Dynamic Decision Making of Mobile Robot under Obstructed Environment

Our path planner is divided into two phases: First, let the robot study its maneuvering environment before giving a task, and then based on the studied map the robot is able to generate a safe and smooth path to its desired goal by using our potential field based collision avoidance algorithm. The environment exploration is done by adopting reduced generalized Voronoi graph which creates a topological map containing only GVG edges unrelated to boundary points. By setting sub-goals at meet points on the map the dead-lock problem caused by existing concave environment can be solved. The local minima problem encountered while the robot is traveling along subgoals is also solved by redefining the repulsive potential field. The system controller (look-ahead control) gives the robot a capability of controlling the distance between the reference point and the center of the robot. As a result, the potential field method performs effectively under our system and allows the mobile robot to follow a smooth trajectory in a flexible manner for attaining the desired goal autonomously whether in a static or dynamic environment