The GPS and Vehicle Dynamics Lab at Auburn University is currently working to improve the navigation performance of a low-cost UGV. These mobile robots (Figure 1) are to be used in roadway safety applications to automatically deploy and position safety barrels accurately. The robot consists of mostly commercial off-the-shelf components with a production goal of $200 per robot. From initial performance tests, the robot is unable to adequately navigate itself to be used in such position critical areas. Simulations have been developed to determine what error sources arise in the use of the robot as well as to test various navigation methods. The simulation uses the same control algorithm as the UGV with various error sources injected such as wheel slip, motor output errors, UGV parameter errors, etc. The resulting final position error was compared to performance tests run by UNL. Figure 2 shows a comparison between 100 actual data runs and a statistical simulation with expected errors injected. The next step in the project is to develop a sensor fusion algorithm which will use a GPS receiver, wheel odometry sensors, and inertial sensors to improve the navigation of the UGV. The simulation will be used to determine which algorithm and what sensors are necessary to improve navigation performance. Following this analysis, the actual system will be updated and tested for use in highly demanding situations such as roadway safety situations.
Small UGV with GPS-INS Board
Navigation Errors from Wheel Odometer Dead Reckoning