Yesterday the U.S. government announced a new round of autonomous vehicle (AV) guidelines at CES 2020 called AV 4.0. Those new guidelines quickly upset safety advocates, however, as they merely “promote voluntary consensus standards.”
The Associated Press reported that the proposed AV 4.0 guidelines were announced by Transportation Secretary Elaine Chao, who said that “It recognizes the value of private sector leadership in AV research, development and integration."
Chao also said the new guidelines would unify work on AVs across 38 federal departments and agencies and establish a list of government principles. According to the AP, those principles include (cyber)security, privacy, protecting users and communities, promoting efficient markets by protecting intellectual property and modernizing regulations, and facilitating coordinated standards and policies. The document also says the government will enforce existing laws to ensure companies don’t make deceptive claims about the capabilities or limitations of AV technology.
However, some said the guidelines would fall short of expectations in areas such as innovation and safety. This includes auto safety advocates such as the National Transportation Safety Board, which investigates crashes and makes safety recommendations. The NTSB already condemned a lack of regulation for testing AVs. On the other hand, the agency says it doesn’t want to stay in the way of innovation.
Specifically, the guidelines lack details on how the government would enforce them or accomplish its goals. Instead, the U.S. government “will promote voluntary consensus standards,” without specifying what those standards should be. The document states: “Voluntary consensus standards can be validated by testing protocols, are supported by private sector conformity assessment schemes, and offer flexibility and responsiveness to the rapid pace of innovation.”
Taking a step back, Mobileye has previously criticized the U.S.’s lack of autonomous vehicle regulations. Mobileye CEO Amnon Shashua has explained that this creates difficulties with dealing with accidents, for example. That's probably at least partly why the company doesn't want to rush it to start up its Mobility-as-a-Service (MaaS) plans in the U.S., as is evidenced by its timeline to begin its robotaxi activities in 2023--a year later than the company’s current plans for Israel, France, China and South Korea. (More details on its plans for South Korea were announced at CES.)
The guidelines will be published in the Federal Register, followed by a public comment period.
The Boeing 373 Max crashes have shown that even safety-critical industries are incapable of regulating themselves. Some years prior to the introduction of the Max, Congress cut back the FAA's role in certifying new aircraft, and that's the result we got. Ultimately, it turned out to be bad for both the public and the industry.
So, just don't have regulations? Because, waiting will only have the effect of players getting bigger and having more clout.
I'm sorry, lack of regulations not an option. It's best to have a fair and open process around creating them. It's not that hard to do.
However, in this case, the Trump administration is trying to undermine electric vehicles. They could care less about people's safety. This regulation is specifically targeting Tesla because their current lead in both autonomous and electric vehicles. The current data around autonomous driving is that a Tesla driving on Autopilot is 5.4 X less likely to be involved in an accident than a human and this product isn't even feature complete. Halting this technology would cause more death and harm already.
No, these regulations will be similar to the "regulations" Germany applied to autonomous vehicles that limits the steering angle. A feature purposefully intended to make the entire system useless and more dangerous thereby nerfing one of the best features of a vehicle that is cutting into their homegrown car manufacturers lagging behind the competition.
Republicans don't just suggest regulations unless they support institutionalized industry "leaders." You know better than that.
That's part of the problem. Tesla just decided for itself when the technology was good enough to unleash on the public, and then decided for itself the particular ways that they deployed it.
That's another false dichotomy. Nobody is saying that development shouldn't continue - just that deployment should be according to thoughtful and consistent policies and procedures. At their best, that's really all regulations are.
How do you know what shape regulations would take? US regulations rarely align exactly with Germany's. Why should they suddenly match up, in this case?
Let's also keep in mind that localities hosting AV testing usually set rules on participants too. The lack of federal regulation does not mean there aren't local regulations. What works in one region may not work in another with different weather, infrastructure maintenance, road layouts, and other variables.
The current dominant companies will get bigger and expand their influence even in the presence of regulations. AV regulations should ratchet down once the industry starts settling on a couple of proven solutions. Right now nobody can agree on the right mix of sensors, maps, and algorithms.
What I do think would be helpful are standards that show the best implementations of specific pieces of sensor technology -- standards that may start bringing all the differing implementations closer together. Leadership from someone like NIST or UL is needed here.
This is really a cop out answer. You seem to be saying that unless deaths can be completely prevented, it's not worth doing anything about it. Fortunately, most people are smarter than that.
The actual numbers of deaths, injuries, and property damage matter. How the technology is deployed obviously affects that.
That's just establishing liability, which I'm sure clever lawyers are already trying to work around with EULAs and what not. As for PR, recall how Boeing mounted a big PR campaign to establish pilot error, in the case of the 737 Max crashes. PR is an unreliable tool.
This argument is utterly toothless. In the vast majority of companies, people don't actively set about to harm their customers (or other members of the public), but financial & competitive pressures can result in decisions that have exactly such consequences. The point of regulations is to set guard rails, so that corners can't be cut that would make their products unnecessarily or unacceptably unsafe.
I didn't think we were talking about testing & development. Sure, there's going to be some special arrangements for that, and that's reasonable. No, this is really dealing with deployment.
Tesla didn't limit its automated driver assistance to just specific localities - they rolled it out to their entire fleet. The public doesn't limit their driving to local roads and city boundaries or county lines. When these systems are rolled out to the public, there need to be federal regulations, because trying to rely on a patch-work of local ones just isn't realistic.
Unless you're going to limit where people can drive these vehicles, then yes. there needs to be federal regulations.
One fundamental point of disagreement seems to be the value or necessity of this technology. Sure, I'd agree it has value to the public. However, it's not being developed to satisfy overwhelming public demand - it's being developed because it's possible, and there's a race because rich VCs and technologists want to be first, meanwhile other big automotive industry players don't want to be shut out. As such, I don't mind if industry progress is slowed, although I also don't think that's necessarily the consequence of regulation. A risk, but not an inevitability.
Anyway, it seems you're getting your wish. However, perhaps we'll all pay a high price for that, in the event of a mass-cyber attack or a raft of accidents that occur after some defect gets pushed out, in a software update. We'll see just how well litigation and PR work to prompt industry reform. I think it's sad that people will have to get hurt, before sanity hopefully prevails.