Automatic Motion Recognition and Skill Evaluation for Dynamic Tasks

Abstract

Along with the development of computerized training systems has come increasing demand for objective methods of evaluating skill. For scenarios where skill is related to specific user motions, we have developed a system using hidden Markov models (HMMs) to recognize motions performed in a virtual environment with a haptic device. The output of this system is a list of motions used during completion of a task. We first explore which observations are most important for accurate recognition of user motions. Second, we use a sequence of motions to evaluate skill by analyzing the performance of individual users over multiple repetitions of a dynamic task. The results reveal that the system is able to achieve consistent recognition with a wide variety of observations.