Research On Laser SLAM Navigation Technology For Rotorcraft In Underground Spac

Laser SLAM is an important technology for autonomous navigation of rotorcraft in underground space.Environment perception and path planning are the key issues of laser SLAM navigation.The traditional SLAM algorithm has obvious problems in memory consumption and calculation.In view of the above problems,this paper introduces other observations,uses graph optimization to build maps,and improves the path planning of the rotorcraft at the same time,and studies the laser SLAM navigation technology of rotorcraft for underground space.The main contents of this article include:1.Environment perception method based on multi-sensor fusion.Aiming at the problem of incomplete construction of single sensor in underground complex space map,the lidar point cloud data and IMU feature data are processed to form a combined laser point cloud.The point cloud cluster set is formed by threshold segmentation,and then the original line segment set is selected by linear fitting method.Secondly,the circular fitting algorithm is used to fit the nonlinear segment set to form a circular obstacle set.Finally,the corridor boundary model is established according to the linear obstacles set.Experiments prove the correctness and feasibility of the multi-sensor fusion environment perception modeling method.2.Point cloud data fitting algorithm based on moving least squares.Aiming at the problem that the fitting effect of complex surface point cloud data is not obvious,the surface fitting function is established,theconstraint conditions are added in the iterative process,and the objective function is solved by Lagrange multiplier method,and then the point cloud data after filtering is obtained.Then the least square method is used to solve the quadratic surface coefficient.and the quadratic surface fitting is obtained.Then the least square method is used to solve the quadratic surface coefficient,and the quadratic surface fitting is extended to polynmial fitting to construct a complete surface.Finally,the fitting function is established,and the weighted discrete norm is considered to obtain the moving least squares fitting function.At the same time,the support domain of the weight function in the moving least squares fitting method is set.The sensitivity analysis of the algorithm shows that the fitting results of the moving least squares method tend to the original surface.3.The path planning algorithm for rotorcraft based on improved D* algorithm.Aiming at the problem that the traditional path planning algorithm has too many turns,the cost in the vertical direction and diagonal direction of the chessboard distance is set to obtain the optimized cost estimation function.Secondly,on the basis of the selected first-level priority nodes,the selection method of the second-level priority nodes is set.Finally,the smoothness function is introduced on the above optimized cost estimation function,and the smoothness function is divided according to the status of the planned path.The simulation results show that the number of path turns is reduced,and the safety performance of the rotorcraft is improved.4.In this paper,the first three tasks are integrated,the experimental platform is built,the underground parking lot is selected as the experimental site,and the laser SLAM navigation scheme is designed.ZD850 six-rotor is selected to carry RS-LIDAR-16 laser,which is designed from hardware and software.The experimental results show that the environmental perception method based on multi-sensor fusion can outline the environment information of the experimental site in detail,and the rotorcraft equipped with sensors can fly safely and smoothly in the underground environment.