CELLULAR AUTOMATA AND OPTIMAL PATH PLANNING
Cellular automata are deterministic dynamical systems in which time, space, and state values are discrete. Although they consist of uniform elements, which interact only locally, cellular automata are capable of showing complex behavior. This property is exploited for solving path planning problems in workspaces with obstacles. A new automaton rule is presented which calculates simultaneously all shortest paths between a starting position and a target cell. Based on wave propagation, the algorithm ensures that the dynamics settles down in an equilibrium state which represents an optimal solution. Rule extensions include calculations with multiple starts and targets. The method allows applications on lattices and regular, weighted graphs of any finite dimension. In comparison with algorithms from graph theory or neural network theory, the cellular automaton approach has several advantages: Convergence towards optimal configurations is guaranteed, and the computing costs depend only linearly on the lattice size. Moreover, no floating-point calculations are involved.