Session: 07-11-03 Mobile Robots and Unmanned Ground Vehicles III

Paper Number: 71088

Start Time: Tuesday, 07:40 PM

71088 - Navigation Method for UGV on Aerial Elevation Maps for Autonomous Missions in Sandy Deserts 

The majority of Unmanned Ground Vehicle(UGV) navigation research has been conducted in organized settings, such as sidewalks, indoor spaces, and warehouses, where the UGV is supposed to drive inside clearly defined regions. Outdoor GPS-based navigation currently is mostly used in agriculture on well-known, machine-accessible soil in predictable conditions. Planning stable trajectories for autonomous UGV in unstructured, challenging terrains such as deserts, swamps, mud, or lands covered in snow is difficult both for UGV’s safety, mission execution, and the operators. Autonomous navigation in unstructured terrain that relies solely on the fusion of sensors installed on the UGV (external -LiDAR, GPS, camera, internal - IMU, and Odometry) itself are impractical for predicting the upcoming terrain out of the sensors reach when attempting to reach GPS goal location. As a result, in previous experiments, our autonomous GNSS RTK based UGV faced wheel slippage and full immobilization in the sand. Situations with potential for rollover or immobilization include crossing a crest, after ascending a dune, and ditch crossing. In some action-oriented missions, the unpredictability of such sort may be the main reason for mission failure, given that autonomous UGV stranded in the already difficult terrain requires human assistance defeating the purpose of an autonomous mission. This paper presents a method of generating the heuristic, manoeuvrable path paired with proposing the nearest acceptable GPS point if the desired GPS goal is in the unreachable position. Path planning algorithm with implemented UGV’s constraints, paired with mission actions and GPS goals will generate accessible waypoints. Also, if one or more mission goals are inaccessible it will propose the nearest safe points for an operator to approve, securing the success of the autonomous mission. 

To address the problem of an unknown and dynamically shifting terrain we are utilizing a combination of Unmanned Aerial Vehicle (UAV) imager, conversion to a 3D point cloud terrain, generation of mission based, UGV constrained waypoints. The method requires 3 steps: Data collection - launching of a UAV in autonomous mode, fitted with a camera and the GNSS RTK allowing sub-centimeter accuracy geotagging. Capturing photographic images of a desert covering the general area of the desired UGV operation. Data processing - raw data is photogrammetrically converted into the site’s point cloud 3D landscape, where an elevation map is extracted. Waypoint generation - Relying on the terrain elevation, using pathfinding algorithm to generate waypoints based on the GPS goals and UGV’s limitations for rollover angle, track width, and wheelbase. Depending on the mission UGV’s can have variable weight affecting its driving capabilities, changing steepness angle it’s able to drive on, and ability to start after it stops, which is also going to be accounted for in this paper. 

By following this approach, a UGV equipped with LiDAR for autonomous obstacle avoidance needs only to follow generated waypoints delivered in precise GPS coordinates. The UGV is paired with GNSS RTK, minimizing localization errors, allowing it to smoothly drive to the proposed saved path. Our experiments are conducted in the sandy desert, which, with its dynamic dune formation is considered to be one of the most difficult terrains to drive on. Considering that topography changes in the desert, UAVs are sent out before every mission. The proposed UAV-assisted robotic waypoint generation is expected to greatly reduce human operator involvement, securing autonomous mission success, optimizing energy usage. We show the advantages of our approach compared to the existing approaches by the measurable reduction of immobilization events. 

 

Presenting Author: Marko Bjelotomic Dubai Electricity and Water Authority

Authors:

Marko Bjelotomic Dubai Electricity and Water Authority
Prashanth Subramaniam Dubai Electricity and Water Authority
Hesham Ismail Dubai Electricity and Water Authority
Abdallah Aljasmi Dubai Electricity and Water Authority