Freeway Congestion Management With Reinforcement Learning Headway Control of Connected and Autonomous Vehicles

Abstract

Adaptive cruise control (ACC) systems are increasingly offered in new vehicles in the market today, and they form a core building block for future full autonomous driving. ACC systems allow vehicles to maintain a desired headway to a leading vehicle automatically. Recent research demonstrates that (1) shorter headways lead to higher throughput, and (2) the effective use of ACC can improve traffic flow by adapting the desired time headway in response to changing traffic conditions. In this paper we show that, although shorter headways result in higher capacity, flow breakdown still occurs if traffic densities at bottlenecks are allowed to exceed the critical density. Therefore, dynamic traffic control near bottlenecks is still necessary to avoid bottleneck activation and capacity loss. We propose an adaptive reinforcement learning (RL) headway controller that uses ACC headways to optimize traffic flow and minimize delay. Based on state measurements, the controller dynamically assigns an optimal headway value for each freeway section within a control cycle. In a freeway simulation example, we first demonstrate that different nondynamic headway assignment strategies failed to avoid congestion and traffic breakdown. We then present a dynamic headway control strategy based on deep reinforcement learning (DRL) that adapts the desired headway according to the changing traffic conditions on both the freeway and the ramp to effectively maximize traffic flow and minimize system delay. We quantitatively demonstrate that our DRL dynamic headway control strategy improved traffic and reduced system delay by up to 57% compared with the examined nondynamic headways.

Freeway congestion management on multiple consecutive bottlenecks with RL‐based headway control of autonomous vehicles

“…• Ramp only control yielded better results than ramp and mainline control: It has been observed that the delay performance was the best when the ramp sections only were controlled as opposed to controlling the ramp and mainline sections together. This has been observed in our earlier work [35] and can be justified by the fact that the freeway delay contributes heavily to the system delay as opposed to the ramp delay due to the relatively higher freeway demand. Also, animated traffic simulations have shown that any speed disturbances on the freeway (to adjust the headway) will cause spillback and affect the whole freeway stream thereby affecting the delay of more vehicles as opposed to the ramp delay which may have a local effect on the corresponding ramp vehicles only.…”

“…This approach is conceptually similar to the RL control studies in [30][31][32][33][34]. In our previous work in [35], we demonstrated that although tighter headways improve road capacity and traffic congestion, a breakdown can still occur at higher densities, which required further attention and more elaborate dynamic optimal headway control near bottlenecks. Therefore, we proposed and developed an adaptive DRL headway controller that varies headways at the infrastructural level, that is, on a segmentby-segment basis, to optimize traffic flow and minimize delay.…”

“…In this paper, we further extend our work in [35] such that RL-based headways are controlled on a network with multiple bottlenecks represented by multiple consecutive ramps. This is done to capture the effect of possible congestion spillback when controlling several consecutive bottlenecks as well as to ensure the system delay enhancement despite the increase in local delays as a result of control.…”

“…Therefore, we proposed and developed an adaptive DRL headway controller that varies headways at the infrastructural level, that is, on a segment‐by‐segment basis, to optimize traffic flow and minimize delay. Our work in [35] examined the impact of the proposed control strategy on a generic freeway scenario with one on‐ramp bottleneck and demonstrated that the proposed strategy improves traffic and enhances system delay compared to the nondynamic headway scenarios. The assumption in [35] is that bottlenecks are independent of each other, and the controller, therefore, focuses on only one on‐ramp bottleneck.…”

“…Our work in [35] examined the impact of the proposed control strategy on a generic freeway scenario with one on‐ramp bottleneck and demonstrated that the proposed strategy improves traffic and enhances system delay compared to the nondynamic headway scenarios. The assumption in [35] is that bottlenecks are independent of each other, and the controller, therefore, focuses on only one on‐ramp bottleneck.…”