Neural dynamics based multiple target path planning for a mobile robot

Abstract

A mobile robot must be able to plan efficient routes to locations that it is required to visit. In several applications, several target locations are required to be visited. This is more complicated than the path planning problem where only a single destination exists. In multiple target path planning, the problem is similar to the traveling salesman problem. Existing solutions solve the problem using offline approaches, limiting their usefulness in dynamic environments. This paper presents an online solution for multiple target path planning in static, prioritized and dynamic environments. The basis for the solution is a shunting model neural network. Simulation results show that while the solution is not optimal, the algorithm can provide an acceptable solution in even dynamic environments.