Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Self-Driving Cars Must Meet 15 Benchmarks in U.S. Guidance

bloomberg.com

Post reply on HN
Print view View original

281–290 of 300 posts

Re: Self-Driving Cars Must Meet 15 Benchmarks in U.S. Guidance

#281

Post by paulmd »

wott wrote: post #271
seanmcdirmid wrote: post #254

Earlier quoted context omitted.

An autopilot in a Cessna can most definitely not take you around the world. Also, autopilots even in a commercial jet keep the plane steady along a route, they don't deal with collisions at all...you could call it "lane keeping" if you want to sue the airline industry for misleading us over a term they created. Autolanding is not autopilot. In fact, autolanding is not even certified if multiple autopilots are not ava…

> Also, autopilots even in a commercial jet keep the plane steady along a route, they don't deal with collisions at all... Yes, they do. Then the problem is when the 2 autopilots of the 2 planes take the same evasive manœuvre...

Ah, the awkward old dance where two people try to walk through the doorway and end up mirroring each other's movements trying to get out of each other's way...

The aviation world actually solved this problem a long time ago. Everyone turns to their right. FAR 91.113:

> (e) Approaching head-on. When aircraft are approaching each other head-on, or nearly so, each pilot of each aircraft shall alter course to the right

(among other collision-avoidance regs in that section)

Re: Self-Driving Cars Must Meet 15 Benchmarks in U.S. Guidance

#282

Post by Practicality »

crististm wrote: post #274
lisivka wrote: post #266

Earlier quoted context omitted.

IMHO, autonomous car should use at least two autopilot systems from independent vendors, to avoid single point of failure.

Two is not enough for voting. If you have two clocks you don't know what time it is.

Sounds like we need three, à la Minority Report.

Re: Self-Driving Cars Must Meet 15 Benchmarks in U.S. Guidance

#284

Post by yoav_hollander »

seren wrote: post #264
yoav_hollander wrote: post #87

Earlier quoted context omitted.

A few comments on this: First, as etendue says, it is not easy. The problem of mixing “Boolean” verification with probabilistic, less-deterministic verification is especially hard. I discussed this a bit in [1], if you care to take a look. Also, I think most current AVs are not driven by DNNs at the top level (comma.ai [2] is one exception). See [3] for some discussion of that, and of verifying machine-learning-based…

Thanks for your input, really interesting topics on your blog as well.

Thanks. I did a second pass through the policy paper, and put a summary of the verification implications here: https://blog.foretellix.com/2016/09/21/verification-implicat...

Re: Self-Driving Cars Must Meet 15 Benchmarks in U.S. Guidance

#285

Post by lisivka »

ekianjo wrote: post #275
lisivka wrote: post #266

Earlier quoted context omitted.

IMHO, autonomous car should use at least two autopilot systems from independent vendors, to avoid single point of failure.

Dangerous. What if they end up with different interpretations, which one is actually correct ?

Brake.

Re: Self-Driving Cars Must Meet 15 Benchmarks in U.S. Guidance

#286

Post by lisivka »

crististm wrote: post #274
lisivka wrote: post #266

Earlier quoted context omitted.

IMHO, autonomous car should use at least two autopilot systems from independent vendors, to avoid single point of failure.

Two is not enough for voting. If you have two clocks you don't know what time it is.

It's enough to catch errors to alarm human and brake.

Re: Self-Driving Cars Must Meet 15 Benchmarks in U.S. Guidance

#287

Post by Spooky23 »

dogma1138 wrote: post #261
Spooky23 wrote: post #241

Earlier quoted context omitted.

People said the same thing about cold boot attavks against encryption keys. Yet today the police and others are using that and other NAND attacks regularly. HIPPA is a very easy compliance standard to meet. If it seems difficult to meet those requirements with your standard tool configurations, you should think about what that means with respect to the integrity of your data.

I would like to see a case of a cold boot attack by the police.

Memory forensics is a thing.

Google around with terms like forensics and "Volatility" or "Volatility toolkit" and you should find some presentations and other references.

Re: Self-Driving Cars Must Meet 15 Benchmarks in U.S. Guidance

#288

Post by kwhitefoot »

jaxbot wrote: post #219
ekianjo wrote: post #216

Earlier quoted context omitted.

Why was this not widely reported?

Last I heard, nobody has been able to verify if the car was actually in autopilot mode. However, the emergency braking also clearly failed, if the police report that no attempt to stop was made is true.

Surely the car keeps a log?

Re: Self-Driving Cars Must Meet 15 Benchmarks in U.S. Guidance

#289

Post by sharemywin »

sharemywin wrote: post #98
sharemywin wrote: post #82

Earlier quoted context omitted.

Regulatory Capture will by definition always result in Regulations working in the interest of the major companies(or special interest) in an industry, rather than the public interest. Definition: Regulatory capture is a form of government failure that occurs when a regulatory agency, created to act in the public interest, instead advances the commercial or political concerns of special interest groups that dominate t…

The question is does regulatory capture always happen?

or a better question does a free market alternative exist?

Re: Self-Driving Cars Must Meet 15 Benchmarks in U.S. Guidance

#290

Post by dogma1138 »

Spooky23 wrote: post #287
dogma1138 wrote: post #261

Earlier quoted context omitted.

I would like to see a case of a cold boot attack by the police.

Memory forensics is a thing. Google around with terms like forensics and "Volatility" or "Volatility toolkit" and you should find some presentations and other references.

I know what memory forensics is and I use Volatility and Second Look and quite a few other things pretty often, I've asked specifically about an instance of cold boot attack that you claimed in a hyperbole that are used often or at all by the police.

You know what I don't need a case, please find me a jurisdiction in which cold boot attacks have passed forensic certification, e.g. a link to the process like for example from a body equivalent to the ASTM https://www.astm.org/Standards/forensic-science-standards.ht... would suffice.

Post reply on HN