Multiagent reconfiguration using hybrid optimal control.

摘要

The study of multiagent systems has gained great attention due to its flexibility and reconfigurability. In this thesis, the goal is to transfer n vehicles from random initial configurations onto a target line in minimum time, preferably without collisions. The single-agent optimal path computed by other researchers is applied to each of the n vehicles, however, collisions may occur.; This collision avoidance problem is opened and an intermediate step is to detect whether a collision will happen. A methodology is developed to determine all initial configurations that if a vehicle starts within this set, a collision will happen. This set of initial configurations is defined as the hazard region.; Another interesting problem is to reconfigure vehicles to a target line without specifying its position. A method is developed to find this optimal line, the target line where all the vehicles will get on in minimum time.