A neural network architecture for path planning

The potential field approach to path planning is computationally intensive but highly parallel. The authors propose a general neural network architecture called the wave expansion neural network (WENN), specifically designed to implement such potential field operations useful for path planning. The authors describe WENN-based neural network systems which implement certain types of artificial potential fields for path planning. The neural network takes the workspace which needs to be navigated as input and generates artificial potential fields which are then subsequently used to find a path. Neural network implementation results in greatly reduced computations.<<ETX>>