Simultaneous Localization And Map Building For Three Wheeled Omnidirectional Mobile Robot

In the field of robot research,Simultaneous Localization and Mapping(SLAM)is the key to autonomous navigation of robots.Among them,the method of creating maps is of great significance.This paper analyzes the core issues of three-wheeled omnidirectional mobile robots positioning and mapping,and establishes a solution model for the entire system.According to the needs of practical engineering,the multi-sensor data is fused,and the depth camera is used to assist the positioning of the laser sensor,and the positioning correction is performed by using the linear features of the environment,so that a more accurate environment map can be obtained.While using the open source dataset to verify the improved algorithm,a three-wheeled omnidirectional mobile robot platform was built,which verifies the effectiveness and practicability of the algorithm from the practical point of view.In view of the inaccuracy of the robot attitude angle caused by the low accuracy of the code wheel mileage data in actual applications,this paper uses the extended Kalman filter method to fuse the code wheel data and the gyro data,and the accuracy of the robot's predication pose is doubled.To a large extent,the accumulated error of the front end is reduced.The traditional SLAM based particle filter only uses the odometer to predict the pose of the robot,which may cause the prediction pose error to be too large,and the algorithm's accuracy and complexity depend on the number of particles.This paper adopts the discarding method based on the odometry prediction pose.The particle's high-speed scanning matching method uses laser observation data to estimate the pose of the mobile robot by least squares matching,and then reduces the single front-end error to a range of 1 cm and 1 degree.In order to solve the problem that traditional SLAM method based on particle filter can not perform loop detection,this paper adopts a visual depth sensor to perform the loop detection method to assist the loop detection and ensure that the front error of the mobile robot is too large.G2 O performs global closed-loop pose correction,eliminating the cumulative error of the mobile robot.In order to solve the problem of the distortion of the wall surface of the map created by the traditional SLAM algorithm in a corridor with a long straight line environment,the method of extracting the straight line features in the environment is adopted in this paper.The alignment of the robot is corrected by the straight line matching,which in turn makes the map wall more straight and improves.Long-line environment map accuracy.In order to verify the practicability and feasibility of the algorithm described in this paper,an actual robot experimental platform was set up to implement related SLAM functions,and its effectiveness was verified through open source data sets and real-world environments.