Performance Characterization, Development, and Application of Artificial Potential Function Guidance Methods

摘要

The primary objective was to examine artificial potential function (APF) guidance performance when applied to systems with limited control authority in a dynamic environment and develop a hybrid guidance to improve algorithm convergence and computational cost. Performance with respect to both computation time and cost was improved by hybridizing the APF approach with receding horizon planning. Results showed that for the hybrid algorithm, computation time was improved from the optimal control solution while improving the convergence and cost from the APF solution. While the hybrid method greatly improved performance for a saturated system in dynamic environment, this was limited to a fully actuated system. When applied with indirect control, performance was improved, but did not converge. Based on this initial data, the hybrid approach shows promise in regard to implementation within a real-time guidance scheme, however, there is still work to be done before it will be fully effective. The secondary objective was to determine what classes of problems are well-suited to APFs or APFhybrids. The data suggests that APFs and the hybrid algorithm proposed are best applied to fully actuated systems. Additionally, if external dynamics or substantial saturation exist, APF guidance performs better when supplemented with an alternative method.