Automatic planning of robot pushing operations

Abstract

A method for automatic planning of robot pushing operations is described. The operation is performed by a two-point pusher. It has been shown that such an operation can entirely constrain polygonal workpiece in a plane and can deal with objects in the presence of initial uncertainty. The operation is discussed with regard to bounds on initial error in position and orientation. The success of an attempt to push the object depends on correct pushing direction and starting point of the pusher. The goal is to catch the object between the fingers and to eventually bring it to known position and orientation. The starting point and the pushing direction define an operation space of all possible tasks. A bounded region in this space is constructed, such that any point inside this region guarantees the object will not be pushed away. The planner also accounts for the uncertainty in positioning, suggesting the pushing operation that will succeed even if worst case error occurs.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>