Artificial potential field-based motion planning/navigation, dynamic constrained optimization and simple genetic hill climbing

In this paper we show a relationship between artificial potential field (AP F) based motion planning/navigation, and constrained optimization. We then present a simple genetic hill climbing algorithm (SGHC), which is used to n avigate a point robot through an environment using the APF approach. We com pare SGHC with steepest descent hill climbing (SDHC). In SDHC, candidate mo ves are evaluated within a 360-degree radius and the best candidate is sele cted by the robot. One would think that SGHC would be at a disadvantage; ho wever, the performance of SGHC is comparable with SDHC. SGHC has an advanta ge in that it is capable of evolving (learning) the appropriate step size a s well as the appropriate angle of movement.