An outdoor delivery robot requires autonomous navigation technologies, such as map generation, driving area definition, path generation, and control. However, integrating the technologies of each field can be difficult due to verification in different experimental environments and hardware. This study presents a viable approach for outdoor mobile robots by integrating mapping, planning, and experiments using Autoware with a low-cost LiDAR sensor. To achieve this goal, we compare the performance of various LiDAR SLAM algorithms to generate precise 3D point cloud maps. This enables us to further create high-definition (HD) maps which are used for safe navigation and positioning of outdoor mobile robots. Then, we validate the performance of the mapping, localization, and planning algorithm in Autoware through simulations using CARLA and real-world experiments. To validate the driving performance of our autonomous mobile robot, we performed a driving test on road and sidewalk navigation, utilizing an HD map of a university campus generated over a travel distance of approximately 5.68 km. Furthermore, to enhance stability in the sidewalk test scenario, we developed and tested a road segmentation-based dynamic obstacle avoidance algorithm. Through analysis of the experimental and simulation results, our paper provides additional insights into precautions for operating outdoor mobile robots.
