几年前，全球领先人体模拟测试设备（ATD，俗称碰撞测试假人）供应商 Humanetics Innovative Solutions 首席执行官 ChrisO’Connor 开始注意到，任何自动驾驶会议上似乎总少不了一个共同的话题。

“似乎很多业内人士和国会议员都在说‘有了自动驾驶汽车，我们就不会再遇到交通事故了’，”O’Connor回忆道，“既然不会再有车祸，那我们也不再需要担心车上人员的安全了。”

当时，自动驾驶汽车革命刚刚崭露头角。“自动驾驶汽车可以避免所有的交通事故” — 这种诱人的宣传，让汽车行业领导者和政策制定者难以拒绝。

“后来我们都意识到，这种说法是很荒谬的，”O’Connor 表示，“自动驾驶汽车也会发生车祸，它们的电子系统可能出现故障，软件也可能会有问题，这些都是我们必须面对的现实。”

O’Connor 并不是自动驾驶汽车的反对者，也从不曾否认自动驾驶汽车在显著减少道路事故、伤害和死亡人数方面的潜力。事实上，O’Connor 是一名货真价实的自动驾驶汽车倡导者，而且他对汽车碰撞测试领域及相关分析工具的深入了解和丰富经验，非常有助于行业解决围绕自动驾驶汽车乘客安全的相关挑战。在接受《汽车工程》采访时，O’Connor 指出，如今汽车行业许多项目的情景规划都基于每年3.7万名车祸死亡人数，而仅在美国每年就会发生 600 万起各类交通事故。

“随着自动驾驶汽车在道路中的比例越来越高，我们必须始终注意乘客安全的问题，”O’Connor说，“我们都见过一些创新自动驾驶汽车的座椅排布，特别是在一些车展上的概念车中。这些创新座椅的共同特点是乘客都是面对面排列的。很显然，这是因为没人需要再驾驶汽车或关注道路情况了。而这，就是第一大挑战。”

展望未来

事实上，O’Connor 提到的座椅设计并不新鲜。自 20 世纪 50 年代的“喷气式飞机时代”和 60 年代的“太空时代”以来，这种座椅就已经为公众所熟知。比如，通用汽车公司曾展示过Motorama 和 Futurama 概念车，可以沿着部署在高速公路下的电磁线路进行无人驾驶，而车上的乘客则在无忧无虑地打牌、吃饭或聊天。这样的场景已经在公众心中留下了不可磨灭的印象。

但 O’Connor 也同时指出，要想让自动驾驶汽车真正实现无障碍和无事故的行驶，我们还需要多年的工程、科学和安全系统测试。与此同时，还必须解决许多技术障碍。

举个例子，自动驾驶穿梭巴士或 SAE 4 级自动驾驶汽车（乘客面对面乘坐）是否将给车辆的安全测试、测试标准和更专业的 ATD 开发带来新的挑战？

O’Connor 进一步解释说，“假设你坐在一个靠背保持在 23 度的汽车座椅上，这也是现有测试过程中的标准座椅靠背角度。这时，你希望将座椅靠背继续下推至 33 度或 43 度吗？这种调整可能并不算太大，但事实上，每增加一次角度，车上乘客受伤的概率都将急剧上升。”

更夸张的调整还包括直接改变座椅的朝向和角度，或者让座椅向侧面倾斜，而这种座椅在想象中的自动驾驶的世界里是非常普遍的。

O’Connor 提出，在不远的将来可能将出现一种全新的场景。想象一下，你正乘坐一辆自动穿梭巴士前往机场，你的座位靠近车辆中部，朝向与车辆的行驶方向相反，就像乘火车一样。这种情况下，一旦发生事故，你的头将直接暴露在正面碰撞区。与此同时，与你一起乘坐巴士的另一名乘客则将自己的座椅调整至面向车窗的角度，希望能更好地欣赏沿途的风景。很显然，这位乘客的情况也可能非常危险。

在此背景下，车辆的整个结构都必须重新设计，以应对最严重的车祸冲击。

“这不是一件简单的事情，”O’Connor 继续说道，“举个例子，自动出租车制造商 Zoox 推出的产品共有四个座椅，一排两个，两两相对，汽车可以向前或向后行驶。在这种座椅排布下，单单是调整座椅靠背角度，都可能会显著增加人员在事故中受伤的概率。

“更重要的是，在自动驾驶行程中，乘客更容易放下座椅靠背，因为他们完全不必操心路况或驾驶，完全可以躺着看电影或睡一会。”O’Connor 表示，“我们必须好好研究这些车辆概念，从而真正保证车辆的整体安全性，这已经刻不容缓了，我们必须在早期设计阶段参与进来。”

重新思考碰撞测试

Humanetics 公司的总部位于密歇根州法明顿山。公司首席执行官 O’Connor 表示，Humanetics每年可以生产大约 250 部 ATD 设备。事实上，ATD设备的成本非常高昂，每个型号都必须经过 10 到 15 年的开发，然后还需要通过 NHTSA、NCAP 及各种政府机构的漫长审批。根据具体设备不同，一款 ATD 设备的成本最高甚至可以达到 80万美元。目前，Humanetics 公司的主要市场在美国以外地区，占公司总销售额的 75%。

O’Connor 的工程师一直与自动驾驶汽车OEM保持着非常紧密的合作，积极针对这些异于传统的座椅配置，打造专门的测试流程。“我们对此非常主动。事实上，为了实现这个目标，我们拿出了公司最新最先进的THOR 系列测试假人，并对其进行修改，使其更加适合自动驾驶汽车测试。”O’Connor表示，“这点非常重要，因为在新型座椅排列下，传统安全带和安全气囊能发挥的作用都相对有限。”

O’Connor 表示，“所有原厂都明白，未来他们将不得不调整，甚至完全重新设计这些安全系统。因此，他们也需要合适的测试设备完成评估工作。”

O’Connor 预测，新的 THOR系列将用于下一代 NCAP 新车评估循环测试，及 NHTSA 的最新倾斜测试循环。在最新的测试流程下，被测车辆必须在静止状态下，接受时速 56 英里（90公里/小时）滑车的侧面撞击，撞击面积对为正面碰撞的35%，碰撞角度则互为 15 度。在这个场景下，车上乘客很容易从安全带等保护设备中甩出。很显然，这种测试可以为未来的自动驾驶汽车结构、客舱和安全系统设计提供非常有价值的信息。

目前，随着汽车原厂纷纷投入自动驾驶汽车的发展浪潮，O’Connor 及其团队也开始为欧盟的 2020 年 NCAP 循环测试进行准备。据了解，新的测试循环流程全面修改了车辆的安全评级标准，增加了很多新的测试与评估项目，并且修改了与当前测试相关的评分条件，与NHTSA 提议的 NCAP 循环测试项目比较类似。

“过去，美国在汽车安全方面一直处于全球领先地位。然而，在过去 10 年里，由于其他市场在行人保护及其他关键测试方面的要求更加严格，我们已经不再具备优势了。”O’Connor 直言道，“最近的美国 NCAP 测试循环议案将是我们迎头赶上，甚至重新占据领先优势的机会。”

正是出于这个原因，目前，Humanetics 牵头了一个私企联盟，专门解决一些自动驾驶汽车独有的安全问题，例如自动驾驶汽车特殊的座椅排列可能带来的安全隐患。显而易见，无论是 Humanetics，还是联盟中的其他成员公司，私营企业已经开始主动解决一些传统意义上应该由政府出面解决的问题。

“我们必须立刻行动，”O’Connor 表示，“我们是虚拟测试的大力支持者，而且一直在向原始设备制造商提供各种虚拟计算机模型，帮助他们在真正启动真实碰撞测试前，进行充分的模拟测试。然而，我们也明白，只有通过实车测试，才能确保测试结果准确无误，且能达到预期的结果。”

THOR 测试假人家族

目前，Humanetics公司的 THOR-50M 系列是业内最先进的正向碰撞 ATD 设备。THOR 是 Test device for HumanOccupant Restraints 的首字母缩略，即人类乘客约束系统测试装置；后缀 50M 中的M 表示男性（Male）、50 则表示中位数水平。目前，THOR 系列中的产品划分越来越精密。这些假人经过专门设计，可以承受严重撞击并提供重要数据。目前，THOR 家族已经取代了上世纪 80 年代同样由 Humanetics 公司开发的 Hybrid-3 系列假人。

公司首席执行官O’Connor 介绍说，尽管最新版本的 Hybrid-3 假人已经装配了高达 20 个传感器，“但以今天的技术来看，它已经是老古董了。”

THOR系列产品的设计则更加人性化，可以提供更高的生物逼真度、动态响应性、可重复性和耐用性，内部数据采集系统能够提供高达 150 条信息渠道，专门用于评估事故伤害。

Humanetics公司 THOR 家族中的最新成员，是一款经过专门设计的女性假人（如图所示），可以更加精确地评估从生理方面更易因事故受伤甚至死亡的女性乘客的安全。目前，除了这款女性假人，Humanetics 公司THOR 家族中的“定制”ATD 产品还包括“老年”假人和“肥胖”假人等。事实上，此类人群在车祸中的表现都与中位数男性有很大不同，因此这些“定制”假人具有非常重要的意义。

“目前，‘老年’假人的设计主要是基于一名 70 岁女性的数据，”O’Connor解释道，“‘她’看起来有点胖，这在老年人中非常常见，可以为车内约束系统的安全设计提供非常重要的数据。”

O’Connor表示，数据显示，老年人在车祸中更易于死于肝脏或脾脏破裂。此外，老年人在车祸中受伤和死亡的概率更高，而且通常不会在事故中当场死亡，而是更容易在几天后死于内部器官破裂。

O’Connor指出，“我们不仅调整了‘老年’假人的腹部设计，甚至定制了内部器官。”O’Connor 表示，“这将有助于确定车辆安全约束装置或汽车构件是否会导致老年人的器官损伤，甚至是最终死亡。”

A few years ago, Chris O’Connor started to notice a common talking point emerging from within the many autonomous vehicle conferences and meetings he was attending.

“It seemed that many people in the industry, and those in Congress, were saying: ‘With autonomous cars, we won’t have traffic accidents anymore,’” noted O’Connor, the CEO of Humanetics Innovative Solutions, the leading developer of anthropomorphic test devices (ATDs), commonly known as crash-test dummies. “And since there won’t be crashes, we will no longer have to worry about occupant safety,” he recalled many pronouncing.

At the time, AVs were just emerging as The Next Revolution in mobility. Industry leaders and policymakers alike were being seduced by the growing hype—that self-driving vehicles will eliminate virtually all road accidents.

“Since then, of course, we’ve all come to realize such statements are ridiculous,” O’Connor said. “There will be crashes involving autonomous vehicles. They will have electronic systems that will fail, and software problems. These are realities we’ll have to face.”

Connor is not a critic of AV technology and its potential to significantly reduce road crashes, injuries and fatalities. In fact, he’s a vocal advocate whose deep knowledge of and experience in the vehicle crash-testing sector and the analysis tools used in the industry is helping to bring the emerging challenges around AV occupant safety into practical focus. Speaking with Automotive Engineering, he is quick to point out that much of the industry’s scenario planning is being done based on an occurrence of approximately 37,000 annual fatalities—while six million accidents of all types occur every year in the U.S. alone.

“We cannot stop thinking about occupant safety as autonomous vehicles become more common on our roads,” O’Connor asserted. “We’ve all seen some of the seating layouts for AVs, particularly within the concept cars at shows. They feature passengers sitting facing one another, because no one needs to drive or focus attention on the road. That’s challenge number one.”

Future visions

The passenger compartment to which he refers is not new. It’s been capturing the public’s imagination since the “jet age” 1950s and the “space age” 1960s, when General Motors showed its Motorama and Futurama concept cars gliding driverless along electromagnetic wires embedded in the highway. The occupants of these futuristic cars sat playing cards, dining, or chatting. Such images became indelible in the public’s mind.

But the truly hassle-free and nearly accident-free reality will require years’ more engineering, science, and safety systems testing, O’Connor notes. In the meantime, many technical hurdles must be solved.

What will an autonomous shuttle, for example, or SAE Level 4 vehicle with their occupants riding facing each other, present to vehicle safety testing, the standards behind the testing, and the development of even more specialized ATDs?

“Let’s say you’re sitting in a car seat that has its backrest position set at 23 degrees. There is currently a standard test for that seating position. Recline the seat to 33 degrees or 43 degrees; this is not an extreme difference, yet your rate of injury could go up dramatically under each of these circumstances,” he said.

Then spin the seat around so it’s facing in the opposite direction. Or angle it sideways. These positions are being portrayed as common and acceptable in the new world of autonomous driving. 

O’Connor suggests another likely scenario in the not-too-distant future: traveling in an autonomous shuttle heading to the airport. You’re seated in a middle row seat that’s facing the rear of the vehicle, like riding in a train. In this position, your head is directly facing the frontal crash zone. At the same time, one of your co-riders sharing the shuttle has rotated her seat, as it was designed to do, around to the side so she can watch the scenery pass by. 

With such a vehicle, the entire structure will have to be re-envisioned completely—to cope with worst-case impact scenarios.

“It’s not a simple thing,” O’Connor continued. “Look at the [robot taxi maker] Zoox model. Their car has a quad seating configuration where the two rows of seats are facing each other. The car can travel forward or backward, it doesn’t matter. Reclining the seat alone, in such a configuration, has the potential to be more dangerous. 

“You’re more apt to recline the seat in an AV because you don’t have to pay attention to the road; you’re watching a movie or want to take a nap,” he noted. “These vehicle concepts have to be studied, and now is the time to consider the overall safety, in the early design phase.” 

Rethinking crash testing

Based in Farmington Hills, Mich., Humanetics manufactures about 250 ATDs per year, O’Connor said. Each model can take 10-15 years of development, followed by extensive approval processes at NHTSA, NCAP, and various government agencies, and can cost up to $800,000 each, depending on equipment. About 75% of the company’s sales are outside the U.S.

O’Connor’s engineers have been working with the OEMs to create test procedures for the unconventional, often out-of-position seating configurations of AVs. “We’re being very proactive about this. We’re taking our latest and most advanced THOR dummy [see sidebar] and modifying it for autonomous vehicle testing,” he said. “Because in these altered seating configurations, the traditional safety belts and airbag restraints will not work effectively. 

“The OEMs all know they are going to have to modify and rethink these systems and they need a proper test device to evaluate them,” he said.

O’Connor predicts that the new THOR series will be used in the next-generation NCAP (New Car Assessment Program), as well as in NHTSA’s new oblique testing protocol. Under these new guidelines, a stationary vehicle will be struck by a crash sled at 56 mph (90 km/h) with a 35% overlap, and the test vehicles positioned at a 15-degree angle relative to each other. In this scenario, vehicle occupants are easily tossed out of position from the restraint system—offering valuable insights for the design of future AV structures, cabin, and safety systems.

With AV development moving rapidly at a growing list of vehicle OEMs, O’Connor and his team are also working on developments to prepare for European NCAP’s 2020 program. The new protocols include a comprehensive revamp of safety ratings, with numerous new tests and assessments added, and the scores associated with the current tests modified. Euro NCAP 2020 is similar to NHTSA’s proposed NCAP program.

“In the past, the U.S. led the way in automotive safety. However, in the last 10 years we’ve lost that lead to other regions, as other markets continued to add pedestrian protection and other critical tests and higher standards,” O’Connor asserted. “The proposed new U.S. NCAP is the chance for the U.S. not only to catch up, but to move ahead.”

For this reason, Humanetics is currently leading a consortium of private enterprises to address safety issues specific to autonomous vehicles, such as non-traditional seating positions within these cars. This is notable for Humanetics and other consortium members, as it shows private industry is voluntarily taking a proactive lead in addressing issues which have traditionally been government’s domain. 

“We have to move at a high rate of speed,” O’Connor stated. “We’re a big proponent of virtual testing and we make all virtual-dummy computer models available to the OEMs, to help them run simulations in advance of actual crash tests. However, only by testing the physical product can we be sure the test is being done correctly and the expected results are achieved.” 

Meet the THORs

The industry’s newest state-of-the-art frontal crash ATD is Humanetics’ THOR-50M. The acronym stands for Test device for Human Occupant Restraints; the suffix indicates 50th percentile male. It’s one of an extensive family of increasingly sophisticated human replicants engineered to endure severe impacts and deliver vital data. THOR replaces the previous Hybrid-3 dummy, developed by Humanetics in the 1980s.

Although the latest iteration of Hybrid-3 has been updated with up to 20 sensors, “it’s really primitive compared with the technologies available today,” CEO Chris O’Connor explained. 

The primary design objectives for the more human-like THOR included greater biofidelity, dynamic response, repeatability and durability. Its internal data-acquisition systems, capable of delivering up to 150 channels of information, are specific to injury assessment. 

The latest Humanetics’ THOR model (shown) is a female version, designed to ensure better protection for women who are physiologically more prone to crash-related injuries and fatalities than men. The female THOR joins other specially tailored Humanetics ATDs, including models for elderly and obese people, who also experience automobile accidents differently than the 50th-percentile male.

“We made the elderly dummy proportionally correct based on data for a 70-year-old woman,” O’Connor explained. “‘She’ looks a little obese, but it’s exactly average. This is very important for restraint systems.”

He said data show more elderly fatalities result from a liver or spleen being ruptured. And not only are injury and fatality rates higher for the elderly, but the deaths typically don’t happen at the time of the accident but rather days later as a result of internal organ ruptures.

“We instrumented not only the abdomen [in the new elderly THOR] but also the organs,” O’Connor noted. “This is important for determining whether a restraint device or part of the car is causing an organ injury and ultimately the death of a senior person.” –L.B.

Author: Lindsay Brooke

Source: SAE Autonomous Vehicle Engineering