Research On Path Planning For UAV Based On Improved RRT Algorithm

In recent years, UAVs’ path planning has been developed rapidly at home and abroad, gradually applied to military and civil aspects, and has achieved great benefits in these aspects. However, with the environment of task becoming more and more complex and uncertain factors increasing, UAVs’ requirements for route planning will be higher. Therefore, the UAVs’path planning technology has become one of the hotspots for scholars at home and abroad. In order to meet the requirements of path planning for the performance of the algorithm, this paper proposes the crossover particle swarm algorithm and adaptive RRT algorithm based on dynamic step for the path planning.In view of the long planning time and easily falling into local optimum of the traditional rapidly exploring random tree (RRT) algorithm, this paper makes some improvements. Firstly, it introduces the dynamic step into path planning. This improvement can dynamically modify the exploratory step of tree according to the ratio between the angle which is formed by the new exploratory node and exploratory direction and the maximum turning angle of UAV if the new exploratory node is failing node. It will modify the current exploratory step into the initial exploratory step when the next exploratory node is not failing node in order to accelerate it to jump out of local minimum area and improve the efficiency of path planning. Secondly, it introduces the adaptive weighting strategy. The node whose next exploratory node is a failing node has some certain inhibition to its periphery, and the size of inhibitory effect is indicated by the inhibitory factor. Thus, we can calculate the weight of each node according to the inhibitory factor and select the node which has the maximum weight as the growth node of tree. Therefore, it can avoid large amount of useless exploration, and improve the efficiency of path planning.In view of the slow convergent rate and easily falling into local optimum of the particle swarm (PSO) algorithm, this paper makes some improvements. Firstly, the introduction of adaptive inertial weight is to control the convergent performance of algorithm to balance the searching ability between local optimum and global optimum. Secondly, it can make a cross operation between the optimal extremum of each particle and the global optimal extremum of particle swarm to accelerate it move toward the global optimal direction. Thirdly, the local mutational operation of particle swarm is conducted to guarantee the global optimality of feasible path.Finally, this paper respectively use the crossover particle swarm (CPSO) algorithm and the adaptive RRT based on dynamic step(DRRT) to carry out systematic analysis and study of UAVs path planning in the complex environment, and use the simulation platform that combined VC6.0/MFC with Matlab to prove the effectiveness and feasibility of improved algorithm. In the case of mountainous terrain environment, after3000iterations, the length of CPSO planned path is244.673km, the planning time is25673ms, and CPSO has significant improvement on the length of planned path and planning time compared with the traditional PSO. As well, the length of DRRT planned path is241.385km, the planning time is16529ms, and DRRT also has significant improvement on the length of planned path and planning time compared with the traditional RRT. Besides, the planning time reduced9144ms, and the length of planned path reduced3.288km by using DRRT, compared with CPSO.