SLAM Loopback Detection and Down Sampling Optimization in Unattended Patrol Environment

Aiming at the optimization problem of SLAM mapping algorithm in substation environment that requires no patrolling, considering the design requirements of patrolling environment map for highlighting obvious features and mapping accuracy, this paper proposes a SLAM algorithm that uses loopback detection based on global descriptors and optimizes down sampling. Considering the accuracy of loop detection in a large environment, this paper designs a loop detection method using an efficient bucket function, which converts single frame LiDAR point cloud data into a two-dimensional matrix with a numerical value of the maximum height of the laser point in the bucket, thus achieving excellent loop detection; An adaptive down sampling filter based on voxel down sampling and curvature down sampling is designed. The optimized algorithm has been optimized by 20% compared to the initial scheme in terms of relative pose error and absolute trajectory error, proving the effectiveness of the proposed method. Compared with the mapping results of the traditional LEGO-LOAM algorithm, the SLAM algorithm proposed in this paper uses the loopback detection based on the global descriptor and optimized down sampling to achieve better mapping accuracy and reliability.