AGV System Error Correction Method And SLAM Algorithm Implemention

Wheeled differential AGV is widely called as Automated Guided Vehicle(AGV).AGV usually drives autonomously along the specified guiding path according to the pre-compiled instructions in the different operating environment with wireless communication through industrial computer and ranging sensors.The diversified demands and advantages of realizing multiple functions have promoted the rapid development of AGV related research.As the most widely used robot type,AGV plays an increasingly important role in many fields,such as warehousing and handling,express delivery,independent p arking,power station inspection and so on.In this thesis the kinematics analysis of the built-in differential AGV is carried out,and the odometer error model is established.The bidirectional square experiment is combined with the UMBmark system error checking method.The laser sensor triangulation method is used to measure the two fixed target points.The coordinates of the starting position and the end position of the robot are calculated.Then the wheel diameter and the wheelbase of the differential AGV are calibrated by the distance deviation value.And we can get the diameter and the wheelbase of the correctable system error.The differential AGV trajectory tracking control based on kinematics model is studied.A new control law is proposed for the ideal model.The simulation results show that the control law designed in this thesis can effectively improve the trajectory tracking accuracy.The Cartographer algorithm of Google is studied and applied.The unmanned Kalman filter is used to fuse the data of odometer,IMU and laser sensor to predict the pose of the robot.The data of the laser scan is coordinate transformed to extract the constraint relationship of the pose change of the robot;The nonlinear least squares estimation method is applied to solve the attitude optimization problem.Through the attitude matching algorithm constantly finds the best position in the optimization window.Then using the sparse pose adjustment algorithm to calculate the optimal position;By the branch upper bound method of depth-first search to accelerates matching and improve the optimization efficiency.Using the distributed network architecture features of the ROS system to build a flexible and stable robot communication architecture.According to the interface defined by ROS,using the visualization tool RVIZ to display the 3D model,the surrounding environment,the motion trajectory and other information.The model is used to perform simulation verification under Gazebo.Finally,the wheel diameter and track correction experiments of the differential AGV and the localization and mapping experiments are designed to test the effect of the simultaneous localization and construction of the differential AGV platform.