Relative angle based obstacle-avoidance

The Local Minima (LM) is the main problem of well-known artificial potential field (APF) method in the scope of obstacle avoidance. In this paper, we prove the existence of the local minima, which causes the unavoidable collision. Furthermore the concept of relative angle is added in repulsion calculation which makes repulsion change with angle among destination, the robot and obstacles. In addition, a hybrid path planning method, called RRT-APF, is proposed through first calculating circumvention point by the modified APF with the relative angle and then generating obstacle-free path via Rapid-exploring Random Tree (RRT). The method not only preserves the efficient search of RRT, but also decreases the randomness of RRT. Simulation results demonstrate that ADAPF is valid on collinear local minima problem. Moreover, in terms of planning time consumption and minimum distance, the proposed RRT-APF method is superior to both RRT and APF.