Are CAVs the cure for traffic inefficiencies on our roads?
Autonomous driving offers a solution for many current traffic issues. But until we will reach high penetration rates of CAVs, we still need to resolve a lot of uncertainties, for example when it comes to the control algorithms and variables implemented in the vehicles. Introducing CAVs on our roads, where they will first coexist with human-driven vehicles, will potentially result in more traffic inefficiencies. For safety reasons, autonomous vehicles will need to keep larger safety distances to conventional vehicles. This reduces the number of vehicles travelling the network and leads to inefficiencies.
Accelerating the future of mobility with more CAVs and fewer delays
Together with Jochen and Peter from PTV Group, I investigated how the throughput of vehicles recovers, as the penetration rate of CAVs increases. We have developed a simplified network to prove the reduction in throughput for lower penetration rates and analysed which percentage of CAVs is necessary for it to recover again. Our calculations are based on a function of the headway kept between CAVs and human-driven vehicles.
We also considered that the higher the penetration rate the greater the performance impact of CAVs, since the vehicles can make use of communicating and cooperating to flow in platoons. For low penetration rates, there are less opportunities to form platoons. To keep our approach attractive for both users and city planners, following the idea of an economical and practical implementation of CAVs without costly road infrastructure modifications, we have utilised the existing network designs to promote length-constrained platoons. This increases throughput and reduces control delay at signalised intersections, even when most of the vehicles are still operated by human drivers.
“I’m passionate about emerging technologies and new modes of transport. As an engineer and researcher, I combine the two to unlock the potential of CAV platooning for a safer and more efficient mobility of the future.”
Testing platooning scenarios with microsimulation
In PTV Vissim, we implemented an existing infrastructure design at signalised intersections, where we used the additional lane provided to cross the intersection. To allow length-constrained platoon build-up following, an ad-hoc strategy was implemented through formula-based routing and green pre-emption. Platoons are promoted in one of the approaching lanes. Since this extra lane is a bay with a specific length, platoons will also have a specific length. In a base scenario, where the same existing network and green pre-emption strategy was kept invariable, vehicles are distributed randomly to each of the lanes.
Limiting the length of platoons increases traffic efficiencies
The comparison of both scenarios for all penetration rates – the one with the formula-based routing and the base scenario, where vehicles are distributed randomly – showed that the platoon formation with a limited length is a good solution avoiding the costly implementation and maintenance of expensive roadside units for vehicle-to-infrastructure communications.
With our formula-based routing strategy at the intersection, the throughput is improved for the first stages of CAV introduction in the traffic stream because of the platooning, and for all penetration rates. Looking at the high demand scenarios, the simulation results showed that the delay can be reduced up to 18% with low penetration rates of 15%.
Improving traffic efficiency without implementing roadside units for vehicle-to-infrastructure communications
We are really satisfied that even assuming the most conservative behaviour for CAVs, that is keeping headways of two seconds to a human-driven vehicle, the accurate utilisation of an existing network design can provide such good results. Now we can’t wait for new implementations and practice-ready solutions to ease the introduction of CAVs on our roads and see how our research results are put into practice.