. They measured the time it takes for one person to hand off objects to another person (haptically simulated) within certain spatial targets. Here the two people are not physically connected, they feel only the sensations the haptic device can simulate. Similarly, two athletes pass a baton without looking and rely completely on their haptic sense to determine when the exchange is complete. Both Elhajj et al. [5] and Hespanha et al. [6] have explored humans communicating in a teleoperational haptic environ- ment over the internet. Barnes and Counsell [7] explored these same issues locally, much like Sallnas and Zhai’s setup with the handoff computer simulation. In each of these cases, the two people were not physically or rigidly connected. Proc. of the 2005 IEEE 9th International Conference on Rehabilitation Robotics (ICORR) June 28 - July 1, 2005, Chicago, IL, USA 569