| Summary: | 博士 === 國立成功大學 === 航空太空工程學系碩博士班 === 95 === This study presents a method, using a fast graph-search algorithm, of finding a feasible flight path traversing multiple targets (waypoints) above a three-dimensional (3-D) real terrain map for air vehicle. This problem involves the flight safety clearance, the real flight constraints, the performance limits of the air vehicle, and the order of target visitation.
We deal with this problem by solving two-point path planning problem at first. The flight path must satisfy the many constraints required to make the flight safe and efficient. The considerations of sufficient safe altitude and horizontal distance are indispensable in flight safety concerns. In practice, the preferred directions of take-off/landing, the limits of climbing/turning rates, and the fuel consumption ratio should also be taken into account. We construct a virtual terrain as a search space above the real terrain, to take into account real flight conditions and the limitations of the vehicle’s performance. The safety concerns of flight path and the phase of take-off and landing are also included. The idea of a virtual terrain could also eliminate a significant amount of search space, from 3-Dimensions to 2-Dimensions, which takes much less computational time, but which may have a shortcoming in rugged terrain where most path points are higher than the cruise altitude. Additionally, the digital indiscrimination problem will show up when discretization is applied. Hence we propose further post processes, which take less than a second with no extra computational load, to overcome these problems. A dimensionless fuel consumption ratio between climbing and level-turn is proposed to deal with the case of level flight between valleys. If climbing requires greater fuel consumption than taking a level turn, the algorithm chooses the level altitude flight path, hence improving the vertical smoothness of the flight. Using all these methods, including multi-resolution terrain and a fast searching method using a heuristic, we have successfully reduced the computational time to an acceptable level, and the simulation results show that our two-point algorithm is feasible.
We next extend our two-point algorithm to solve the multi-point path planning problem. Particularly, the turning rate limit should be considered not only in each path segment itself but also at each junction between any two sequential path segments. This complex problem can be solved efficiently by utilizing both the heuristic and hierarchical schemes in a pre-constructed virtual terrain as well. Furthermore, to determine the optimal order of target visitation, the turning rate constraint at the junctions is relaxed while the distance matrix is computed. Subsequently, the order permutation associated with the minimum total distance is calculated. The quality of the planned path is improved by several fast post-process methods. The experimental simulation results showed that our algorithm is practical with an acceptable computational time.
|
|---|
