Google Car总监：别妄想100%有效的汽车安全技术！

你可能怀疑过：我们到底是否真的需要自动驾驶车辆？汽车厂商一直把“安全性”做为开发自动驾驶车辆技术的诉求，但问题是，我们该如何预期 Google Car 或是其它自动驾驶车辆达到降低交通事故率、保护人身安全的目标？ 那些号称为自动驾驶车辆应用所设计的芯片、软件与子系统供货商，偏爱用一种类似的开场白来介绍他们的解决方案：也就是先提出美国与全世界每年有多少人不该因为交通事故失去宝贵生命；在日前于东京举行的智能运输系统世界大会(ITS World Congress)上发表演说的Google自动驾驶车辆安全技术总监Ron Medford也不例外。

Gooogle自动驾驶车辆安全技术总监Ron Medford
DW5esmc

因为Medford原本担任美国国家高速公路交通安全局(National Highway Traffic Safety Administration，NHTSA)副局长，他所提出的数据看来更有说服力，包括：

2011年美国交通事故件数
Source：NHTSADW5esmc

2011年美国交通事故发生原因统计
Source：NHTSADW5esmc

本文授权编译自EE Times，版权所有，谢绝转载 第2页：世界上最佳的汽车安全技术就是安全带 第3页：自动驾驶车发生一起事故，将会使该技术的发展延迟十年

相关阅读：
• 为什么电动车反而更安全？
• 每日一报8月22日：特斯拉汽车安全评分“爆表”
• 汽车安全应用将迈入智能化和交互化DW5esmc

{pagination} 不过Medford的演示文稿内容带来了一个令人惊讶的转折：在2011年，美国有3万2,367人因为交通事故而丧生，其中有高达54%比例是因为没有系安全带：

2011年美国交通事故死亡人数
Source：NHTSADW5esmc

他 在演讲中表示：“到目前为止，世界上最佳的汽车安全技术就是安全带。”但实际上，人们仍然会忘记(或是拒绝)系上安全带，以至于在交通事故中失去宝贵生命 ──这个消息所带来的启示是，就算是最佳汽车安全技术，也不可能拯救每个人的生命，安全技术的有效性，最后仍取决于人们如何使用它(或是不使用它)。 Medford 指出，在1975至2011年间，有29万2,471人因为安全带而保住生命；但同时仍有45%在交通事故中丧生的小轿车前座乘客，是系了安全带的。儿童 安全座椅的使用情况也类似，根据统计，虽然在1975至2011年间有9,875条生命因为儿童安全座椅而获救，但同时也有71%的婴儿与54%的幼童， 就算使用了安全座椅仍然在交通事故中丧生。 本文授权编译自EE Times，版权所有，谢绝转载 第3页：自动驾驶车发生一起事故，将会使该技术的发展延迟十年

相关阅读：
• 为什么电动车反而更安全？
• 每日一报8月22日：特斯拉汽车安全评分“爆表”
• 汽车安全应用将迈入智能化和交互化DW5esmc

{pagination} 类似情况是，虽然自2011年起，所有的车辆都要求配备电子车身稳定控制系统(Electronic Stability Control，ESC)，该技术有效降低了交通事故死亡率，但根据统计，在2011年，配备了ESC的车辆乘客在交通事故中死亡率仍达到49%。 简而言之，Medford的结论是，就算产业界对于自动驾驶车辆安全性抱持高度希望，但预期100%的效果是不切实际的；毕竟每一种现有汽车安全技术虽然都有助于降低事故死亡率，还是无法保证百分之百的效果。 “我曾听过有人说，若是自动驾驶车辆发生一起事故，将会使该技术的发展延迟十年。”但Medford认为，这种对新技术并不公平、不切实际的期望只会带来伤害，而非有益于自动驾驶车辆技术的发展与演进。 所以我们现在被打过预防针了：像是Google Car那样的自动驾驶车辆不可能带来道路上的零交通事故死亡率──至少现在不可能、短期内也不可能，除非人们能绝对不坐在车子里！ 本文授权编译自EE Times，版权所有，谢绝转载 编译：Judith Cheng 参考英文原文：Google Car Director Tamps Down Safety Expectations，by Junko Yoshida

相关阅读：
• 为什么电动车反而更安全？
• 每日一报8月22日：特斯拉汽车安全评分“爆表”
• 汽车安全应用将迈入智能化和交互化DW5esmc

{pagination} Google Car Director Tamps Down Safety Expectations Junko Yoshida, Chief International Correspondent TOKYO — Have you ever wondered: Who, really, needs a self-driving car? "Safety," of course, is the big pitch, and the strongest argument, the automotive industry has trumpeted in its case for autonomous cars. The question now is how truly effective we expect Google cars, or any other self-driving cars, to be -- in terms of saving people's lives. Marketing presentations for chips, software, and subsystems -- all supposedly designed to enable the building blocks of autonomous cars -- tend to start with a similar setup: a litany of depressing numbers illustrating how many people are needlessly killed every year in traffic accidents in the United States and worldwide. So, there were no surprises there, when Google's director of safety for self-driving cars, Ron Medford, started his speech at ITS World Congress in Tokyo last week by running through a set of depressing slides repeating the familiar premise. Ron Medford, Google's director of safety for self-driving cars, at ITS World Congress Tokyo Considering Medford's previous position (he was the former National Highway Traffic Safety Administration's deputy director until he was recruited by Google earlier this year), the figures he rattled off, based on 2011 numbers in the United States, were both scary and convincing: Traffic accidents in the United States in 2011 (Source: NHTSA) Causes for traffic accidents in the United States in 2011 (Source: NHTSA) But his presentation made a surprising turn when Medford showed the following slide. Death tolls in traffic accidents in the United States in 2011 (Source: NHTSA) Of 32,367 people who died intraffic accidents in the United States in 2011, 54 percent of them were not wearing seat belts. As he put it in his speech, "the best automotive safety technology ever invented" is the seat belt. And yet, in reality, people are still dying in droves simply because they forget (or refuse) to buckle up. Medford's message was clear. Even the best automotive safety technology can't possibly save every life. The limit to the effectiveness of safety technology is the way people use it (or don't use it). Between 1975 and 2011, Medford said, 292,471 lives were saved by seat belts. And yet, during the same period, 45 percent of people killed in front seats in passenger cars were wearing seat belts. The same applies to child seats, according to Medford. Despite 9,874 lives saved between 1975 and 2011 because of child seats, 71 percent of infants and 54 percent of toddlers killed in accidents died in their child seats. Thanks to Electronic Stability Control (ESC) -- required in all vehicles by 2011 -- fatality rates have been dropping. Still, in 2011, death tolls in ESC-equipped cars were 49 percent in single-vehicle accidents. In short, sure, the safety expectation for self-driving cars is very high. But expecting it to be 100 percent effective isn't really realistic, concluded Medford. After all, every automotive safety measure has helped to reduce fatality rates, but none has proven to be 100 percent effective. "I've heard people saying that one accident in a self-driving car will set back the technology for 10 years," said Medford. But in his opinion, that sort of unfair and impractical expectation can only harm, rather than advance, the development of technologies for autonomous cars. So, we have been forewarned. Google cars will not result in zero fatalities on the road -- not now, not soon, not until you can get people out of cars entirely.