Automated guided vehicles are used for several decades in a number of different industrial applications where they allow for a robust and efficient operation. These automated vehicles typically follow fixed predefined lanes and stop movement in case of unexpected obstacles blocking their way. Hence, the algorithms used in such systems for local path planning and central control are rather simple. This class of automated vehicles are therefore not suitable for dynamic environments with people. Thus, the great potential of autonomous vehicles cannot be exploited while driving among people in public spaces, such as shopping centers, hospitals or public infrastructures.

The AutonomousFleet project explores the autonomy of vehicles as a system-of-system problem, which hasn’t been studied so far: the coordination and management of a significant number of autonomous vehicles for dynamic environments populated with people. Here, the coordination system of an autonomous vehicle poses similarities with the traffic control system for vehicle drivers on roads - it may only affect the driving behavior but cannot control it. In addition, other challenges arise since environments that are frequented by pedestrians have greater degrees of freedom concerning the choice of lanes. The coordination system must be able to recognize obstacles or blocking in time to initiate adequate countermeasures. Moreover, the holistic view of the system in a populated environment is of vital importance for its acceptance. Processes have to be plausible and people need to be able to contribute their knowledge to the current system.

The main objective of the AutonomousFleet project is to develop the scientific basis for the coordination and navigation of several autonomous vehicles. The research activities focus on a coordination system which provides system-aware routes for autonomous vehicles as well as on the analysis of the local situation for path planning. An important component of the system is integration and processing of on-board data from vehicles and inclusion of expert knowledge. These different sources of information will be processed in a multi-layer knowledge base that serves as a basis for the various levels of the system.

The research results will enable efficiently performed transport tasks with a fleet of several autonomous vehicles in environments shared with people. Additionally, the autonomous vehicle fleet provides real-time data on the built environment (e.g. structural changes) and events (e.g. estimation of traffic state and crowd flows).