Intersections where drivers rapidly decelerate to obey the traffic signals can lead to increased fuel use and travel time. Connected and Automated Vehicle (CAV) technologies offer an opportunity to resolve this problem. This study estimated the potential impact of CAVs through data collection, model development, algorithm designs, simulations, and limited field tests. The three main technologies studied in this project included eco-routing, shared autonomous rides, and adaptive traffic signal controls.
Energy consumption and trajectory data were collected from 500 vehicles over one year (2017-2018), with a total mileage of eight million. In addition, factors impacting drivers’ decision making in choosing routes were examined in a controlled field environment at the University of Michigan's MCity testing facility that is equipped with Dedicated Short-Range Communication (DSRC) Road Side Units (RSU).
Thirty-two participants observed driver information display hardware to collect data on driving speed, throttle position, and brake pedal use and experienced seven different intersection scenarios for red, yellow, and green signalization. Next, an Eco-Routing turn-by-turn navigation app that provided three different routes to the driver varying the time and the fuel consumption on the routes (eco vs fast vs “balanced”) was developed and tested by 43 participants. A route choice survey was conducted offering drivers selections from recommended driving routes prior to beginning their trips (738 valid trips). Another survey was distributed to characterize the possible change of travel behaviors introduced by the CAVs. In total, 396 responses were collected.
A microscopic traffic simulation model of Ann Arbor was developed along with an adaptive traffic signal control algorithm to investigate the energy savings and mobility benefits from CAV penetration rates ranging from 3 - 12 percent+. The model evaluated the travel behavior to determine the number of shared automated vehicles that would replace individually owned vehicles. In each iteration of the simulation, the commuters chose their departure time and travel modes based on the observed traffic condition in the current iteration. These traffic conditions were updated based on the traveler’s choices and finally equilibrium was reached. In addition, the model also used eco-routing algorithms to determine fuel saving potential. Vehicle‐actuated control was applied in the simulation as the benchmark for comparison with the CAV‐based control. Average values of throughput, vehicle delays, and CO2 emissions were recorded with various levels of traffic demand (using the product of the basic demand and a demand factor). Both under‐saturated and over‐saturated conditions were investigated.
- Microsimulation model results showed that eco-routing in Ann Arbor can lead to six percent fuel saving.
- Agent-based model results revealed that if different families share a fleet of vehicles, the total number of vehicles needed could be reduced by 75 percent, which means one shared automated vehicle can replace four individually owned vehicles. However, the vehicle miles travelled would increase by 25.6 percent (due to the connections between different trips).
- The CAVs' adaptive traffic signal control algorithm was found to reduce delay by an average of 13 percent and reduce fuel consumption by 10 percent. CAV-based control also improved intersection capacity, although the throughput increase is insignificant with under-saturated traffic demand.
- Human drivers were found to follow the Eco-driving suggestions roughly 70 percent of the time. Results of the analysis showed that drivers were more likely to select the eco route when its distance was shorter and gas consumption per mile was higher.
+: The study focused primarily on optimizing traffic signals with low penetration rates of CVs (< 10 percent), since this is what will make an immediate impact on the state‐of‐ the‐ practice.