Periodic Reporting for period 1 - C-AVOID (Connected – Autonomous – Vehicles Orchestrated with Intelligent Decisions)
Berichtszeitraum: 2020-10-01 bis 2022-09-30
Recent studies have shown that accidents on the road infrastructure result in more than $1.35$ million worldwide casualties annually, which justifies the fact that safety is one of the top priorities for the automotive sector. Several efforts have been made in order to improve the situation. The most attractive solutions nowadays involve the use of wireless connectivity at the vehicles and the introduction of autonomous vehicles.
Through wireless connectivity, vehicles exchange messages with each other and with the infrastructure, obtaining additional information on the road infrastructure’s status. Such additional information is critical to improve the safety of transportation networks, since: (i) it allows vehicles to “see” events that are outside the field of view of the driver and of any sensor that modern cars have on board (camera, LIDAR, radar, …); and (ii) it allows predicting, or receiving a prediction, on an event that is about to happen, e.g. increasing the chance of avoiding an imminent danger. Autonomous vehicles, instead, have the potential of increasing the efficiency of modern transportation networks. Thanks to the use of advanced technologies, autonomous vehicles can get safely closer to each other and, as a result, they can reduce congestion and travel time, and they can improve fuel efficiency, potentially bringing consistent benefits to society.
The activities of C-AVOID aim at exploiting, simultaneously or independently, both wireless connectivity and autonomous vehicle. Thanks to an optimized use of these revolutionary and disruptive technologies, C-AVOID proposes novel solutions to improve the safety and the efficiency of the road transportation of the future.
Through wireless connectivity, vehicles exchange messages with each other and with the infrastructure, obtaining additional information on the road infrastructure’s status. Such additional information is critical to improve the safety of transportation networks, since: (i) it allows vehicles to “see” events that are outside the field of view of the driver and of any sensor that modern cars have on board (camera, LIDAR, radar, …); and (ii) it allows predicting, or receiving a prediction, on an event that is about to happen, e.g. increasing the chance of avoiding an imminent danger. Autonomous vehicles, instead, have the potential of increasing the efficiency of modern transportation networks. Thanks to the use of advanced technologies, autonomous vehicles can get safely closer to each other and, as a result, they can reduce congestion and travel time, and they can improve fuel efficiency, potentially bringing consistent benefits to society.
The activities of C-AVOID aim at exploiting, simultaneously or independently, both wireless connectivity and autonomous vehicle. Thanks to an optimized use of these revolutionary and disruptive technologies, C-AVOID proposes novel solutions to improve the safety and the efficiency of the road transportation of the future.
Thanks to the support provided by the faster, more reliable, and more capable new generation of cellular networks, i.e. 5G, the C-AVOID project delivered two main frameworks for the automotive sector:
• a methodology that allows predicting collisions at urban intersections well in advance, so that human drivers and autonomous vehicles can avoid the imminent danger;
• an optimization framework that coordinates autonomous vehicles in congested road sections, so to improve safety and use efficiently the transportation network.
The high level of accuracy in detecting collisions ahead of time achieved in C-AVOID is possible thanks to vehicles’ trajectory predictions at urban intersections of unprecedented precision, even if drivers have a large set of possible maneuvers ahead. This is obtained thanks to wireless connectivity, which allows collecting of vast data from the area of interest from both vehicles and infrastructure. Trajectory predictions are then built “learning” the drivers' future decisions based on: the status of surrounding vehicles, the lane and the speed used by vehicles approaching the intersection, and the status of the road infrastructure, e.g. of the traffic lights.
Thanks to wireless connectivity, C-AVOID also proposes a new optimization framework that is able to coordinate autonomous vehicles approaching an intersection. The framework accounts for sensor measurement inaccuracies, e.g. due to imprecise GPS readings, leaving among vehicles a dynamic safe distance that ensures no collisions. The overall objective of the proposed solution is to reduce the average vehicle crossing time, effectively improving the efficiency of the road transportation networks. The formulation of the obtained optimization framework is quite general, and it was easily adapted also to 3 dimensions, i.e. to Unmanned Aerial Vehicles.
Finally, in the above-mentioned applications, C-AVOID explored different trade-offs: between solution optimality and communication/computational overhead; between running the algorithms at the vehicles, or running them in a centralized fashion within the cellular network infrastructure.
The project thus far delivered seven scientific publications, which include three prestigious journal publications to: (i) the IEEE Transactions on Intelligent Transportation Systems, (ii) the Journal of Intelligent & Robotic Systems, and (iii) the EURASIP Journal on Wireless Communications and Networking; and four conference publications to: (i) the IEEE Vehicular Technology Conference (twice); (ii) the IEEE International Intelligent Transportation Systems Conference, and (iii) the International Symposium on Modeling and Optimization in Mobile, Ad hoc, and Wireless Networks. Three more scientific publications are either under submission or close to be submitted.
• a methodology that allows predicting collisions at urban intersections well in advance, so that human drivers and autonomous vehicles can avoid the imminent danger;
• an optimization framework that coordinates autonomous vehicles in congested road sections, so to improve safety and use efficiently the transportation network.
The high level of accuracy in detecting collisions ahead of time achieved in C-AVOID is possible thanks to vehicles’ trajectory predictions at urban intersections of unprecedented precision, even if drivers have a large set of possible maneuvers ahead. This is obtained thanks to wireless connectivity, which allows collecting of vast data from the area of interest from both vehicles and infrastructure. Trajectory predictions are then built “learning” the drivers' future decisions based on: the status of surrounding vehicles, the lane and the speed used by vehicles approaching the intersection, and the status of the road infrastructure, e.g. of the traffic lights.
Thanks to wireless connectivity, C-AVOID also proposes a new optimization framework that is able to coordinate autonomous vehicles approaching an intersection. The framework accounts for sensor measurement inaccuracies, e.g. due to imprecise GPS readings, leaving among vehicles a dynamic safe distance that ensures no collisions. The overall objective of the proposed solution is to reduce the average vehicle crossing time, effectively improving the efficiency of the road transportation networks. The formulation of the obtained optimization framework is quite general, and it was easily adapted also to 3 dimensions, i.e. to Unmanned Aerial Vehicles.
Finally, in the above-mentioned applications, C-AVOID explored different trade-offs: between solution optimality and communication/computational overhead; between running the algorithms at the vehicles, or running them in a centralized fashion within the cellular network infrastructure.
The project thus far delivered seven scientific publications, which include three prestigious journal publications to: (i) the IEEE Transactions on Intelligent Transportation Systems, (ii) the Journal of Intelligent & Robotic Systems, and (iii) the EURASIP Journal on Wireless Communications and Networking; and four conference publications to: (i) the IEEE Vehicular Technology Conference (twice); (ii) the IEEE International Intelligent Transportation Systems Conference, and (iii) the International Symposium on Modeling and Optimization in Mobile, Ad hoc, and Wireless Networks. Three more scientific publications are either under submission or close to be submitted.
Even if wireless connectivity and autonomous vehicle adoption are getting traction, C-AVOID obtained results that filled important gaps present in existing solutions. First, it exploits the cellular network support to improve safety at urban intersections. Typical collision predictions rely on onboard sensors or on communication from nearby vehicles. In C-AVOID, an intelligent mechanism monitors an area of interest through time, “learning” the best model to predict what will happen in the short-term future and to detect collision in advance. Second, exploiting frequent communications from autonomous vehicles, the autonomous intersection management approach presented in C-AVOID allows coordination between vehicles even in presence of sensor measurement errors and imprecise trajectory predictions.
Contrary to most of the state-of-the-art approaches, in order for the outputs of the project not to have only a theoretical innovation potential, in C-AVOID, several practical implementations have been proposed. First, event-triggering techniques were embedded in the proposed solutions, so to reduce both the communication and computational complexity. Second, the networking part of the proposed collision avoidance solution was successfully tested in the open-source cellular implementation of Open-Air Interface.
Finally, the activities carried out in C-AVOID potentially affect cellular operators, and car manufacturers/transportation authorities. For the former, network-assisted safety applications represent a new market possibility, since they open up the use of 5G networks to the automotive vertical. For the latter, improving the safety of the road transportation networks is nowadays of utmost relevance, given the number of casualties that road accidents cause every year. Similarly, autonomous vehicle coordination has the potential to increase the capacity of the road transportation of the future, reducing congestion and vehicles’ travel time.
Contrary to most of the state-of-the-art approaches, in order for the outputs of the project not to have only a theoretical innovation potential, in C-AVOID, several practical implementations have been proposed. First, event-triggering techniques were embedded in the proposed solutions, so to reduce both the communication and computational complexity. Second, the networking part of the proposed collision avoidance solution was successfully tested in the open-source cellular implementation of Open-Air Interface.
Finally, the activities carried out in C-AVOID potentially affect cellular operators, and car manufacturers/transportation authorities. For the former, network-assisted safety applications represent a new market possibility, since they open up the use of 5G networks to the automotive vertical. For the latter, improving the safety of the road transportation networks is nowadays of utmost relevance, given the number of casualties that road accidents cause every year. Similarly, autonomous vehicle coordination has the potential to increase the capacity of the road transportation of the future, reducing congestion and vehicles’ travel time.