An efficient obstacle avoidance scheme in mobile robot path planning using polynomial neural networks

Application of Polynomial Neural Networks (PNN) in mobile robot path planning with an obstacle avoidance scheme is proposed. Given an environment and a desired goal location (position and orientation), PNN's are built from some selected starting locations to reach this goal. These PNNs comprise the memory of our model. An efficient associative retrieval technique is then applied to make the robot follow a minimal cost polynomial path. In the movement, when it faces an obstacle, the robot uses a contour finding algorithm to get away from the obstacle. The major advantage of using the PNNs is its interpolating capability with a moderate size of data space. Also no preprocessing of the range data is necessary