A path conditioning method with trap avoidance

Abstract

This work presents a sliding-mode method for robotic path conditioning. The proposal includes a trap avoidance algorithm in order to escape from trap situations, which are analogous to local minima in potential field-based approaches. The sliding-mode algorithm activates when the desired path is about to violate the robot workspace constraints, modifying it as much as necessary in order to fulfill all the constraints and reaching their limit surface at low speed. The proposed path conditioning algorithm can be used on-line, since it does not require a priori knowledge of the desired path, and improves the conventional conservative potential field-based approach in the sense that it fully exploits the robot workspace. The proposed approach can be easily added as an auxiliary supervisory loop to conventional robotic planning algorithms and its implementation is very easy in a few program lines of a microprocessor. The proposed path conditioning is compared through simulation with the conventional potential field-based approach in order to show the benefits of the method. Moreover, the effectiveness of the proposed trap avoidance algorithm is evaluated by simulation for various trap situations.