Research On Relocation Method Of Mobile Robot And Development Of Inspection System

Nowadays,the storage management of hazardous chemicals in colleges and universities is characterized by a small number,a large variety,multi-campus districts,and relatively dispersed storage of hazardous chemicals.While the safety patrol and security inspection for the laboratories and storage warehouses of hazardous chemicals in colleges and universities rely on manual work and the installation of cameras,which is not only inefficient but also has blind areas for patrol inspection.Every year,the lab safety accidents and storage accidents caused by this will cause huge losses and risks to colleges and universities.With the continuous development of society and technology,the patrol inspection based on mobile robot can meet the needs of patrol inspection work well in different working environments.Therefore,a robot security patrol system is designed and implemented,which can meet the requirements of all-weather security patrol and emergency response in university labs,hazardous chemicals warehouses,etc.At the same time,focusing on the‘Robot Kidnapping’ problem,a relocation method combining vision and laser is proposed to solve the relocation problem of the mobile robot in the case of unknown initial pose or abnormal location information during positioning.Firstly,the mathematical model of the robot system is established,including the robot coordinate system model,the motion and dead reckoning model and the observation model based on lidar.Then,the Gmapping algorithm based on filtering SLAM and the Cartographer algorithm based on graph optimization SLAM are briefly analyzed and experimentally verified.The concept of relocation is introduced,and the relocation solutions based on the above two different algorithms are implemented and compared.It is found that the two methods have some shortcomings in solving the relocation problem.Secondly,aiming at the shortcomings of indoor relocation based on lidar,a relocation method combining vision and laser is proposed.The establishment and parameter calibration of the camera model and lens distortion model are completed.The process of feature point extraction and inter-frame matching in the image are described in detail.In the SLAM based on vision and laser,EPnP is used to realize the estimation of camera motion pose between frames,synchronously collect and record the robot pose and camera pose of the mobile robot in the SLAM process.When relocation is started,the robot uses the camera to obtain image frames of the current environment and performs global environment feature matching to obtain the global p ose estimation of the camera in the relocation mode.The optimal robot pose is retrieved from the mapping table of the camera pose and the robot pose recorded simultaneously,and the pose is used as the initial pose of the robot under the AMCL algorithm,and particles are scattered around it.The movement information and observation information of the robot are used to make the particles finally converge to the real pose of the robot.The entire relocation process of the mobile robot has completed.Finally,a robot security inspection system is designed and implemented,a new mobile robot platform is built,and the proposed relocation method is integrated.The following of the track free inspection path in the actual inspection needs is introduced in detail and verified by experiments.The environmental anomaly detection based on multi-sensor is completed,mainly including temperature detection of thermal imaging object,detection of thermal imaging intruder and online measurement of environmental parameters.In order to meet the needs of remote users for patrol inspection,a human-computer interaction system based on Web user interface is developed.The test results verify the effectiveness and practicability of the inspection system.