Mario Zanon

Chalmers University

Tuesday, July 04, 2017, 11:00

Room 02-012, Georges-Köhler Allee 102, Freiburg 79110, Germany

The coordination of autonomous vehicles approaching an intersection has recently attracted increasing research interest. The main motivation behind this research topic is to use automation in order to (a) reduce the amount of accidents, (b) reduce pollution and energy consumption and (c) increase the capacity of the infrastructure.

In order to achieve these three goals, it is necessary to introduce communication between the involved agents and the infrastructure, and to design a suitable algorithmic framework in order to compute a solution. The coordination problem can be framed as a distributed mixed-integer optimal control problem. While this class of problems has a high complexity and is known to be NP-hard, by designing tailored optimisation algorithms and heuristics one can aim at computing at least approximated solutions in a rather short time. This seminar introduces one of the steps towards such an algorithm: we assume a prescribed crossing order is given and focus on the continuous part of the problem, i.e. the solution of the distributed optimal control problem.

Note that the problem we address is not restricted to vehicles approaching an intersection, but can be framed more generally as a resource allocation problem subject to dynamic constraints. Such problems find applications in e.g. production facilities, traffic management, and logistics.