Research On Robust Multi-sensor Fusion SLAM System

In recent years,intelligent mobile robot technology has developed rapidly.And simultaneous localization and mapping(SLAM)is the key technology of the mobile robot,which plays an important role in ensuring the autonomous exploration of robots in unknown environments.Visual-inertial SLAM and laser-based SLAM are the main research fields of SLAM.Although the inertial measurement unit(IMU)significantly improves the performance of camera motion estimation,tracking failures occur when in illuminate variation or texture-less region for a long time.In addition,lidar has the problem of motion distortion caused by continuous measurement and self-motion.To tackle these problems,this paper presents a visual-inertial-laser SLAM system,which includes robust and precise odometry,loop closure,proximity detection,and map reuse.The main work of this paper is as follows:1.In order to improve the precision of motion estimation,a visual-inertial-laser odometry is proposed.Starting with a tightly-coupled visual-inertial odometry,followed by a scan matching to refine the estimation,and register the point cloud to the global map.2.For the sensor degradation in the illuminate variation,texture-less and structure-less environment,implement a system structure which can be flexibly and automatically adjusted.3.In order to solve the problem of error accumulation in state estimation,global constraints are added by visual-based loop closure and laser-based proximity detection,and then the accumulated errors are removed by global pose graph optimization to maintain global consistency.4.Map reuse is implemented.Combining visual and laser measurements to locate on the loaded map.