Simulation Study Found 82 Percent Reduction in Predicted Number of Crashes with Automated Emergency Braking, Lane Departure Warning, and Automated Driving Systems.
Automated Driving Systems Simulated Using Real-World Data in Tsukuba, Japan, Predicted Safety Benefits.
Tsukuba
Multi-agent traffic simulations to estimate the impact of automated technologies on safety
Summary Information
Automated driving systems offer great potential for improving safety for transportation system users. The objective of this study was to develop a multi-agent traffic simulation methodology to estimate the potential road safety improvements of automated vehicle technologies. The multi-agent traffic simulation program developed in this study combined road infrastructure data with a large number of flow entity data including vehicles, drivers, and pedestrians. Human errors were induced by modeling inattention, aimless driving, insufficient safety confirmation, misjudgment, and inadequate operation. The program was applied to simulate traffic in a designated area in the city of Tsukuba, Japan, using real-world data from 2012-2017 period.
METHODOLOGY
In the simulation, a 100 percent manual driving scenario was initially set to simulate traffic for a total vehicle travel distance of 80,000 kilometers (about 50,000 miles). The crashes from this simulation were compared with real-world crash data from the study area. Additional scenarios were then implemented to estimate safety impacts, featuring increasing levels of automation with Automated Emergency Braking (AEB), Lane Departure Warning (LDW), and Highly Automated Driving (HAD) corresponding to SAE Level 4 (high driving automation). The crash types simulated included rear-end Vehicle-to-Vehicle, head-on Vehicle-to-Vehicle, intersection Vehicle-to-Vehicle, on-road Vehicle-to-Pedestrian, and single-car lane departure crashes. Specifically, the following scenarios were considered in the simulation:
- 100 percent manual driving
- 50 percent manual, 50 percent AEB
- 25 percent manual, 25 percent AEB, 50 percent AEB+LDW
- 25 percent manual, 50 percent AEB+LDW, 25 percent HAD
- 25 percent AEB+LDW, 75 percent HAD
FINDINGS
- The analysis results showed that, under Scenario 1, the system simulated a total of 859 crashes including single-car lane departure, car-to-car, and car-to-pedestrian crashes, compared to 156 predicted crashes under Scenario 5. This corresponded to an 82 percent reduction in crashes with the use of automated driving systems.
- In general, the results revealed an increasing trend in the crash avoidance improvement corresponding to increased level of automation penetration.
- The results also indicated that, both the average crash speeds and the relative crash speeds decreased with increased automation level, corresponding to a 28 percent (29.4 km/h in Scenario 1 to 21.3 km/h in Scenario 5) and 61 percent (17.2 km/h in Scenario 1 to 6.8 km/h in Scenario 5) reduction, respectively.
- This study also estimated a 72 percent reduction in the number of fatalities, by comparing the estimated number of 18 for Scenario 1, to the estimated number of five for the highest automation level simulated (Scenario 5).
