Quantitative human and robot motion comparison for enabling assistive device evaluation

Abstract

A promising new application area for humanoid robots is in the area of assistive device testing. Humanoid robots can facilitate the experimental evaluation of assistive devices by providing repeatable, measurable, human-like motion while removing the difficulties and potential safety hazards associated with human trials. To ensure that the humanoid robot is providing a valid test platform, the robot must move in a similar way as a human while wearing the assistive device. This challenge is made difficult due to the inherent variability in human motion. In this paper, we propose an approach for a quantitative comparison between human and robot motion that identifies both the difference magnitude and the difference location, and explicitly handles both spatial and temporal variability of human motion. The proposed approach is demonstrated on data from robot gait and sagittal plane lifting.