Stability Control Improvement and State Detection for Autonomous Vehicles

Existing sensors and algorithms have the potential for protecting Unmanned Ground Vehicles (UGVs) from impending collisions, rollovers, and loss of control and for warning and assisting drivers of manned vehicles of these same potential occurrences. However, more effort is needed to develop, refine, and combine existing methods so they are reliable and run in real-time on real hardware. In this phase II project, radar, stereo camera, and lidar data will be fused to generate a 3D model of obstacles and dangerous terrain ahead of a vehicle. In simulations of yaw and roll dynamics, a vehicle model with adaptive tire parameters will be run forward in time to predict future vehicle behavior. Collision, rollover, and yaw instability dangers will result in alerts to the driver or automatic correction in the case of an unmanned vehicle. A third mode of the system will assist a driver when these same conditions are anticipated. Nonlinear Model Predictive Control will be used as a framework to incorporate the sensing and control actions. Successful completion of this project will provide benefits to both manned and unmanned ground vehicles. Commercialization efforts will be an important part of this project.