Evolutionary action maps for navigating a robot in an unknown, multidimensional, stationary environment. II. Implementation and results

Abstract

Here, the BGM that is suggested in the first part of this paper used to construct a complete path planner for an agent of arbitrary shape that is operating in a totally unknown, multidimensional, stationary environment. The planner does not require any a priori knowledge about the environment to guarantee that a path to the target will be found. Any a priori information about the environment, regardless of its degree of fragmentation or sparsity, can be integrated into the planning process to accelerate convergence and enhance the quality of the path. Details for constructing the planner along with experiments to demonstrate its capabilities are supplied.