Study stopped due resource interruption.

Our study goal is to assess natural patient-therapist interactions in order to map such human-human activities to robot-human interactions. Critical to accomplishing this mapping will be determining the feasibility of a humanoid robot interacting with a patient in a more intuitive and flexible way, while concomitantly investigating the issue of safe contact and release.

Stroke survivors with varying levels of motor deficit impairment in their upper extremity (UE) will be assessed using standardized clinical scales for cognition and motor impairment. Following assessment, the patient and therapist will complete a therapy session consisting of activities of daily living (ADL).

During the therapy session, visual 3D motion tracking will be used via Kinect, audio will be recorded, and basic measurements (i.e. the distance between a participants wrist and elbow) will be taken. After the patient-therapist interaction has been mapped, members of the research team will build computer based models of the therapist and patient using the insights gained from the human-human interaction. The models developed in this proposed study will be used as a template for programming safe and intuitive humanoid-patient interactions for future study. The model of the therapist will be implemented on the Robot (w/o a patient involved) and in a simulation environment where it will be tested with a computer-based model of the patient.

Inclusion Criteria:

• Stroke survivors 18 years of age with hemiplegia and varying levels of impairment
• Healthy subjects greater than 18 years of age with no motor disabilities

Exclusion Criteria:

• Under 18 years of age