Simulation study on human interactions with CACC vehicles finds that local coordination methods scale more efficiently to market penetration than ad hoc methods.

University researchers model the impacts of CACC at varied market penetration rates (MPRs).

Date Posted
10/24/2019
TwitterLinkedInFacebook
Identifier
2019-L00915

Clustering Strategies of Cooperative Adaptive Cruise Control: Impacts on Human-driven Vehicles

Summary Information

Cooperative Adaptive Cruise Control (CACC) which enables vehicular platoons to travel in a closely coupled formation is expected to drastically increase mobility and provide for safer and more convenient travel. However, most analyses of CACC focus on the impacts of a completely connected environment and do not analyze the impact of CAVs on a mixed-operation network. Researchers from the University of Delaware and the New Jersey Institute of Technology sought to understand the effect of CACC on human-operated vehicles (HVs) using a VISSIM model with varying CAV market penetration rates (MPR). Researchers also examined whether the results were different between ad hoc coordination, where CAVs simply cluster with other CAVs encountered on the road, and local coordination, where CAVs actively seek out other CAVs to form platoons. The simulation emulated a 5-mile-long stretch of I-66 near the Washington, DC beltway. In order to simulate human behavior, a Wiedemann model was used, which was calibrated based off of the results of an earlier study on human reactions to CAVs performed in 2016 by Leidos.

Lessons Learned

Findings



Overall, the analysis found that local coordination had a significantly higher positive impact on throughput and productivity than ad hoc coordination. Both approaches resulted in roughly linear increases in throughput as the MPR increased.



The analysis also looked at hard braking observations in which HVs braked in response to a CAV. The study found that when the ad hoc approach was used, hard braking events occurred at an effectively constant rate for MPRs between 10 and 40 percent. When the coordination approach was used, braking events were found to increase with higher MPR, peaking at 30 percent market penetration. While the number of hard braking events was similar between the two coordination approaches at 30 and 40 percent MPR, the coordination approach resulted in dramatically fewer braking events at 10 and 20 percent MPR. A statistical analysis found that the two distributions were significantly different overall.



Finally, the researchers also looked at lane change activity, which was found to increase with higher MPR. Because higher market penetration by CAVs necessarily means there are fewer HVs on the road, the researchers used an average lane change frequency rather than the overall count of lane changes. For ad hoc coordination the average lane change rate increased linearly between 10 and 30 percent MPR, and increased more slowly between 30 and 40 percent MPR. For local coordination, the average lane change rate peaked at 30 percent MPR, and slightly decreased between 30 and 40 percent MPR. However, local coordination was found to cause a higher average lane change frequency than ad hoc coordination at low MPRs. At 30 percent MPR, local coordination began to be more effective.

Goal Areas

Keywords Taxonomy: