Hybrid evolutionary path planning via visibility-based repair

This paper introduces a hybrid evolutionary system for globel path planning within unstructured environments. This hybrid system combines a novel representation for obstacles within an environment, the concept of evolutionary search and a new concept we refer to as visibility-based repair to form a hybrid which quickly transforms infeasible paths into feasible ones. Our hybrid evolutionary system differs from other evolutionary path planners in that (1) more emphasis is placed on repairing infeasible paths to develop feasible paths rather than using simulated evolution exclusively as a means of discovering feasible paths and (2) a continuous map of the environment is used rather than a discretized map. In this paper, we demonstrate the effectiveness of this new hybrid system by using three challenging path planning problems.