This research project developed and deployed a novel vehicle modeling approach suitable for surfaces that aren't flat.
This approach captures nonplanar effects on smooth surfaces but remains computationally tractable for model-based control.
It is founded on the observation that many vehicles remain tangent to the surface they drive over.
This assumption is standard for existing control-oriented models, but is limited to flat surfaces.
We generalize to a nonplanar surface and derive a physics-based model from first principles.
We then apply standard kinematic and dynamic vehicle models and tackle previously unachievable control problems.
Applications include lane keeping on nonplanar surfaces such as off-camber turns, optimal racelines, overtaking, and path planning.
Models and Predictive Control for Nonplanar Vehicle Navigation
Vehicle Models and Optimal Control on a Nonplanar Surface
Overtaking Maneuvers on a Nonplanar Racetrack
https://github.com/thomasfork/Nonplanar-Vehicle-Control
This work studied forecast error of two vehicle models and the effect of numerical discretization on accuracy.
It also investigated numerical tractability and performance at varying speeds.Β
Closed loop control (MPC) was developed using the models and tested
Kinematic and dynamic vehicle models for autonomous driving control design