Obstacle avoidance procedure and lee algorithm based path replanner for autonomous mobile platforms

Abstract

The paper proposes solution for two important issues connected to navigation of independent mobile platforms in an unknown environment. First issue relates to obstacle map, estimated based on stereovision images. It provides a basis for further platform path-planning. The main problem that has to be solved in obstacle map derivation is elimination of artifacts resulting from depth estimation. Thus a two-step artifact filtering procedure is proposed, which exploits both within-frame spatial correlations as well as temporal, between-frame correlations to do this task. Second procedure, based on well-known Lees algorithm is designed for obtaining vehicle collisionless path. Such routes need to be updated on-the-fly to take into account moving obstacles or newly detected objects. The main idea of the proposed approach is to identify regions where environment has changed and to execute a procedure of selective path updates. As a result, an optimal path can be derived at a computational expense comparable to the heuristic Lifelong A* search. Experiment results demonstrate efficiency of the two discussed approaches for platform operation control in real environments, where both static and moving obstacles are present.