A Middle Way to Traffic Enlightenment

TL;DR

The paper tackles the mismatch between infrastructure VSLs and real-world multi-lane traffic by introducing a cooperative automated vehicle controller that both respects posted limits and adapts to local, real-time traffic. It combines a middleway speed setpoint with a safety-filtered low-level controller to avoid unsafe slowdowns while maintaining safety, validated through field tests on I-24 with two control Rav4s. Key contributions include a formal middleway speed law, a CBF-based safety override, and real-time local traffic estimation via onboard radar in lieu of latency-prone external data. Results show substantial mode usage in live congestion, with the controller spending most time in safety-following while also leveraging VSL and middleground modes as traffic waves evolve. The work demonstrates practical viability for scalable, low-cost deployment of VSL-coordinated, mixed-autonomy traffic management on open highways.

Abstract

This paper introduces a novel approach that seeks a middle ground for traffic control in multi-lane congestion, where prevailing traffic speeds are too fast, and speed recommendations designed to dampen traffic waves are too slow. Advanced controllers that modify the speed of an automated car for wave-dampening, eco-driving, or other goals, typically are designed with forward collision safety in mind. Our approach goes further, by considering how dangerous it can be for a controller to drive so slowly relative to prevailing traffic that it creates a significant issue for safety and comfort. This paper explores open-road scenarios where large gaps between prevailing speeds and desired speeds can exist, specifically when infrastructure-based variable speed limit systems are not strictly followed at all times by other drivers. Our designed, implemented, and deployed algorithm is able to follow variable speed limits when others also follow it, avoid collisions with vehicles ahead, and adapt to prevailing traffic when other motorists are traveling well above the posted speeds. The key is to reject unsafe speed recommendations from infrastructure-based traffic smoothing systems, based on real-time local traffic conditions observed by the vehicle under control. This solution is implemented and deployed on two control vehicles in heavy multi-lane highway congestion. The results include analysis from system design, and field tests that validate the system's performance using an existing Variable Speed Limit system as the external source for speed recommendations, and the on-board sensors of a stock Toyota Rav4 for inputs that estimate the prevailing speed of traffic around the vehicle under control.

Figures (12)

• Figure 1: Recurring Dilemma: Variable Speed Limit (VSL) gantry shows a 30 mph speed limit on Interstate-24. Prevailing traffic is shown to regularly exceed the VSL by a large margin. In this work we demonstrate an automated vehicle controller that follows the VSL on the gantry when nearby vehicles do, and adopts a higher speed when prevailing traffic is moving much faster than the posted speed.
• Figure 2: Environment: Overview of the control environment. The VSL system measures downstream traffic for aggregate traffic information. A control vehicle (blue) can measure timely information about highly local traffic in front and adjacent to the vehicle (red). Our controller changes the set speed based on information from both of these sources.
• Figure 3: Speed Selection: The system architecture to determine the speed setting based on the VSL gantry, the state of the cruise controller, GPS information, and the radar data. The equation for the middleway algorithm is shown in equation \ref{['equ:middleway']}.
• Figure 4: Speed Controller: this acceleration-based controller is a replacement for the OEM cruise controller. The controller takes an input speed setting, $v_{des}$, and sends acceleration commands to the vehicle through libpanda. The speed controller is based on rate limiting $v_{des}$, a nominal proportional controller, and a CBF to perform dynamic filtering to provide car following and prevent collisions.
• Figure 5: Experimental Deployment: Four vehicles, pictured here, are launched into early morning congestion on Interstate-24. From right to left, they enter into the traffic flow. Vehicles 2 and 4 are instrumented for experimental control, and vehicles 1 and 3 are operated under human-piloted control.