Computational Approach for Near-Optimal Path Planning and Guidance for Systems with Nonholonomic Contraints

摘要

A computational approach is developed for optimal path planning for constrained nonlinear dynamical systems. In the approach developed here, the continuous-time optimal control problem is transcribed to a finite-dimensional nonlinear programming problem. In this research, we develop novel methods for discretization based on Legendre-Gauss and Legendre-Gauss-Radau quadrature points. Using this approach, the finite-dimensional approximation is kept low- dimensional, potentially enabling near real time or real time solutions. The approach is demonstrated on sample problems and is found to be a highly accurate and computationally efficient way to discretize constrained nonlinear optimal control problems.