Applying visual programming to robotics

Abstract

A visual robot programming environment called PROVIDE (programming robots by visual dialogue) is described. PROVIDE offers much of the flexibility and intuitive feel of programming by visual demonstration without requiring complex scene analysis of task planning: complex scene analysis is short-circuited by having most of the demonstration performed not in the real world but within the digitized images seen by the robot's vision system. The programmer/teacher indicates directly in the digitized images robot actions, constituent parts, grip-points, relative orientations, approach routes, insertion-axes, etc. Learning the visual sensing needed to perform the taught task is an integral part of this demonstration. The contrast between this and pure visual demonstration is illustrated. Complex task planning is avoided by providing an environment which enables human problem-solving skills to support, at a high-level, the robots understanding and execution of a task.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>