ONLINE COLLISION PREDICTION FOR MOBILE ROBOT COLLISION AVOIDANCE UNDER UNCERTAINTY

An efficient method for predicting collisions between two wheeled mobile vehicles is presented. This problem has been incorporated [1,2] in the framework of an on-line collision avoidance strategy for mobile robot environments. Both the robot and the moving objects are modeled as convex polyhedra. The use of L1 or L(infinity) norms instead of the Euclidean norm, makes the problem a linear programming problem. Uncertainty is induced by sensing and the unknown dynamics of the moving obstacles. Thus, the following stochastic problems are formulated: First, filtering of the minimum distance between the robot and the moving object. Second, prediction of the minimum distance in future instants. Finally, prediction of possible collisions between the mobile vehicles and estimation of the collision time. The computational and numerical aspects of the proposed approach are discussed.