Real-time Autonomous Guidance for Self-piloted Vehicles

摘要

This paper presents an efficient strategy for solving mission planning and trajectory optimisation problems in real-time. It is generally applicable to autonomous vehicle or robot guidance problems, and is discussed and demonstrated here for a self-piloted aircraft. The objective is to determine an optimal flight path given a set of mission goals (way-points), terminal conditions, geometric obstacles, and vehicle performance limitations. Mission Planning is accomplished by a heuristic optimisation technique, which identifies the optimal way-point sequence while exploring only a fraction of the solution space. Obstacles are avoided by a recursive procedure that optimises path segments locally. Trajectory Optimisation is performed by a gradient projection technique that enforces terminal conditions while reducing inequality constraints. Convergence to the optimal solution is guaranteed by certain restrictions on the penalty function and a path parameterisation using orthogonal polynomials. The efficiency of this approach is demonstrated through simulated aircraft missions.