Leader: A. Robertsson, LU
Participants at LiU: L. Nielsen
Project description: Modeling and control of vehicle dynamics has successfully been used for energy-optimal
driving strategies, (semi-)autonomous driving and e.g., development of safety systems in automotive industry.
However, the need for fast re-planning of evasive maneuvers in critical traffic situations calls for a crucial interplay between efficient optimization algorithms and the use of simpler models capturing the essential dynamics together with fast feedback based on online measurements to achieve robustness to model uncertainties and driving conditions.
A complicating factor is being close to safety limits, leading to an intricate interplay between model complexity and expressiveness together with control and optimization. Similar problems arise in online re-planning and control of mobile and industrial robots subject to workspace sensing.
The goal of the project is thus to obtain techniques to handle situations with complicated nonlinear dynamics and
significant model uncertainty to be solved in time-critical situations.
The results obtained so far have mainly been investigating the results of different tire friction models and vehicle
behavior models in simulations and offline optimization scenarios. The investigated systems include both standard models reported in the literature and models identified from real test drives. Based on the previous results the new project will focus on the online aspects for re-planning and control and the introduction of more complex vehicle models representing e.g., truck and trailer systems, which comprise non-trivial extension due to significant differences in dynamic behavior. The traffic safety-related issues of heavy vehicles is one important driving factor behind this. Experimental facilities at the Vehicular Lab systems, Linköping University, and corresponding research infrastructure at RobotLab, Lund University, will be employed in e.g., the evaluation of the online path tracking algorithms.
The research is planned along two different routes which can be characterized as “online optimized maneuvers” and “narrow lane driving”, respectively. The former focuses on investigating and developing models for optimal control in critical situations (e.g. evasive maneuvers), as well as studying the maneuvering strategies that are obtained from these optimizations and how they can be utilized in future advanced safety systems and autonomous driving. The second is related to fundamental properties of path following relaxed to a “narrow lane” and the optimal control problems of both reaching and traversing along this lane subject to e.g. actuator saturation. Project research and outcomes will have synergies with, among others, theWallenberg Autonomous Systems Program (WASP) on safety and autonomous driving.