Mobile Robot Path Planning and Obstacle Avoidance in Unknown Environment with Fuzzy Obstacles

In this paper Hopfield neural network is used for path planning and obstacle avoidance in an environment with fuzzy (soft) obstacles. The 2-Dworkspace of the robot is divided into small cells (grids)and each cell is modelled by a neuron in a Hopfield network. The model assumes that an external inputspecifies the target neuron and the obstacles in the neural map. After training the network, the robot can find the shortest path from any arbitrary start positionto target avoiding fuzzy obstacles within its workspace. Proof for stability and uniqueness of the surface's peak are included. Computer simulations are performed to verify analytical results