Zhao Yuqing, Ito Daisuke, Mizuno Koji
a Graduate School of Engineering , Nagoya University , Nagoya , Japan.
b Department of Mechanical Science and Engineering , Nagoya University , Nagoya , Japan.
Traffic Inj Prev. 2019;20(1):100-106. doi: 10.1080/15389588.2018.1533247. Epub 2019 Mar 1.
Though autonomous emergency braking (AEB) systems for car-to-cyclist collisions have been under development, an estimate of the benefit of AEB systems based on an analysis of accident data is needed for further enhancing their development. Compared to the data available from in-depth accident data files, data provided by drive recorders can be used to reconstruct car-to-cyclist collisions with greater accuracy because the position of cyclists can be observed from the videos. In this study, using data from drive recorders, the performance and limitations of AEB systems were investigated.
Data of drive recorders involving taxi-to-cyclist collisions were collected. Using the images collected from the drive recorders of those taxis, 40 cases of 90° car-to-cyclist intersection collisions were reconstructed using PC-Crash. Then, the collisions were reconstructed again utilizing car models with AEB systems installed while changing the sensor's field of view (FOV) and the delay time of initiating vehicle deceleration.
The angle of FOV has a significant influence on avoiding car-to-cyclist collisions. Using a 50° FOV with a braking delay time of 0.5 s resulted in avoiding 6 collisions, and using a 90° FOV resulted in avoiding an additional 14 collisions. Even when installing an ideal AEB system providing 360° FOV and no delay time for braking, 8 collisions were not avoided, though the impact velocities were reduced for all of these remaining collisions. These collisions were caused by the cyclists' sudden appearance in front of cars, and the time-to-collision (TTC) when the cyclists appeared was less than 0.9 s.
The AEB systems were effective for mitigating collisions that occurred due to driver perception delay. Because cyclists have a traveling velocity, a wide-angle FOV is effective for reduction of car-to-cyclist intersection collisions. The reduction of delay time in braking can reduce the number of collisions that are close to the braking performance limit. The collisions that remained even with an ideal AEB system in the PC-Crash simulation indicate that such collisions could still occur for autonomous cars if the traffic environment does not change.
尽管用于汽车与自行车碰撞的自动紧急制动（AEB）系统一直在研发中，但为了进一步推动其发展，需要基于事故数据分析对AEB系统的效益进行评估。与从深度事故数据文件中获取的数据相比，行车记录仪提供的数据可用于更精确地重建汽车与自行车的碰撞，因为可以从视频中观察到自行车的位置。在本研究中，利用行车记录仪的数据，对AEB系统的性能和局限性进行了调查。
收集了涉及出租车与自行车碰撞的行车记录仪数据。使用从那些出租车的行车记录仪收集的图像，利用PC-Crash重建了40起90°汽车与自行车在十字路口的碰撞案例。然后，在改变传感器的视野（FOV）和启动车辆减速的延迟时间的同时，使用安装了AEB系统的汽车模型再次重建碰撞。
视野角度对避免汽车与自行车碰撞有显著影响。使用50°视野且制动延迟时间为0.5秒可避免6起碰撞，使用90°视野可额外避免14起碰撞。即使安装了提供360°视野且无制动延迟时间的理想AEB系统，仍有8起碰撞未避免，尽管所有这些剩余碰撞的撞击速度都有所降低。这些碰撞是由骑自行车的人突然出现在汽车前方引起的，且骑自行车的人出现时的碰撞时间（TTC）小于0.9秒。
AEB系统对于减轻因驾驶员感知延迟而发生的碰撞是有效的。由于骑自行车的人有行驶速度，广角视野对于减少汽车与自行车在十字路口的碰撞是有效的。制动延迟时间的减少可以减少接近制动性能极限的碰撞数量。在PC-Crash模拟中即使使用理想的AEB系统仍存在的碰撞表明，如果交通环境不变，自动驾驶汽车仍可能发生此类碰撞。
