BRIAN SANTO: I’m Brian Santo, EE Times Editor in Chief, and you’re listening to EE Times on Air. This is your Briefing for the week ending February 21st.
BRIAN SANTO: 我是EE Times的主编Brian Santo，您正在收听ETimes On Air。以下是截至2月21日的本周播报。
In this episode.
We keep getting told that autonomous vehicles will save lives. Okay, yes. But so will driver-assist technology — and probably by the same margin. Generally speaking, the companies that favor the adoption of driver-assist technologies tend to be the established automotive companies. These are the companies that have been around for decades and decades, and have an enormous amount of experience with road safety.
The companies that tend to favor autonomous vehicles — AVs — are the high-tech companies that have sprung up in recent years. Generally speaking, high-tech companies tend to subscribe to the credo: “Move fast and break things” — a line made famous by Facebook founder Mark Zuckerberg.
EE Times has been covering technology for nearly 50 years. We can attest that moving fast and breaking things has been marvelously effective for companies that have developed web browsers and smartwatches and videogame systems and robot vacuum cleaners. We think it is a reckless, irresponsible and dangerous attitude to have, however, when it comes to building systems that have to be operated safely, and that most certainly includes cars.
We’re not the only ones. Semicast Research analyst Colin Barnden is considered one of the world’s leading experts in vehicle sensors — the fundamental technology that enables modern vehicles to sense their surroundings and avoid collisions.
我们不是唯一持有此观点的机构。Semicast Research的分析师Colin Barnden被视作全球车辆传感器的顶尖专家之一——车辆传感器是确保现代车辆能够感知周围环境，并避免碰撞的基本技术。
EE Times international editor Junko Yoshida called up Colin to talk to him about his latest articles for EE Times, in which he has clearly lost his patience with the false piety and the hubris from AV proponents.
EE Times国际编辑Junko Yoshida联线Colin，谈论他在EE Times最新发布的文章。Colin的文章表明，他已经对AV拥护者错误的虔诚和自负失去耐心。
JUNKO YOSHIDA: I would like to get everybody’s attention to what you have been writing about. I wouldn’t call it a “soap box,” but this is something you and I both agree, and I think there are a lot of things to chew on. So explain the premise. What prompted you to write this piece? Tell us the premise. Explain.
JUNKO YOSHIDA: 我想引发大家对你所写内容的关注。 我不是说你站在“肥皂盒”（俚语中意指将对话变成某人的演讲）上演讲，这是我们双方都同意的观点，我认为还有很多事情需要仔细思考。所以请解释一下，是什么促使你写下这篇文章的？给我们讲讲写作的原因，说明一下。
COLIN BARNDEN: I think it would be fair to say that I’ve listened now to I think a lifetime’s worth of podcasts in which I listen to people from the AV industry tell me that autonomous driving and AV is all about saving lives. It’s just sounding too wee work to me, you know, elevating the world’s consciousness, a bit of angelical, and I’m really just putting down some of my thoughts about an alternative point of view.
COLIN BARNDEN: 我想可以这样说，我到现在为止已经听了，我认为能受益一生的大量播客。在这些播客中，我听到来自AV行业的人士声称，自动驾驶和AV所做的就是挽救生命。对我来说，这听起来没甚用处，看起来好像提升了全世界的意识，带点天使视角的意味。实际上我只是在表达自己的一些不同看法。
JUNKO YOSHIDA: It is true. The last time you wrote the column, just stop saying you’re in it for saving people’s lives. Just say you’re in it for money, right?
JUNKO YOSHIDA: 确实如此。上一次你撰写专栏时声称，与其说你参与这个行业是在挽救人们的生命，不如说是在赚钱，对吧？
COLIN BARNDEN: Exactly that. What I’m looking at really, is this about saving lives, or is this about controlling lives? If I want to travel somewhere, if I want the choice of whether I take a human-driven taxi, a machine-driven taxi or if I want to drive myself. And I can see a future really with the tech industry. What it looks to me what they’re working towards is essentially removing the steering wheel from human hands so that we can be driven around by them. And I’m not really comfortable with that. As somebody who likes freedom, who likes the ability to drive where I want, when I want, how I want. And that’s really the concern that I’ve got here.
COLIN BARNDEN: 就是这样。我现在看到的这个行业，到底是真正在拯救生命，还是在控制生命？如果我想去某个地方旅行，或者想要选乘人工驾驶的出租车、机器驾驶的出租车或是选择自己开车，我可以真正从中看到科技行业的未来。 在我看来，实际上人们正在努力，将人们的双手从方向盘上解放出来，以便我们可以被“驾驶”。我对此不太满意。作为崇尚自由的人，我喜欢拥有将车子开往我想去的地方的能力，何时开、怎么开由我决定。这确实是我所担忧的问题。
JUNKO YOSHIDA: I just realized that I was going back, the history of Waymo, or since Google started this project, actually it’s been 10 years, right? And last year Waymo claimed that their autonomous vehicles are driven more than 10 million miles in the real world and more than 10 billion miles in simulation. I think that’s their claim. Actually, I take pleasure in looking at the advertisement video clip they put together. Because what they’re claiming inside, a software engineer comes out and talks about how the autonomous cars can not only do perception very well, but they can think ahead and predict what the next move is with the person who is standing right next to the road or something. I just can’t believe their audacity to say that this is all already true! Because this is not true! Right?
JUNKO YOSHIDA: 我刚刚意识到，我回想Waymo的历史，或是自Google启动此项目以来，实际上已经有10年了，对吗？去年，Waymo声称他们的AV在现实世界中行驶了超过1000万英里，在模拟空间中行驶了超过100亿英里。 我认为这只是他们声称的。 实际上，我很乐意观看他们的广告视频。因为在他们的声明中，一位软件工程师现身讨论了AV是如何做到，不仅可以很好地感知，而且可以提前思考，并预测路旁行人的下一步行动。我简直难以置信，他们居然胆大妄为地说这一切已经成真！ 因为这不是真的！对吗？
COLIN BARNDEN: The use of these vehicles looks to be incredibly limited to me. We’ve seen them in sort of Chandler / Gilbert geofence [Chandler and Gilbert are towns in Arizona used as a test area — ed.]. There’s some uses of this technology is San Francisco. There are different companies that are trialing in different areas. But I always look at this, and I would say to myself, You know, so what? I mean, congratulations, Waymo. So you’ve reinvented the human taxi driver. How exactly is this elevating the world’s consciousness? Or how is this going to go about saving lives?
COLIN BARNDEN: 这些车辆的用处对我来说似乎是非常有限的。我们已经在钱德勒/吉尔伯特限定区域中看到这些车辆（钱德勒和吉尔伯特是亚利桑那州的城镇，被用作试验区）。这些技术在旧金山有一些用途。有不同公司正在不同的地区进行测试。但是看到这些，我会对自己说：你知道的，这又怎么样呢？反正恭喜Waymo。你已经重新发明了一种出租车司机，这究竟是如何提升世界意识的？ 或者说这将如何挽救生命？
My entire premise is, let’s get driver assistance and monitoring technology into absolutely every vehicle with four wheels or more first, and then let the market decide if there is any validity or any advantage in self-driving technology whatsoever. And that’s essentially the premise. And really that’s what we’re seeing in Europe, is that position being driven very much from Euro end cap and the European Commission. And the United States seems to be on a completely different path.
JUNKO YOSHIDA: Speaking of path, actually you explain in your latest column, the left fork and the right fork. Explain what left fork is and what right fork is.
JUNKO YOSHIDA: 说到“道路”，实际上你在最新的专栏中进行了说明，即左岔道和右岔道。请解释一下什么是左岔道，什么是右岔道。
COLIN BARNDEN: Left fork, essentially, is all about the swashbuckling, flamboyant AV tech industry and its wonderful promises of a utopia of accident-free travel in which we have flawless machine drivers and everybody’s happy and nobody dies and there is no suffering. Eventually. And whenever is “eventually,” I don’t know, and I don’t think [Waymo CEO] John Krafcik knows, and I don’t think really anybody in the autonomous driving community knows.
COLIN BARNDEN: 从本质上讲，左岔道是拥有传奇历险情怀，蓬勃发展的AV技术行业，他们承诺了一个美好的乌托邦，在这里都是无事故旅行，我们拥有无懈可击的机器司机，每个人都很高兴，没人丧命，没人受伤。最终，我都不知道何时才到他们所称的“最终”，而且我也不认为John Krafcik（Waymo的 CEO）知道，我认为自动驾驶行业中没人真正知道。
And then I’m looking at the right fork, which is the boring, predictable, traditional automakers. They take their time and they think about what they’re doing. And they’re going down a totally different path now it looks to me about what you could call collaborative driving where the human driver and the machine monitor — the machine backup, if you’d like — they work together in synchronization, and the humans are responsible for reasoning and the machines are responsible for driver assistance and monitoring for fatigue in the human driver. And that looks to me like very achievable, very deliverable, very sensible. Which is precisely really what you would expect from the traditional automakers who’ve been around a long time and kind of know a thing or two about technology change and events happening in the world.
JUNKO YOSHIDA: I have a couple of follow-up questions on that. When I was first reading your column, I was thinking, Okay, here’s Colin talking about left fork and right fork. As Yogi Berra said, “When you come to a fork in the road, take it.” So isn’t it the reason why companies like Ford and GM have a separate company like Argo AI or the Cruise, it’s because they feel that this is an opportunity. They may not be seriously thinking this is going to make big money, but they really need to take it now so that they won’t be far behind. Is that the theory do you think?
JUNKO YOSHIDA: 我对此还有几个疑问。当我第一次阅读你的专栏文章时，我在想，好吧，这是Colin在谈论左岔道和右岔道。 正如Yogi Berra所说：“当你在路上遇到岔路时，请抓住机会。”这难道不是福特和通用汽车这样的公司拥有Argo AI或Cruise之类的独立公司的原因？因为他们都认为这是一个机会。他们可能没有认真考虑过这是否会赚大钱，但他们确实需要立即着手做这件事，这样他们才不会落后于其他对手。你认为这个理论成立吗？
COLIN BARNDEN: That’s a possibility. But essentially what it really looks to me like some of these companies are hedging their bets. A lot of the auto makers are backing separate companies to develop this technology. But really, if we look at where the advisory and the legislative bodies are pushing the technology now, it’s very much towards much more simple driver assistance and monitoring technologies, which really, they fit into Level Two. And where or not there’s really a future for Level Four, it might be that that’s true at some point in the future. But we saw it from Zed F at CES. They came out and they said that there’s really no requirement for consumer AV vehicles, privately owned autonomous vehicles. And essentially really what it comes down to is that the technology is just too expensive. Humans have got perfectly good vision systems, perfectly good reasoning systems already. Essentially what we need help with is distraction and fatigue and essentially driver assistance and driver monitoring. And that very much to me looks to be the direction that the technology’s moving in now.
COLIN BARNDEN: 那是一种可能性。但对我来说，从本质上来讲，这些公司中有些只是见风使舵。许多汽车制造商都在支持独立的公司来开发这项技术。但实际上，如果我们看看咨询和立法机构，他们现在正推动该技术的发展，那么该技术将朝着更加简单的驾驶辅助和监控技术的方向迈进，实际上，它们已被纳入第二级。不管第四级在未来有没有真正的立足之地，也许在将来某个时候确能实现。但我们在CES的Zed F上看到了这项技术。Zed F说，人们对AV消费级汽车、私有AV没有要求。从根本上说，最终结果是这项技术太昂贵了。人们已经拥有了非常好的视觉系统和推理系统。从本质上讲，我们需要的帮助是能够在驾驶时分心，得到帮助来缓解驾驶疲劳，以及获得驾驶辅助和对司机的状态监控。在我看来，这非常有可能成为该技术的发展方向。
JUNKO YOSHIDA: You actually mentioned that increasing the number of ADAS features that are now being required down the line by Euro NCAP, for example, like automatic emergency brake (AEB). I wrote a story recently based on some of the conversations that I had with different people, that AAA last year did the testing, and the results were actually astoundingly bad. And part of the reason was I think a lot of automakers are using not-so-expensive equipment and not so much redundancy in there. When I saw the video, I could not believe the ADAS car could not see the person right in front of it. What’s your take on that? Why is AEB so bad?
JUNKO YOSHIDA: 实际上你已经提到，欧洲NCAP现在增加了所需ADAS功能的数量，例如自动紧急制动（AEB）。最近我根据与不同人们的一些交谈写了一个故事，去年AAA进行了测试，实际上结果非常糟糕。部分原因是我认为许多汽车制造商使用的设备并不昂贵，也没有那么多冗余。当我看到这段视频时，我不敢相信ADAS汽车竟然看不到前面有人。你对此有什么看法？为什么AEB如此糟糕？
COLIN BARNDEN: There’s so much work still to be done. But essentially that’s what you can do with the 100, maybe 200 bucks budget, which the OEMs can put aside for that. So if anybody follows Oliver Cameron on Twitter, CEO of Voyage, he’s been talking a lot recently about a technology called SafeStop. And essentially it’s a Velodyne lidar. I think it’s an Nvidia GPU, and this is hundreds of dollars’ worth of technology. It’s just not cost-effective in a mass market production where you might be looking to do this for $150 and maybe within a six watt power budget. So you can do all sorts of wonderful things with very, very cutting-edge technology. But essentially what can you do in the power, price and performance budget of the OEMs? That’s the challenge.
COLIN BARNDEN: 还有很多工作要做。但是实质上，这就是你可以用100美元，或者200美元的预算做的事情，OEM可以为之留出资金。如果有人在Twitter上关注Voyage的CEO Oliver Cameron会发现，他最近一直在谈论一种称为SafeStop的技术。 这技术本质上是Velodyne公司的激光雷达，我认为其使用的是Nvidia 的GPU，这是价值数百美元的技术。 在大规模市场生产中，这可能不具成本效益，你可能要花150美元，而且可能在6瓦的电力预算之内进行生产。因此，你可以使用非常尖端的技术来完成各种奇妙的事情。 但从本质上讲，在OEM的功率、价格和性能预算方面，你能做些什么？这就是挑战。
An F-35 fighter plane is more maneuverable than an A380. But what does that actually tell us?
JUNKO YOSHIDA: Exactly. There’s another thing that related to ASAS as we are on this topic. You’ve seen a lot of comments being made by experts like Missy Cummings from Duke, talking about the difficulty of Level Three. In other words, more and more features inside the car are getting automated. People become complacent, and people will stop really monitoring the system, I mean monitoring the road ahead. How do you feel about this advancement of ADAS, actually, in this case?
JUNKO YOSHIDA: 确实如此。在这个话题上，还有另一件事与ASAS有关。 杜克大学的Missy Cummings等专家就三级考试的难度发表了很多评论。 换句话说，汽车内越来越多的功能正在实现自动化。人们变得自满，人们会停下真正的监控系统，我的意思是监控前方的道路。事实上，在这种情况下，你对ADAS的发展有何看法？
COLIN BARNDEN: When we’re at Level Three, what we’re talking about is the machine-to-human handover. So that’s the problem. So going from the human to the machine isn’t a problem. But it’s coming back from the machine to the human that is the trouble. And there’s something like a 45-second handover period. And if you think of a car traveling on the freeway at 55 miles an hour, how far it travels in 45 seconds or so, there’s been some work done at Leeds University here in the UK talking about this handover period.
COLIN BARNDEN: 当我们处于三级时，我们所说的是机器对人的交互。这就是问题所在。从人到机器都不是问题，问题是从机器到人。大概有45秒的交互时间。而且你想，如果一辆汽车在高速公路上以每小时55英里的速度行驶，那它在45秒左右的时间内行驶了多远，英国利兹大学已经做了一些工作来讨论这个交互时间。
So really what we’ve got here is Level Three, where essentially the OEM is liable for some of the time. This is essentially the problem. And really, this is why we’ve seen the arrival of what’s called Level Two Plus and various other names. Level Two Plus Plus, Level Three Minus. There’s all sorts of names for this. But essentially that’s what we’re seeing, is these highway assist systems where you’ve got really got quite advanced functionality of the system, but it is in the human driver that remains responsible at all times. And Super Cruise is really the most advanced system around this level, but with a driver monitoring system and the operational design domain limits as well.
所以我们这里真正的是第三级，基本上OEM对这交互时间负有责任。这基本上就是问题所在。的确，这就是为什么我们看到了所谓的二级加和各种其他名字的出现。二级++，三级-，有各种各样的名称。 但本质上，这就是我们所看到的，这些高速公路辅助系统确实具有非常先进的系统功能，但始终由驾驶员来负责。 而且，Super Cruise确实是这一级别上最先进的系统，但仍具有驾驶员监控系统和操作设计领域的限制。
JUNKO YOSHIDA: In your opinion, that’s where the driver monitoring system needs to take a lot of responsibility in a way.
JUNKO YOSHIDA: 在你看来，驾驶员监控系统需要在某种程度上承担很多责任。
COLIN BARNDEN: Yes. Really, the future that I see is driver monitoring systems in absolutely every vehicle. Because really what we need is… That’s essentially where technology can really help humans, is around distraction and fatigue. And there’s starting to be some handover between the automated driving features and the human driver. Call that Level Two Plus, Level Three Minus, whatever. That essentially is where driver monitoring is absolutely critical. And that’s really what we’re seeing with Super Cruise and the recently launched system in the Cadillac Escalade, which is a seeing-machine system. And that’s using infrared vision to track the head position, the eye gaze of the driver, and looking for evidence of fatigue in terms of the blink duration of the driver. And really you can use both systems in highway assist Level Two Plus, or you can use it down into Level Two systems as well, to keep the driver engaged in the driving task and for the system to know at all times what the driver’s engagement level and awareness is.
COLIN BARNDEN: 是的。实际上，我所看到的未来是每辆车上的驾驶员监控系统。因为我们真正需要的是……本质上，技术真正可以帮助人类的地方，是围绕注意力分散和疲劳缓解。自动驾驶功能和人工驾驶之间已经开始有一些交接。所称二级+，第三级-。从本质上讲，驾驶员监控是绝对关键所在。这就是我们在Super Cruise和凯迪拉克凯雷德（Cadillac Escalade）近期推出的系统中看到的东西，这是一种视觉机器系统。利用红外视觉驾驶员的头部位置和视线，并根据驾驶员的眨眼时间寻找疲劳迹象。实际上，你可以在二级+公路辅助系统中同时使用这两个系统，或者也可以将其用于二级系统中，以使驾驶员始终参与驾驶任务，并使系统始终知道驾驶员的参与程度和意识是什么。
JUNKO YOSHIDA: But this is my biggest concern, actually: Waymo calls its autonomous car (they actually trademarked it) “the world’s most experienced driver.” I just laughed because it’s got a little TM mark at the top of that writing. But if an automaker claims they are shipping “the world’s most experienced driver,” and if the world’s most experienced driver gets into an accident, it is the automaker who has to be responsible, isn’t it?
JUNKO YOSHIDA: 但这实际上是我最大的担忧：Waymo称其自动驾驶汽车（实际上是它的商标）是“世界上最有经验的驾驶员”。 我被逗笑了，因为在该文章的开头有一个TM（trademark：商标）标记。 但是，如果汽车制造商声称他们正在交付“世界上最有经验的驾驶员”，并且如果世界上最有经验的驾驶员发生了事故，那么必须由汽车制造商来负责，不是吗？
COLIN BARNDEN: Exactly that. And this is why I think the AV tech industry with their huge budgets, but really their naiveté and their swashbuckling “we can do everything,” the traditional automakers, they’ve kind of been here before. And this is essentially really what we can see, is that this industry, the automakers, have really taken a bit of a step back and had a few deep breaths and said, We just don’t want to do this. And we can see from the videos on YouTube of the crazy things that happen with autopilot use in the real world. And essentially the way that humans try and trick the system into doing crazy things with the water bottles and getting into the back of the vehicle and sleeping. And the traditional automakers, that is not the sort of business that they want to get into. That might work for Tesla, but the real sensible, the GMs and the Fords of the world, the Toyotas, they don’t want to get involved in that at all.
COLIN BARNDEN: 就是这样。这就是为什么我认为AV行业拥有巨大的预算，但实际上他们天真且自命不凡“我们可以做任何事”，传统的汽车制造商也经历过这个阶段。 而这正是我们所能看到的，这个行业，汽车制造商，实际上已经退后一步，深呼吸了一下说：我们只是不想这样做。从YouTube上的视频中我们可以看到现实世界中使用自动驾驶时所发生的疯狂事情。本质上是人类试图欺骗系统做疯狂的事情，并钻进车后排睡大觉。而传统汽车制造商不同，这不是他们想要涉足的业务。特斯拉可能想去做，但真正明智的通用汽车、福特汽车和丰田汽车等，他们根本不想参与其中。
JUNKO YOSHIDA: So now we see actually the divide has become much clearer than three years ago I would say.
JUNKO YOSHIDA: 所以目前我们看到的现状，实际上已经比三年前更加明朗。
COLIN BARNDEN: Maybe even three months ago. And I think really even post-CES. I think there’s been this collective wake-up call, really, within the traditional automakers, that they don’t want to do this. And there is a lot… What’s really interesting, about 90% of the light vehicles in use in our world– and there’s about one and a half billion of them by my estimates — they’re Level Zero. So this argument that we need to remove human drivers from the task of driving in order to save lives, when 90% of the vehicles in use are Level Zero, have no automated driving features whatsoever. It seems really, really strange to me.
COLIN BARNDEN: 甚至比三个月前更明朗。我认为甚至是比CES后的情况更明朗。我认为，在传统汽车制造商内部，确实集体敲了个醒钟，让他们意识到其实并不想这样做。而且还有很多这样的例子…真正有趣的是，世界上约有90％的轻型车辆正在使用——据我估计，其中约有10亿5千万是零级。所以这一论点认为，我们需要把人类司机从拯救生命的驾驶任务中剔除，当90%的在用车辆都是零级时，根本就没有自动驾驶的功能。我觉得这观点真的很奇怪。
So what looks to me is, let’s put ADAS into these vehicles, let’s put a driver monitoring system in there, and then let’s let the market decide if we need automated driving at all or not.
JUNKO YOSHIDA: Well said. Thank you so much.
JUNKO YOSHIDA: 说得好。非常感谢你参加节目。
COLIN BARNDEN: Thanks, Junko. Always great to talk to you.
COLIN BARNDEN: 谢谢Junko，和你交谈总是很愉快。
BRIAN SANTO: To be fair, companies developing autonomous driving technology have demonstrated some truly remarkable capabilities. A video shown by MobilEye at CES was just astonishing. We’ve got a link on the podcast page if you want to see it. It is truly impressive.
BRIAN SANTO: 公平地说，开发自动驾驶技术的公司已经展示了一些真正出色的功能。 MobilEye在CES上展示的一段视频令人惊讶。如果您想观看这段视频，可点击本播客页面上的相关链接。视频确实令人印象深刻。
We also have a link to the story that Junko referred to in her conversation with Colin, in which the AAA demonstrated vehicle sensor systems failing miserably. You’ll want to see that, too.
The issue isn’t that AVs can’t do amazing things. The issue is that they cannot do amazing things at a reliability rate we expect from any other mission-critical system, which is better than five-nines. They won’t be anywhere near that reliability for many years to come.
Meanwhile, driver-assistance technology is maturing. New features can be rolled out progressively and — and this is the important part — automakers can start rolling them out sooner rather than later.
Well, that’s your Weekly Briefing for the week ending February 21st. The Weekly Briefing appears every Friday. You can listen on Spotify, iTunes, Stitcher and of course find it on our web site at eetimes.com, where you can find a transcript of every podcast.
Do us a favor: If you like what you’ve been hearing, share the podcast with your co-workers and friends.
This podcast is Produced by AspenCore Studio. It was Engineered by Taylor Marvin and Greg McRae at Coupe Studios. The Segment Producer was Kaitie Huss.
本播客由 AspenCore Studio制作。Coupe Studios的Taylor Marvin和Greg McRae担任设计。Kaitie Huss担任片段制作
I’m Brian Santo. See you next week.