Non-GPS Based Vehicle Navigation


The research in GPS and Vehicle Dynamics Lab (GAVLAB) is investigating vision and IMU sensor fusion for accurate and robust vehicle navigation, which will also take into account vehicle dynamic constraints in areas where GPS is denied. Currently, work at the GAVLAB has shown that navigation by using relative obstacles is effective with the application of Lidar/IMU sensors. The foundation of the research was the use of Lidar to detect fixed points to calculate relative velocity, heading, and lateral error measurements in a corridor environment as shown in Figure 1. Lateral position estimation results from this investigation are shown in Figure 2. The specific environment tested was an indoor hallway, but many natural corridors exist in the form of curbs, ditches, or trees along a road. The current stage of research is investigating each of these scenarios. The above accomplishments can easily be adapted to integrate vision sensors with the IMU measurements. Cameras can be used to assist in the terrain recognition and locating of relative obstacles with the use of IMU navigation information. Finally, the results will improve the accuracy and robustness of the Non-GPS based vehicle navigation system.

Figure 1:


Heading and Lateral Error Calculation from Vision Data

Figure 2:


Lidar-IMU Estimate of Lateral Error