Estimation of Relative Motion between Non-Rigid Bodies on a Semi-Truck

Abstract

This thesis develops methods to estimate the relative motion between the cabin and chassis of a semi-truck without the use of a complex suspension model between the two bodies. In, this thesis only sensors that would be already available on current vehicle using both inertial and GPS sensors on the cabin and chassis bodies are utilized. Automation of Semi-Trucks requires perceiving obstacles that are both stationary and moving that exist close to the truck. The cabin and chassis of a truck have a suspension system between them. This can cause there to be offsets between the perception of obstacles from the cabin vs the chassis which can inhibit autonomy of the vehicle. The thesis analyzes both the use of GPS/INS integration as well as Transfer Alignment techniques to correct an IMU mounted on the chassis with measurements taken from GPS an tennas and IMUs mounted on the cabin. Additional methods are investigated to further improve these techniques such as taking into account the quality of the cabin measurements, adding higher fidelity models for state estimates, or reducing the amount of states. Using real world datasets, the methods are evaluated by comparing the corrected chassis solution to a high qual ity GPS/INS senor that acts as a truth measurement. The results show that the relative motion between the two bodies can be determined. The quality of the final solution is dependent on the performance of both the cabin solution and chassis IMU. The thesis will show with a high quality cabin GPS/INS solution the chassis attitude solution can be found to be within 0.5 degrees of the true value.