OpenStreetMap-Based Autonomous Navigation for the Four Wheel-Legged Robot Via 3D-Lidar and CCD Camera

TLDR

OpenStreetMap provides publicly available road network data widely used for outdoor robot navigation. The study proposes an OSM‑based navigation method that fuses road network data with local perception to address map‑error induced path inconsistencies. The method obtains a global path via Dijkstra on OSM, fuses 3D‑LiDAR and CCD camera data to detect local roads and obstacles, filters and extracts road features to refine the local path, and then uses this refined path for robot tracking. Experiments show the robot’s trajectory deviates only 0.18 m from the road center, demonstrating high accuracy and robustness in complex real environments.

Abstract

OpenStreetMap (OSM) is widely used in outdoor navigation research recently, which is publicly available and can provide a wide range of road information for outdoor robot navigation. In this article, aiming at the problem that the map error of OSM will cause the global path to be inconsistent with the real environment, we propose an OSM-based robot navigation method that combines road network information and local perception information. As a global map, OSM provides road network information to obtain the global path by the Dijkstra algorithm. Multisensor (including 3D-LiDAR and Charge-coupled Device (CCD) camera) information fusion offers local information to detect local road information and obstacles for local path planning. We filter local road information and then extract useful road features to optimize the local path. Finally, this local path is used for robot path tracking to complete navigation tasks. The experimental results show that the average error between the trajectory of the robot and the road center is 0.18 m. This reveals that our method has high navigation accuracy and strong robustness in the real complex environment.