Lane density optimisation of automated vehicles for highway congestion control

Abstract

The rapid development of automated vehicle (AV) technology and increasing deployment in the near future present novel opportunities for congestion control on highways. Collective dynamics of AVs allow for network-wide traffic management, offering more flexible centralised and decentralised control schemes. To capitalise on this technology, control strategies need to be developed to enable traffic management to take advantage of AV potential to manipulate traffic flow dynamics. This paper presents a two-level traffic control method to relieve highway congestion by controlling lateral flows of AVs. The first, a proactive lane density distribution optimisation problem to establish the optimal vehicle density in lanes upstream of bottlenecks. The second, a reactive lane change advisory system to tackle local merging manoeuvres and resolve merge conflicts. The effectiveness of the proposed strategy is demonstrated through microsimulation experiments and comparison to ramp metering strategy. To consider the transition period and gradual uptake of AVs, the proposed control is also tested at various penetration rates of AVs. At moderate to high levels of AV penetration, the control is demonstrated to be superior to ramp metering within a range of traffic demands. The proposed traffic control reduces total travel time of all vehicles and travel time variation among vehicles on the mainline and ramps. The effectiveness is attributed to the delayed onset and severity of congestion and subsequent capacity drop. The success of the proposed strategy is illustrated on a three-lane highway with a single on-ramp and further exhibited on a road network with multiple on-ramps and off-ramps.