The project started by defining its use cases and scenarios for which the interaction between vehicles and infrastructure could have a high impact. This led to 16 use cases, which are categorized into three clusters:
• platoon management,
• signal optimization and
• Infrastructure to Vehicle (I2V) interaction.
This concluded that self-organizing platoons, with only indirect infrastructure management, are optimal. For the negotiation of signal plans, a new principle was developed that is based on mutually beneficial information exchange. This negotiation and the safety-related collective perception concepts quickly implied definition of new or extended message sets. With the prospect of upcoming impact assessment, a novel simulation architecture was developed that focussed on combining high realism of the whole system with fast simulation times to enable large-scale validation. Simpla was developed for fast simulation of automated vehicles in SUMO.
As mentioned before, the V2X message sets developed by MAVEN play a vital role in the proper functioning of the use cases. A new Lane Advisory Message (LAM) was developed, while the Collective Perception Message (CPM) was extended to support infrastructure sensors as well. For lane-specific speed advice a dedicated profile for interpretation of the Signal Phase and Timing (SPaT) and Map messages was developed. Lastly, the Cooperative Awareness Message (CAM) was extended to support the platooning and negotiation functionalities.
On the vehicle side, the automation logic is split into two levels. The trajectory planning ensures the vehicle drives an optimal path on a short horizon. The tactical level takes input from V2X communication to control the vehicle on a higher level. This is where the MAVEN extensions for platooning, lane changes and optimal speed for approaching an intersection were integrated.
For platooning, a detailed algorithm based on a state-machine and the new V2X message sets was developed. With V2X information about environmental perception, ADAS systems were developed and integrated in the control system to enhance safety.
Finally, extended High Definition (HD) maps were generated and adopted to enable automation on MAVEN focussed scenarios, like urban intersections.
The research in queue modelling showed 50% improvement of the queue length estimation at low penetration, while full penetration eliminated the error, which benefited all other use cases. A new stabilization cost function for adaptive control resulted in 25% reduction of average prediction error, while maintaining similar traffic efficiency. A patent was granted for this solution that improved speed advice accuracy.
The impact assessment simulations of the integrated green wave system with speed and lane advice showed that stopping at main corridors was eliminated:
• The total network capacity was increased by 34%,
• average queue length decreased by 74% and
• CO2 emissions reduced by 11%.
The negotiation-driven adaptive traffic control solution showed:
• a large reduction of 52% delay time
• the average queue length reduced by 46% and
• CO2 emissions were cut by 12%.
Lane advice excelled in reducing queue length and delay, but performed suboptimal in terms of stops and CO2 emissions. Deeper analysis pointed out that this could be resolved by recalibrating the traffic controller. Lastly, the system based on actuated traffic control also showed good performance, but with 5% CO2 reduction this was slightly less than the others.
Six integration sprints resulted in successful field trials and real-world demonstrations. These included platooning with interoperability between DLR and Hyundai vehicles as well as interoperability of these vehicles with the DLR and Dynniq infrastructure deployed in Braunschweig and Helmond, respectively. Here, vehicles could follow instructions to adapt their speed and lane fully autonomously. Safety use cases demonstrated that collaborative perception can prevent accidents in case of obstructed sensor view. The traffic efficiency use cases could also be experienced by the public at the final event. This is also shown in the figure of this summary.
These results show a clear benefit of the MAVEN systems and also led to clear exploitable results. The consortium used the BASE/x methodology to elaborate its exploitation strategy and developed models for the many results: new and extended message sets, urban platooning, I2V assisted automated driving at intersections, traffic management integration. These results will find their way into future products like cars, traffic control and traffic management systems. At the same time the knowledge will be incorporated in teaching at universities, while the transition roadmap will be an important guideline for policy makers for years to come.
