Abstract
This paper presents an algorithm for generating Smooth Collision Avoidance Trajectories (SCAT). SCAT generation is a method that allows a mobile robot that is moving along a pre-planned path to alter a section of its path, so that it may smoothly exit the original path, avoid a predicted collision, and return to the original path smoothly and on schedule. The SCAT generation algorithm is an improvement over off-line methods, as it requires minimal a priori information, and is more robust than pre-planned methods by its very nature. The SCAT algorithm is also an improvement over on-line schemes that only alter velocity along a pre-planned path, as it is able to avoid collisions in cases that those methods cannot. Details of the SCAT generation algorithm are developed herein, followed by examples of the algorithm in action. Simulation results show that the SCAT algorithm is very dependable, given that it can be provided with reasonably accurate in-formation about the location of dynamic obstacles in its vicinity.
Bayesian recognition of a moving object information-measuring system state: a priori information weight correction