Research On UAV Trajectory Planning Based On 3DVFH+ And DWA Algorithm

Trajectory planning has been a very active research field for many years.These algorithms have been constantly developing and becoming more perfect,but they still cannot meet users’ requirements for speed and trajectory quality instead of decreasing in today’s increasingly complex and diversified flight environment.In the 3d unknown environment with dynamic high density obstacles,the calculation amount of the current algorithm will increase exponentially as it solves 3d problems in 2d or 2.5-dimensional plane,thus affecting the overall running speed of the algorithm.Moreover,the dynamic constraints of UAV and the degree of freedom in the vertical direction of 3D space are ignored.There may be paths in three-dimensional space but no effective results can be obtained in twodimensional space,resulting in the overall total path being too long and increasing the flight cost of UAV.Therefore,the problem of how to quickly obtain the optimal or nearoptimal trajectory with small computation,no collision and high smoothness in the highdensity dynamic THREE-DIMENSIONAL environment has not been finally solved,and there is still huge potential for improvement.In this paper,a real-time trajectory planning method for controlling the movement of quadrotor UAV is proposed.The algorithm can plan a low-cost feasible THREEDIMENSIONAL trajectory for UAV in a dynamic environment with high density obstacles,and the trajectory can adapt to path constraints,uav dynamic constraints or performance constraints.The algorithm is extended on the basis of 3DVFH+ algorithm,adding the dynamic window（DWA）algorithm which is commonly used in two-dimensional space.As a classical trajectory planning algorithm,DWA uses evaluation function to select the best motion command,which is fast and very effective for dynamic obstacle avoidance.In order to improve the evaluation ability of the evaluation function and adapt to the uav dynamic characteristics,a new evaluation function with adaptive weight is designed by extending the two-dimensional to three-dimensional evaluation function.Combined with the 3DVFH+ algorithm,the point cloud generated by the depth camera is used to extract information from the surrounding environment of uav,and then the 3D octant model is defined to compress the point cloud,avoiding the sudden increase in the amount of calculation caused by the establishment of global map.The algorithm has good real-time performance and is close to the global optimal.The comparison results show that the hybrid algorithm is superior to the traditional 3DVFH+ algorithm and dynamic window algorithm in terms of computational efficiency,smoothness and global path length.In order to verify the results obtained,in addition to the comparison and verification of true in MATLAB,Gazebo was used to simulate the real environment test in the virtual generation environment.Simulation and experimental results show that this algorithm is efficient and fast to achieve planning uav in dynamic uncertainty environment,quickly generate a safe,smooth,small amount of calculation and uav dynamics trajectory,successfully achieve the desired target position.