Autonomous Vehicle front steering control computation saving
Julián Salt Llobregat, Julián Salt Ducajú
TL;DR
The paper tackles reliable lane-keeping yaw-rate control for autonomous vehicles under resource constraints by modeling lateral dynamics as an LPV system and designing a high-order μ-synthesis controller. It then implements the controller with interlaced computation and discrete lifting to split fast and slow dynamics, enabling reduced processor load while maintaining performance, demonstrated with a discretized controller $C_d(z)$ at $T=0.01$ s. The main contributions include a practical interlacing framework, lifting-based analysis, and a real-world validation on a 2017 Lincoln with urban-circuit testing, showing negligible performance loss in path tracking. This work offers a viable path to scalable, robust lane-keeping within Intelligent Vehicle Networks by alleviating network and computation bottlenecks. The approach combines robust control design with dual-rate, lifted representations to enable efficient implementation without compromising safety margins.
Abstract
For autonomous vehicles lane keeping purposes it is crucial to control the vehicle yaw rate. As it is known a vehicle yaw rate control can be achieved handling the steering angle. One option is to consider a robust controller and depending of the requirements the synthesis can drive to a high order controller. Nowadays this kind of vehicles needs a networked based control (IVN -Intelligent Vehicle Network-)with a considerable amount of control loops for different vehicle components. Therefore, in this environment the controllers computation saving could be a good option for unload the network and digital processors. That is the main target of this contribution; in order to accomplish this goal a interlacing implementation technique is considered. Results in a real path tracking illustrates viability of this procedure.