Research On UAV Mapping And Path Planning Based On Multi-Sensor Fusion

In recent years,with the intensive research and development of UAV technology,the application of UAVs to solve problems in many environmental areas has been favored.However,most UAVs rely on global satellite navigation systems for safe flight and operations.With the expansion of the application of environmental areas,it is gradually being used in environmental areas without global satellite positioning signals and autonomous control,which cannot be accomplished for traditional UAVs.The UAV with multiple sensor fusion can perform its own position estimation,environment map building and safe path planning,which is one of the key technologies to solve the safe flight and operation without global satellite signals.Therefore,relying on the sub project of the national key R & D plan "research and development of intelligent geophysical exploration UAV system".This paper studies the mapping and path planning of UAV equipped with multiple sensors.The main research content of the paper is as follows:(1)In order to effectively fuse the measurement data of binocular camera,Li DAR and IMU,accurate binocular camera intrinsics,IMU error parameters,binocular camera and IMU extrinsics,and binocular camera and Li DAR extrinsics are needed.In this paper,firstly,the theoretical basis of the pose transformation is summarized,based on which the mathematical models of binocular camera,IMU error and 2D Li DAR are established.Secondly,a fast IMU deterministic error calibration method based on Dogleg in the trust domain and an IMU random error calibration based on Allan’s variance principle are proposed.Finally,the extrinsics of the binocular camera and IMU and the binocular camera and Li DAR are calibrated by combining the pose transformation theory.(2)For the closed environment without global satellite signal,this paper studies the velocity fusion of optical flow-IMU based on the filtering method and the position estimation scheme of optical flow-IMU-2D Li DAR.And this paper proposes a improved ORB feature extraction algorithm for the problem that the edge feature points cannot be extracted in ORB feature extraction.For the problems that IMU is easy to drift and optical flow images are easily affected by light,the complementary-Kalman filter fusion method is proposed.The speed calculation of the fused UAV has less deviation from the real speed and the robustness of the position estimation has also been improved.Based on this,a 3D dense map construction method based on RGB depth camera is proposed to construct a 3D dense map for the simulation environment,which proves the effectiveness of the proposed method and provides a map basis for the subsequent UAV path planning in unknown environments.(3)An improved artificial hummingbird algorithm is proposed for the path planning of UAVs.Firstly,the range constraint,flight altitude constraint and heading angle constraint models,as well as the path planning model are established according to the structure and environmental characteristics of the UAV.Secondly,the artificial hummingbird algorithm is improved by using Chebyshev chaos mapping to get a more balanced distribution of initial food sources and the solutions obtained are more convergent to the optimum,which accelerates the convergence speed of the algorithm;Levy flight is introduced to expand the search capability of the guided foraging phase and solve the problem of the algorithm falling into local optimum.Again,the path calculated by the algorithm is smoothed and optimized with reference to the generation principle of B-sample curve,and the UAV 3D path planning algorithm for improving artificial hummingbirds is obtained.Finally,the obstacles are replaced by mountain peaks in a simulated closed environment,and the simulation and optimal path solution are performed,and the results show the effectiveness of the improved algorithm.(4)Build a platform for functional verification.The F450 UAV platform equipped with the required sensors for the experiment is designed and calibrated based on the established theoretical methods for the depth camera intrinsics,IMU error,combined depth camera and Li DAR extrinsics,and combined depth camera and IMU extrinsics.Finally,the indoor implementation without GPS signal is carried out by this built UAV platform and the proposed mapping and path planning scheme.