A Global-Local Approach for Trajectory Generation on Rough Terrain

In this paper,we consider trajectory generation problem as an optimal control problem while minimizing a cost function and satisfying any boundary, terrain, path and system dynamics constraints. The optimal state and control profiles are approximated by a weighted average of independent local approximations. Specially derived weight functions are used to construct a globally continuous function. Such a method provides us a way of using local information to construct near-optimal trajectories, which is useful in the context of numerically solving nonlinear trajectory generation problems.