| In recent years,with the development of intelligent robot technology,mobile robots play an increasingly important role in high-risk rescue,military and harsh environments.In some complex environments such as multiple obstacles and unstructured terrain,it is difficult to decide the optimal motion plan only by relying on the intelligence of mobile robots.Therefore,in some complex operating situations,it becomes particularly important to add human control to the robot as an auxiliary means,and the premise of human control is to obtain the on-site environment around the robot.Most of the current methods of manipulating remote mobile robots are based on the live images transmitted by the robots.However,the two-dimensional images lack distance information,which is not conducive to human manipulation,and the image information is difficult to use for precise motion decision-making of the robot.Compared with the two-dimensional image information,the three-dimensional virtual environment reconstructed by the point cloud scanned by the lidar can provide accurate distance information,which is more in line with the human visual sense,and has a flexible viewing angle,which greatly enhances the operator's sense of presence.Therefore,combined with the school-enterprise cooperation project "Intelligent Smoke Screen Release Equipment Research and Development",this paper builds a vehiclemounted mobile platform and a wireless control terminal,and studies the terrain reconstruction technology based on the wireless control terminal.The main contents of this paper are as follows:(1)Based on the C/S architecture,the construction of the system hardware platform is completed,including the vehicle server and the wireless control terminal.Establish a terrain reconstruction coordinate system,analyze lidar and GPS coordinate systems,and complete data preprocessing.(2)The reconstruction of the terrain is completed on the vehicle server,and a terrain data storage method of block map is proposed,which realizes the dynamic loading and updating of terrain blocks.A method of fusion of IMU and GPS data is proposed for position estimation,and based on Kalman filtering,the historical elevation values in the grid are fused to improve the accuracy of terrain reconstruction.(3)For the transmission channel with limited transmission capacity,a solution for the compression and transmission of terrain data is proposed.In order to reduce redundant data,this paper establishes a data acquisition model in the process of vehicle body motion,and solves and transmits the terrain update area.Compression of terrain data adopts 2D-DCT transform and pruning operation of coefficient moments.According to the characteristics of coefficient matrix,sequence data is obtained by ZigZag scanning method.The traditional run-length encoding is optimized,and the sequence data is encoded with the optimized run-length encoding method.(4)For the visualization of the wireless control terminal,firstly,a color difference algorithm of points is proposed to color the terrain grid;then a terrain rendering algorithm based on adaptive downsampling is proposed according to the degree of terrain scaling,so as to save the operation and processing resources of the control terminal;Finally,the grid jumping problem in the algorithm is solved by calculating the new visible range diagonal points.(5)Under the Windows 10 system,the C++ programming language is used to realize the functions of sensor data acquisition,data processing,data compression,data transmission,data analysis,etc.,to complete the reconstruction of the terrain;use the Qt GUI framework to complete the vehicle server and control terminal.The software human-computer interaction interface,using Open GL in the control terminal to complete the visualization of the reconstructed terrain on the vehicle server.(6)Complete terrain reconstruction accuracy experiments,terrain data compression experiments,terrain rendering experiments,and visualization delay tests to verify the effectiveness of this method. |
PDF Full Text Request