Research On SLAM Algorithm Of Mobile Robot Based On Multi-sensor Fusion

The mobile robot SLAM(simultaneous positioning and mapping)technology is one of the key technologies for mobile robots to achieve autonomous navigation.With the continuous improvement of artificial intelligence and hardware computing power,mobile robot SLAM technology is gradually diversified.In addition to the traditional two-dimensional laser radar SLAM,SLAM technology based on visual and visual inertial fusion has become a research hotspot in recent years.However,the environment faced by the mobile robot SLAM problem is relatively complicated and the mobile robot itself is equipped with various sensors such as a camera,a laser radar,an IMU,etc.,which can be used to implement the motion estimation of the mobile robot itself.Combining the advantages and disadvantages of visual SLAM and two-dimensional laser SLAM and the advantages and disadvantages of various sensors,this paper studies and implements the SLAM algorithm of mobile robot based on multi-sensor fusion.The main work of this paper is as follows:1.Research on loose coupling two-dimensional grid map and 3D point cloud map matching optimization method.Aiming at the difference of the algorithm framework of visual SLAM and laser SLAM and the way of map storage,this paper designs a real-time matching laser SLAM track pose and visual SLAM key frame pose and corrects the key frame pose deviation by linear interpolation.Finally,the method of off-line optimization of 3D point cloud map,which reduces the coupling of visual SLAM and laser SLAM,realizes the alignment of 2D grid map and 3D point cloud map,in order to realize the robot in the known map.The location provides a wealth of map information.2.Resear-ch on pose estimation algorithm based on error state Kalman filter.By dividing the advantages and disadvantages of IMU,wheel odometer,camera and lidar sensor,this paper constructs the error state Kalman filter based on IMU error kinematics model and combines wheel odometer and visual laser SLAM pose for mobile robot.Provide high frequency,high stability real-time pose estimation.3.Research on tightly coupled SLAM algorithm based on vision and lidar.Based on the error state Kalman filter pose estimation algorithm,the estimator pose is used as the initial pose between the visual frames and the initial pose of the lidar scan-match to improve the accuracy and speed of matching.The global tight coupling optimization is realized by constructing the pose constraint of the key frame and the local raster map,which improves the consistency of the joint map.