Automated vehicle (AV) features such as adaptive cruise control could significantly increase driver safety and mobility. But in some circumstances, these features can alter vehicle movement and spacing, and interfere with traffic flow. This project analyzed the integration of varying levels of AVs with human-driven vehicles (HVs) and the impact on ramp meter operations to measure the effects and identify potential solutions and modifications.
Minnesota is a global leader in the deployment of ramp meters. To control the flow of traffic onto roadways, MnDOT traffic engineers establish algorithms for ramp metering. Integrating AVs and HVs, which lack AV features, may negatively impact traffic flow and performance on road networks. For example, adaptive cruise control in AVs creates more headway space between vehicles than typical HVs, causing changes to established traffic patterns. As a result, the ramp meters are not properly coordinated with traffic flow, and some performance metrics suffer.
To better prepare traffic control infrastructure for the increasing integration and impacts of AVs, this project examined how AV features affect traffic flow dynamics and how ramp metering adjustments can mitigate these effects.
What Did We Do?
Five test sites were selected representing a range of ramp metering scenarios. A category was assigned to vehicles based on their automated functionality:
- HVs: The driver is responsible for all aspects of driving.
- Low-level automated vehicles (ACCs): The vehicle has one or two automated features, such as adaptive cruise control and lane-keeping assistance.
- AVs: The vehicle can operate most or all aspects of driving.
Using traffic modeling software, investigators simulated peak-hour volume traffic flows at the five sites for seven scenarios, including all HVs, all AVs and mixed vehicle scenarios. Vehicles entered the highway through ramp meters operating with MnDOT’s density adaptive metering algorithm. Roadway traffic conditions were evaluated based on throughput and average travel time. On-ramp conditions were evaluated based on average wait time and queue lengths. The HV scenario served as the baseline.
After the initial round of simulations, the ramp metering algorithm underwent two adjustments to improve performance. The first adjustment focused on decreasing wait times for vehicles entering the highway; the second adjustment attempted to improve throughput and travel time on the highway. Researchers repeated the simulations after each adjustment and compared the results to determine the effectiveness of the adjustments.
What Did We Learn?
The first round of simulations indicated negative impacts were greatest when the vehicles were all operating adaptive cruise control. Vehicle throughput decreased 58.4%, and travel time increased 61.1%. This was primarily due to conservative spacing behavior inherent in adaptive cruise control systems that increases intervehicle headway spacing, reduces traffic throughput and leads to ramp metering inefficiency. Simulations with a high ACC vehicle mix—consisting of 60% ACC, 30% HV and 10% AV—and a high HV vehicle mix—consisting of 70% HV and 30% ACC—also demonstrated negative outcomes. The AVs-only scenario achieved the best traffic metrics, which demonstrated the potential benefits of moving toward high saturation rates of AVs.
“The results of the simulations using adjusted ramp metering algorithms are encouraging and demonstrate tools MnDOT has at its disposal to mitigate traffic flow issues that may arise due to integrating automated vehicle technology,” said Garrett Schreiner, freeway operations engineer, MnDOT Ramp Meters.
Simulation results incorporating the two adjusted algorithms demonstrated successful outcomes. The first adjustment resulted in small improvements in highway throughput while showing significant improvements in average ramp wait time. The second adjustment improved highway throughput and travel time with less improvement in ramp waiting times than after the first adjustment.
Overall, the findings indicate mixed-autonomy traffic will result in different traffic flow characteristics, which will vary based on the vehicle mix on the road. MnDOT’s ramp metering algorithm will continue to function, but it can be adjusted to improve traffic performance metrics, if needed.
What’s Next?
The results from the simulations adjusting ramp metering algorithms are a positive indication for traffic management as more AVs and features make their way onto the road. While MnDOT has no immediate plans to adjust its algorithm, the agency can address traffic flow to ensure performance. It is also important for MnDOT to ensure that traffic management algorithm variables remain adjustable for purposes such as this as the algorithm evolves. Further research could analyze additional algorithm refinement with real-world trials and the development of real-time estimation techniques for a methodology that incorporates site-specific factors and the real-time composition of vehicle types.
Crossroads 