Research On Path Planning Algorithms For Autonomous Prospecting Vehicle

Prospecting is an efficient way of accessing mineral resources, but the rough terrains often make it difficult to drive a prospecting vehicle.There is an attempt to solve this problem by introducing autonomous ground vehicles. Path planning is the core technique for autonomous vehicles.An autonomous vehicle cannot drive by itself without path planning. Thus the research on path planning has both theoretical and practical significance.Graph searching method is used to solve path planning problem. The first step is to build a graph,which means generate a planning map with the terrain model; The second step is to search in the graph, which means search for a feasible path in the planning map.For 2D environment, obstacle sensitive method is used to generate the planning map, Discrete-State Convergence-Accelerated A Star algorithm(DCAA*) and Hybrid-State Convergence-Accelerated A Star algorithm(HCAA*) are proposed to search for the path. In the DCAA* part, three different ways to realize obstacle avoidance are disscussed. To make the path executable, the turning angle is limited and a transitional curve is added to every joint. In the HCAA* part, child nodes generation and the architecture of cost is elaborated. The method to calculate the main heuristic cost and sub heuristic cost is introduced,and the effect of heuristics weight is explained.For 3D environment, trafficability grading method is proposed to generate the planning map, and HCAA* algorithm is adjusted to search for the path. In the grading method, four factors that lead to passing failure are considered: step, slope, serial step, and low adhesion. Accordingly, four parameters: step δ, slope ψ, roughness ω, and geologic index τ is used to evaluate the terrain trafficability. To build different scenes for the simulation experiments rapidly, a module design method is used. We can then process the terrain model we build and generate a planning map, using the grading method. The adjustment to HCAA* lies in the architecture of cost and its evaluation.A series of simulation experiments are conducted. The simulation results confirm that the algorithms proposed in this thesis can accomplish the path planning task in rough terrain and have a good performance.