Grid-based Submap Joining: An Efficient Algorithm for Simultaneously Optimizing Global Occupancy Map and Local Submap Frames

TL;DR

This work tackles grid-based SLAM in large-scale 2D environments by formulating submap joining as a non-linear least squares problem that jointly optimizes a global occupancy map and local submap frames. It proves a key independent property of Gauss-Newton updates that makes pose increments independent of the global map, enabling a pose-only GN algorithm equivalent to full GN, with the global map recovered in closed form after optimizing poses. The approach achieves a favorable balance of speed and accuracy, outperforming Cartographer and SLAM Toolbox and approaching Occupancy-SLAM in quality while dramatically reducing computation, thus enabling large-scale operation. Experiments on simulation and real datasets demonstrate solid scalability and robust map quality across diverse environments.

Abstract

Optimizing robot poses and the map simultaneously has been shown to provide more accurate SLAM results. However, for non-feature based SLAM approaches, directly optimizing all the robot poses and the whole map will greatly increase the computational cost, making SLAM problems difficult to solve in large-scale environments. To solve the 2D non-feature based SLAM problem in large-scale environments more accurately and efficiently, we propose the grid-based submap joining method. Specifically, we first formulate the 2D grid-based submap joining problem as a non-linear least squares (NLLS) form to optimize the global occupancy map and local submap frames simultaneously. We then prove that in solving the NLLS problem using Gauss-Newton (GN) method, the increments of the poses in each iteration are independent of the occupancy values of the global occupancy map. Based on this property, we propose a poseonly GN algorithm equivalent to full GN method to solve the NLLS problem. The proposed submap joining algorithm is very efficient due to the independent property and the pose-only solution. Evaluations using simulations and publicly available practical 2D laser datasets confirm the outperformance of our proposed method compared to the state-of-the-art methods in terms of efficiency and accuracy, as well as the ability to solve the grid-based SLAM problem in very large-scale environments.

Figures (6)

• Figure 1: The description of the grid-based submap joining problem. (a) and (b) show the grid-based submaps. (c) shows the global grid map generated by fusing the two submaps using the pose information only. (d) shows the consistent global grid map generated by our proposed grid-based submap joining method.
• Figure 2: Simulation environments and robot trajectory results. (a) and (b) show the simulation environments (the black lines indicate the obstacles in the scene) and the trajectories of ground truth, odometry inputs, Cartographer hess2016real, SLAM Toolbox macenski2021slam, and our method for one dataset in each of the two simulation experiments. Results from Occupancy-SLAM Zhao-RSS-22 are visually identical to our method and not drawn in the figure.
• Figure 3: The OGMs and point cloud maps generated by poses from ground truth, Cartographer, SLAM Toolbox, Occupancy-SLAM and our approach in one dataset for Simulation 1 (first two rows) and Simulation 2 (last two rows).
• Figure 4: OGMs from Cartographer, Occupancy-SLAM and our method. The first to third rows are Car Park, Deutsches Museum b0 1G and C5 dataset, respectively.
• Figure 5: Point cloud maps from Cartographer, Occupancy-SLAM and our method. The point cloud maps are generated by projecting the scan endpoints using poses.