Integrating DEM Path Planning in the Navigation and Control System for UGVs in Rough Terrain: A Case Study on Multiple Victim Evacuation

Unmanned ground vehicles (UGVs) can play a substantial role in disaster robotics for applications such as transporting supplies and equipment, evacuating victims, or conducting surveillance. However, the rough and irregular terrain found in disaster sites presents a challenge for autonomous UGV navigation. In previous work, we introduced DEM-AIA, a graph-based search algorithm for path planning in digital elevation models (DEMs) that accounts for terrain inclination and vehicle limitations. Nevertheless, executing planned paths with an actual vehicle is challenging due to the complex interactions between the vehicle and the soil. In this paper, we address the integration of DEM path planning within a framework for the autonomous navigation of a heavy-duty UGV on rough terrain during a disaster response exercise with first responders. In the exercise, a skid-steer 8-wheeled vehicle equipped with two onboard stretchers autonomously plans and executes a trajectory to evacuate two victims, who are extricated by rescuers at two different locations. Path tracking is achieved with a variant of the pure-pursuit algorithm using the fusion of GNSS, IMU, and visual odometry. The paper presents the motion control architecture and details of the exercise.