Description

The main objective of the PhD Thesis focuses on the combination of set-based approaches and optimization techniques for the coordination of autonomous vehicles, ensuring safety for the ego-vehicle in dynamic environments.

For this purpose, 4 specific objectives have been defined:
1. Design and development of a motion planner based on optimization techniques for an autonomous vehicle with dynamic environments.
2. Design and development of a motion planner with emergent techniques oriented to critical situations and proposed scenarios where different shared resources such as roundabouts or intersections.
3. Design and development of a motion planner for coordinated vehicles following an agreement protocol between the vehicles involved using common resources.
4. Extension of the developed movement planners to other aspects of safety and security.

For the development of these objectives, special emphasis is being placed on the performance of Reachability Analysis through the use of zonotopes. This is both for the estimation of nearby vehicles and for defining which ego-vehicle states meet safety criteria. Once the safe domain of states is defined, MPC-based approaches are being designed to obtain the optimal motion plan.
In addition, other families of ensemble-based methodologies and their possible application for safety in autonomous driving have been evaluated throughout the thesis. Such as the applicability of the invariance theory or the use of viability kernels.

Finally, it should be noted that during the completion of this doctoral thesis, a 4-month international placement was carried out at Linköping University in the Vehicular Systems section.