Delphi buys self-driving car startup Nutonomy for $400M | Hacker News Reader

Delphi buys self-driving car startup Nutonomy for $400M

techcrunch.com

78 pointsedshiro44 comments

This was an inside deal. Looking at the partners page for Fontinalis (VC firm), you've got Bill Ford Jr. and Jr. Jr.

All of their other investments are unremarkable.

I work for one of the big three - this autonomous stuff is way, way further out that what anyone is saying. I can't put a year on it because markets can stay irrational longer than you can stay liquid but it's all a head fake/game by upper management, who have finally found the Silicon Valley playbook of making wild ass promises now they that they've largely shed their pension obligations (my Dad, a slightly below classified level engineer at GM, will be taking the cash buyout instead of pension. Who can trust GM at this point? They've gone bankrupt before).

Big Three company with that's poised to actually make real profits? FCA. New Wrangler/Grand Cherokee/Ram/Wagoneer/Grand Wagoneer in the next 24 months. Today, those vehicles make the lion share of profits and its not even close. And all of those vehicle redesigns will have mild hybrids for their launch, as a hedge against oil going up. FCA improving quality by going upmarket/simply charging more. Dealer model isn't ideal, but that will change over time.

What do you make of Waymo's Early Rider programme?

https://waymo.com/apply/

I'd like to see a Waymo Pacifica going north on I-75 the morning after we get 6-8 inches of snow.

To which someone on HN will respond - 'well it's dangerous to be driving during that time anyway.'

Well, yes - but people still need to get to places (hospitals). And I've yet to see regulation under what situations autonomous vehicles can and cannot operate. And the fact that hasn't been addressed yet in any legal manner/you're not seeing cars available with it on a lot other than Tesla (who is playing with fire without LIDAR) tells you where this is at.

Not saying it shouldn't be worked on/this isn't coming, but we've got a ways to go with this yet.

@markbspiegel and some of the other Tesla bears (read: those with large short positions) have documented the NHTSA cases that have been filed against Tesla.

I get that you are contributing a lot in this thread, and my position on how soon self driving cars will be widespread is negative. That said "I'd like to see a Waymo Pacifica going north on I-75 the morning after we get 6-8 inches of snow." is a (edited:) poor argument. There are plenty of non-snowing cities to test the crap out of self-driving cars first.

(And clearly, that's what the Waymo early-rider program is trying)

Yeah, but when you sell them to the public, you can't stop people from driving into other states or areas where conditions are less favorable; which I think is relevant and applicable to their original argument that actually having self-driving cars out in the market is further out than what the industry is saying.

But yeah, in relation to the Early Rider program, that's probably one of the reasons you must apply. They were asked what they thought of the program, and my interpretation of their response (not putting words in their mouth, but as I understood it) would be along the lines of, "Must be a controlled trial, since their technology is likely far from being able to handle anything outside of favorable environments."

At least for Waymo, the biggest use case is to test the rideshare program with driverless cars with Waze aka a driverless Uber. They can apparently operate in specific cities first, start to monetize and figure out extreme weather conditions and edge cases.

This was the point I was making. The first gen can just be rideshare (vs buy the car).

IMO self driving car ownership won't be cost competitive with ride-sharing/tncs.

Why would a car that can drive itself be idle 95% of the time?

Self driving cars would still suffer deprecation, wear and tear etc relative to miles driven. I don't see how self-driving will directly change any of those.

This and the other replies are all valid points, and further illustrate the larger picture of the process involved in getting self-driving cars to the market.

You can stop the self driving features from working in some geographic areas.

GM is launching an enhanced cruise control that only works on limited access highways. They lidar mapped all the limited access highways in the US and Canada for the system.

What does it do? The major highways are under nearly constant construction, not to mention sometimes there are accidents, lane closures, and other obstructions in the roadway. I am curious what would need the accuracy of a LIDAR map but is okay with one that's not realtime (and possibly wildly inaccurate).

The only thing I'm sure they use the map for is limiting the feature. They probably also extracted features like bridge abutments (where they might show up as a potential obstruction to the sensors).

The system is just a hands free cruise control, it still has an attention monitor and won't even do things like switch lanes: http://www.cadillac.com/world-of-cadillac/innovation/super-c...

You don't need to sell them to the public. City specific Taxis are still a worthwhile investment

>Yeah, but when you sell them to the public, you can't stop people from driving into other states or areas where conditions are less favorable; which I think is relevant and applicable to their original argument that actually having self-driving cars out in the market is further out than what the industry is saying.

You actually can. For instance - radar detectors are legal in some states and not others. If you cross that state line you can be pulled over and cited immediately.

Or to go one step further, you could be driving a truck carrying a load that's perfectly legal in the state you left, cross state lines, and be stopped: https://ops.fhwa.dot.gov/freight/policy/rpt_congress/truck_s...

> And I've yet to see regulation under what situations autonomous vehicles can and cannot operate.

It wouldn't be implausible in the future for an insurance company to deny a claim based on the fact that the owner knew the autonomous system didn't work in snow, but drove anyway.

"I'd like to see a Waymo Pacifica going north on I-75 the morning after we get 6-8 inches of snow."

This kills the argument.

I agree with you that it's further out than we'd like to think but since you're an insider - do you see shorting Delphi as a decent investment play?

I don't understand this market at all. I'm in cash/Ethereum. I've suspended my 401K contribution and buy a small bit of ETH every paycheck.

For example, Caterpillar -

(Revenue/EPS)

Q3 2012: $16.4B/$2.54 Q3 2017: $11.4B/$1.77 Stock price: 2012: $85 2017: $139

I've suspended my 401K contribution and buy a small bit of ETH every paycheck.

What is the rational here? Interested in ETH, however I haven't seen it move past $400 in months.

Just because its not tracking with BTC at the moment does not mean its never going to increase in value.

You mean $300? ETH seems to lack the drama of BTC and Byzantium is humming along. I see the ETH dev team being active/creating off-chain solutions (whether the market will adopt these solutions is a valid question to ask) - I think these solutions offer value that isn't fully priced in.

I like ETH (especially the lack of fork drama every few months) but the only thing that's stopping me regularly buying more is a fear for what will happen to the price if the ICO market starts receiving sustained negative attention from regulators. Blog posts like this one suggest a (probably necessary imho) storm is coming: https://blog.colony.io/the-colony-token-sale-7ac14c845bc0

I know ETH isn't just about ICOs but I remain skeptical how big of a market there is, outside of the ICO world, for Dapps and smart contracts. That's before considering potential competition from projects like NEO, ARK and LISK; which could further splinter demand.

Wouldn't the pre-tax nature of a 401k mean that an alternative investment would need to return ~25% in the first year in order to break even with a 401k contribution returning 0%?

I'm not arguing that the market is currently rational; just surprised you've suspended 401k contributions because of it.

Yup, unless GP for some reason can only buy 401k into their own company, rather than some broad-market fund, in which case if they're specifically skeptical about their own company then maybe not putting stuff into a 401k.

But yea generally pre-tax dollars are better than any other investment you can reliably make.

It's not rational. It all depends on wall street hype and their interest is to keep increasing value so any remote news that has no play in the market is used to hype something up every week. Companies with very solid earnings and increasing profits/growth like Apple get little love while others keep increasing then all of a sudden they barely bear estimates and the shares climb up 30% in 3 months.

My lesson, just ride the flow

> since you're an insider - do you see shorting Delphi as a decent investment play?

I think that's a very dangerous thing to ask. If you get an answer, run.

Those cars don't make up the lions share of the profits, they make up the profits. Jeep is worth more than FCA because the other brands, with the exception of Maserati, are losing money.

You're right about the driverless thing being farther out than these big companies seem to be leading people to believe. It's somewhat straightforward to get driverless working in 90% of cases, but the remaining parts are basically unsolved as far as I'm aware. I think automated highway driving is the best we're going to see for at least a few years.

'Jeep is worth more than FCA because the other brands, with the exception of Maserati, are losing money.'

The L-cars are profitable (Charger, Challenger, 300). Not as profitable as Jeep, but they make money.

Thanks for your insights. I also believe that we are much further on the L4/L5 front than what most automakers proclaim. I was quite "disappointed" with the $400M acquisition price tag when companies like Cruise sold for $1B to GM and MobilEye were acquired by Intel for $15B.

Do you know how far are Nutonomy to the likes of Waymo/Cruise or even Tesla/comma.ai from a tech point of view? There is so much opacity in this market that it is very hard to determine where each player stands.

Also seems to me that the way to play it smart if you are a startup in automotive space is to build ever more advanced ADAS features and sell them to carmakers or if you are building driverless vehicles deploy them in very safe and sunny environments. Would really appreciate getting your thoughts on this as I am in interested in this sector.

ADAS?

Advanced Driver Assistance Systems - basically things like Lane Departure, etc. Sorry should have clarified this point; more info on wikipedia: https://en.wikipedia.org/wiki/Advanced_driver-assistance_sys...

Anyone know the context behind Elon Musk’s “groan” tweet about this acquisition: https://twitter.com/verge/status/922828400439504897

Maybe because a acquisitions in this phase is rarely a success signal. It's hard to innovate now that they aren't independent anymore.

Nutonomy founder Karl gave a talk at MIT and one of the things that surprised me was that they used very little machine learning. Most the work is done using a pre defined route map and the rrt star algorithm for making local decisions.

That's actually a good thing - I've done a fair amount of work on autonomous robots (not land) and I'd say that the AI/Machine Learning word is thrown about a lot for no reason. Many of their challenges can be solved with tried and proven algorithms like rrt star. My personal opinion is that they're a autonomous car company not a research lab, so the focus is indeed correct.

Most teams use ml heavily for perception but not necessarily for other things. You need explainability when it comes to safety, and maps are needed no matter what if you want your car to get around.

The opposite of Drive.ai and Nvidia approaches (end-to-end deep learning).

That's true of all the autonomous vehicle projects: the planners, etc are all written by hand. The ML is for "image" processing (I say "image" because it's object extraction from multiple sensors, in most cases including LIDAR).

and those outputs probably map into a bayesian network to make decisions.

I read the article and still don't get what Delphi does, I'm pretty sure they are not talking about programming language tho :)

Delphi is one of the biggest Tier-1 automotive suppliers. They effectively supply parts, components and software to major auto manufacturers.

Other Tier 1s: Bosch, Magna, Continental, Denso etc.

https://en.wikipedia.org/wiki/Delphi_Automotive

One point of interest is that Delphi was previously the internal supplier organization for GM, spun off in the 90's.

> Nutonomy’s partnerships with Lyft, Groupe PSA and others will continue and the company will work in parallel with Delphi from its home base in Boston

GM also has a partnership with Lyft, and Google just invested in Lyft as well.

If Lyft were a person, one might describe them as promiscuous.

I honestly thought Delphi went out of business years ago.

Reply on news.ycombinator.com