Skill Learning from Observations


This paper presents a new approach for representing and regktering skills basedon state transition graphs. First, we generate clwters of feasible state transitions from the experimental data such that the union of these clusters represents the feasible srate transition region of the system. Globally Competitive and Locally Cooperative (GCLC) algorithm k… (More)
DOI: 10.1109/ROBOT.1994.351071


8 Figures and Tables