Hopfield net generation and encoding of trajectories in constrained environment

The Hopfield network, transient dynamic is used for resolving a path planning problem. The patterns are provided by the network dynamic in the phase space rather than by the attracting stable behavior in this same space. Two essential characteristics of the Hopfield net are presented: a rich and robust dynamics and a distributed memory capacity. The rich dynamics allow the discovery of trajectories in constrained environments, i.e., robust trajectories, satisfying various constraints, such as join two points, pass along others, turn obstacles, be short. More time-explicit constraints, such as avoid strong accelerations, could be easily added. The distributed memory allows the encoding of several of these trajectories whose respective release depends entirely on the departure point