Research and/or Engineering Questions/Objective Although autonomous vehicle technology has undergone great advances in recent years, it is common for these vehicles to present disconnections in environments such as roundabouts or acceleration lanes and that the human driver must regain control. But this solution is not feasible when dealing with highly automated vehicles. This paper shows a coordinated vehicle management solution in which infrastructure and vehicles cooperate based on situational awareness of the whole scenario. Methodology The purpose is the development of a distributed decision-making system in which vehicles and infrastructure collaborate. For this, it has been necessary to implement novel perception solutions from the infrastructure based on computer vision and LiDAR technologies, the definition of specific communication protocols aimed at autonomous driving enhancing the information delivered in current cooperative systems oriented to informative services, and the development of distributed decision-making algorithms. Then, global decisions are taken at the infrastructure. These decisions provide a preliminary organization with prioritization and optimized vehicles speed profiles. Finally, vehicles in the critical road stretch receive those instructions and their local system adapts their final behavior considering the local perception around them. Results The simulations and tests of the proposed solution carried out in different environments show the effectiveness in traffic ordering. The developed system calculates the vehicles speed and trajectory. Communication specification has been completed and the specific messaging has been developed to support the requirements for each use case. Finally, driver's behavior has been analyzed as a basis for implementing naturalistic decisions. Limitations of this study The solution has been tested but real deployment requires high penetration of perception and communication systems. Furthermore, the distributed decision system is focused on specific road stretches but more relevant impacts are expected if several global decisions modules can work together. This fact involves a deeper knowledge of the scenario and a wider range for the situational awareness. What does the paper offer that is new in the field including in comparison to other work by the authors? The proposed solution integrates several technologies and systems to improve the real deployment of highly automated vehicles in complex environments as it is expected to reduce disconnections and provide vehicles with tools for managing those scenarios. The combination of perception systems from the vehicles and the infrastructure, and V2X communications provides an advantage in obtaining a reliable and complete situational awareness and this provides the possibility of a better choice when organizing traffic at global and local levels. Conclusions: In this paper, a distributed decision-making system between the infrastructure and the vehicle is presented to enable the operation of autonomous vehicles in complex environments with a double objective: to enhance the Operational Design Domain and, therefore, to reduce the situations in which the driver must retake control; and to manage traffic in an optimized way.
Prof. Dr.-Ing. FELIPE JIMENEZ, Professor, Technical University of Madrid