Performance of AEB systems in preventing car-pedestrian collisions

This study quantifies the performance of autonomous emergency braking (AEB) systems in preventing car-pedestrian collisions in Japan.

The study uses data on Japanese traffic accidents compiled by Japan's National Police Agency, restricting the analysis to collisions for which we could determine whether the primary party's car was equipped with an AEB system. Poisson mixed-effects regression analyses are conducted using data for 2022 and 2023 to quantify the collision avoidance performance of cars that were first registered in 2021 and equipped with AEB systems compared with cars without AEB systems that were first registered in 2015 or 2016.

Our analysis of collisions of all injury levels reveals that AEB-equipped cars that were first registered in 2021 had 12.3% (95% confidence interval [CI] 6.2%-18.0%) fewer accidents than non-AEB-equipped cars that were first registered in 2015 or 2016. The reduction for fatal and serious injury collisions alone was higher, at 15.6% (95% CI 2.2%-27.2%). A day/night analysis demonstrates a 7.6% (95% CI -0.9%-15.3%; ns) reduction in daytime collisions and a 19.0% (95% CI 9.9%-27.1%) reduction in nighttime collisions. Analysis of three car-pedestrian collision scenarios reveals a 12.1% (95% CI -0.3%-23.0%; ns) reduction in collisions of all injury levels between forward-moving cars and street-crossing pedestrians, an 9.7% (95% CI -2.2%-20.2%; ns) reduction in collisions between right-turning cars and street-crossing pedestrians, and a 21.8% (95% CI 4.8%-35.9%) reduction in collisions between forward-moving cars and pedestrians walking in the same or opposite direction.

The results demonstrate that AEB systems are effective in reducing car-pedestrian collisions; however, performance remains low compared with the effectiveness of AEB in reducing rear-end collisions. This indicates that considerable room remains for technological improvement.