Robot programming by human demonstration: the use of human variation in identifying obstacle free trajectories

Programming by human demonstration is an intuitive method of robot programming, in which the programmer demonstrates how a task is performed using a human/robot teaching device that measures human motion, and the data gathered is used to generate the robot program. A direct duplication of the demonstrated trajectory would result in unnecessary robot motion due to human “wiggles” and unintended motion. To identify a more satisfactory robot trajectory, a method is presented that uses multiple demonstrations of the same task. Variation in human trajectories between trials is attributed to human inconsistency and is used to define an obstacle free region, by applying the Jordan curve theorem. The shortest path within the obstacle free region is determined, resulting in a shorter robot path than any of the demonstrations. Thus the presence of human inconsistency is used to improve robot performance. The analysis is restricted to planar translational motion