Design Of SLAM Algorithm For Indoor Mobile Robot

In order to achieve robotic autonomous mobility,the problem of simultaneous localization and mapping(SLAM)must be solved first.The laser radar-based robot SLAM algorithm is relatively mature.Mobile chassis and laser radar are the main sensors of the laser SLAM system,and the reduced accuracy of the sensor will result in poor accuracy of the map output by the SLAM algorithm.In this dissertation,the robot SLAM system constructed by low-precision mobile chassis and single-line laser radar is researched.By analyzing the data error and distortion caused by the sensor,the corresponding solution strategy is designed to improve the accuracy.An optimization algorithm is given for the particle filter resampling process.Firstly,the various parts of the SLAM system are modeled.The probability model of the SLAM system based on Bayesian filter is given.The measurement model of the sensor is established to obtain the odometer and environmental measurement data.The system observation model based on the likelihood field model is established to correct the robot pose prediction results.In order to obtain the obstacle information of the two-dimensional indoor environment,a mathematical model of the grid map is given.The above model lays a mathematical foundation for the sensor data processing strategy and particle filter algorithm optimization.The odometer provided by the mobile chassis will produce cumulative errors during the robot's motion.The real-time calibration strategy for the wheeled odometer was designed to eliminate this error.The PL-ICP algorithm is used to match the laser radar scan data to obtain the reference truth value of the robot pose.The original odometer is calibrated based on the nonlinear least squares method,and the cumulative error of the odometer is eliminated.The lidar's scan data produces motion distortion as it moves quickly.The principle of this distortion is analyzed,and the laser distortion removal algorithm based on linear interpolation is proposed.The emission position of each laser beam is corrected by linearly interpolating the pose of robot provided by the odometer.the removal of the laser radar motion distortion is completed.Particle filtering is used as the SLAM backend optimization algorithm.The principle of particle filter algorithm is analyzed,and the basic steps of particle filter based SLAM algorithm are given.Aiming at the particle degradation and dissipation problem of particle filtering,the multi-threshold resampling algorithm is proposed to optimize particle filtering.The MATLAB simulation proves that the optimization algorithm can reduce the degree of particle degradation and dissipation by controlling the fluctuation range of effective particle number,and thus reduce the tracking error of particle filter.At the same time,the relationship between the threshold number and the tracking error of particle filter is given.In the indoor environment,the real-time calibration strategy of the wheel odometer,the laser radar motion distortion removal algorithm and the multi-threshold resampling optimization effect were verified.The robot operating system(ROS)was used as a software platform and the robot was controlled to move indoors.The map from the SLAM algorithm output was given.The experimental results were analyzed and the corresponding conclusions were given.