Research On Map Construction Method Of Nuclear Emergency Robot Based On LiDAR

When a nuclear accident occurs,relevant personnel cannot immediately enter the accident site to understand the accident situation and make judgments.Therefore,the nuclear emergency robot entering the accident scene is of great significance to the environmental mapping of the accident scene.There are a large number of large-scale environments with long corridors in nuclear power plants,and after the accident,in such corridor environments,due to the absence of GPS,poor ground conditions,and lack of feature points,the odometer is degraded,etc.The problem of poor positioning and mapping accuracy in the large-scale environment of the corridor.This paper proposes to solve this problem by using the IMU data at the sensor data input end and the delay decision algorithm in the back-end loopback optimization part.The main research contents are as follows:(1)Research on the model of nuclear emergency robot mapping system.Firstly,the ranging model of the lidar sensor and the attitude calculation and attitude estimation model of the inertial measurement unit,which the nuclear emergency robot relies on in the perception process,are established.Secondly,the detailed derivation of the chassis kinematics model required for the nuclear emergency robot to move.Finally,the commonly used topological map,geometric feature map and grid map are compared and analyzed.The grid map model which can express the unstructured complex environment is selected and its construction process is theoretically deduced.(2)Nuclear emergency robot mapping system design.First,the overall system of the nuclear emergency robot is designed.Then according to the task requirements of the nuclear emergency robot,the hardware system modules of the nuclear emergency robot are designed and selected.Finally,the software platform ROS is outlined and the software system modules are designed.(3)Research on Gmapping SLAM algorithm based on particle filter principle and Cartographer SLAM algorithm based on graph optimization principle.First,the principles of Bayesian filtering,particle filtering,and Gmapping SLAM algorithm are deduced and explained;then,the graph optimization theory and Cartographer SLAM algorithm are studied and analyzed,IMU data is added to the sensor input of Cartographer SLAM,and delay is used in back-end optimization.The decision-making algorithm is optimized by loopback,which is used as the optimized Cartograpehr SLAM algorithm to improve the mapping accuracy and reduce the CPU usage.(4)Experimental analysis of SLAM algorithm based on nuclear emergency robot.First determine the experimental environment and evaluation criteria.Secondly,determine the experimental purpose and experimental plan,and verify the mapping effect of the particle filter algorithm and the graph optimization algorithm under the actual working conditions similar to the nuclear emergency robot by using the Gmapping algorithm and Cartographer,and quantitatively analyze the mapping effect using the evaluation criteria.Finally,a quantitative analysis of the mapping effect of the Cartographer algorithm before and after optimization is carried out.