Topic: Autonomous driving
Weekly Narrative Video – Tech Disrupts Transportation (Apr 29, 2017)
Uber Exec in Charge of Pittsburgh Self-Driving Test Quits (Apr 18, 2017)
GM has filed for and received a tax credit in the sum of $8 million from the state of California in return for investing $14 million in office space and related items this year and committing to hire 1163 employees over the next five years for its self-driving tech subsidiary Cruise. Given how the importance of autonomous driving technology will grow in the coming years and the fact that California is the hub of much of the testing, it’s logical that GM would want to increase its base there significantly. However, these 1163 employees represent a more than three-fold increase in its employee base in the state, and the average salary GM is projecting for those employees is $116,000, so my guess is they’ll mostly be skilled engineers.
Cadillac takes aim at Tesla’s Autopilot with ‘hands-free’ Super Cruise technology – The Verge (Apr 10, 2017)
One of my big objections to Tesla’s Autopilot technology has always been the name, which connotes a level of autonomy the system doesn’t actually aspire to and which it certainly doesn’t deliver. Tesla has partly dealt with that issue by updating its software to require users to keep their hands on the wheel, but does little else to ensure attention, which means that even when the system performs as it should, there’s little guarantee that the human driver will. Cadillac today announced a new Autopilot-like feature but very sensibly named it in a way much more likely to give buyers and users an accurate impression of what it does, tying it to the very familiar cruise control already in almost all new cars. However, the more important thing in my view is that the system also comes with lots of protections designed to ensure that the driver does actually pay attention, which is a huge issue in situations where attention but not activity is required, such as driving a car with this kind of intelligent cruise control running. There’s a long history of scientific research in this area, and it all says that paying attention in a passive way like this is something human beings aren’t good at, and Cadillac’s new system is designed to help the driver stay attentive. The big question about this new system, though, is that although it’s being billed as LIDAR-based, it’s not using a LIDAR in the car but instead using mapping data previously generated by LIDAR, which means it’s non-real-time. That, in turn, means that if anything has changed in the road environment since the map was generated, the car won’t know about it, and GM doesn’t seem to have talked much about how frequently it’s going to update its maps of US and Canadian highways to mitigate this.
via The Verge
We finally have a fleshed-out response from Uber to the Waymo lawsuit over stealing of LIDAR technology, and it doesn’t do much more than muddy the water over this issue. The biggest sticking point here is that Anthony Levandowski, who is alleged to have stolen files from Waymo before he left and used these to develop LIDAR technology at Otto and then Uber, refuses to cooperate with the investigation, and Uber refuses to compel him as an employee to cough up the files. Uber also argues that its LIDAR design is different in key respects from Waymo’s and therefore that it clearly hasn’t been copied from it. The judge seems to be highly skeptical of Uber’s claimed inability to do anything with regard to the Levandowski files, and seems minded to grant at least a temporary injunction against Uber’s LIDAR technology. Uber’s claims that such an injunction would significantly harm its business seem like nonsense – this technology has nothing to do with its core business today and is merely being tested in a few cities. A longer-term injunction would obviously be more damaging because it would stop Uber from advancing the technology, but in and of itself that’s not a valid argument against such an injunction should the judge determine that the design was copied. Lots more to come on this, no doubt.
via Business Insider
This article is based on a study by a company called Navigant Research, and it seems to be an evaluating of companies’ strategic assets rather than any actual capabilities today, so it’s worth noting that context for their rankings of companies here. Notably, they rank traditional carmakers in the first six spots, with Waymo apparently the first non-traditional / tech company in the rankings. That’s notable, because all the numbers suggest Waymo is out in front in testing of autonomous driving technology in California by a long way, and although we don’t have equivalent data for Michigan, where Ford does much of its testing, I’d be surprised if it had done many more miles. So this is mostly an evaluation of the benefits the big automakers derive from their existing massive scale and capabilities in building vehicles and bringing them to market, something none of the pure tech companies has (Tesla, of course, has some small-scale manufacturing capability and is looking to ramp fast, but comes in 12th in the rankings nonetheless). This jives with my perception that, even as these tech companies do increasingly well in developing their own technology, they’re very unlikely in most cases to build the cars, and as such the traditional car companies are still in a position of strength and potential leadership when it comes to actually building and deploying the technology.
via USA Today
I think the headline here would more appropriately read “Uber crash shows bystanders see human traits in self-driving software” because there’s no evidence that the car actually gunned a light turning yellow in this case, but that’s how witnesses perceived it. (I joked on Twitter earlier that perhaps Uber’s autonomy software had been taught the company’s values, one of which is “always be hustlin'”.) The reality is that self-driving cars are often taught to emulate human behavior rather than driving in some idealized perfect way, because that’s what makes human passengers feel comfortable and ultimately trust the technology. But I very much doubt that Uber’s cars are taught to accelerate through lights in the process of turning yellow or red. It appears that police concluded Uber’s technology was not at fault in this crash, and after a brief break over the weekend, its cars are back on the roads in the various cities where they’re operating. But given Uber’s failure rates relative to Waymo’s, and the fact that Uber cars are carrying paying customers, there’s certainly potential for a lot more crashes, some of them actually Uber’s fault.
Uber’s autonomous cars drove 20,354 miles and had to be taken over at every mile, according to documents – Recode (Mar 16, 2017)
One of the great things about autonomous driving technology is that regulators require companies to keep track of how those cars are performing, and in the case of California that data is published annually, providing a great insight into how each company’s technology is advancing. However, occasionally, internal documents on testing emerge that provide lots of detail too, and such documents have apparently been leaked to Recode (and BuzzFeed). There’s lots of interesting data here, and it suggests progress is being made and Uber is driving lots of miles in its various cars. It’s worth comparing some of the numbers here for Uber with those reported by other companies in California by way of putting them in context: Uber says its rate of disengagements per mile is 0.8, for example, whereas Waymo’s cars in California are now at a rate of 0.2 per 1000 miles, or some 4,000 times better. Now, Waymo’s cars have been driving in the state for much longer than Uber’s, but that’s still a massive discrepancy in performance. And it’s also worse than Tesla’s rate of 182 disengagements over 550 miles driven in 2016. So it appears Uber has a long way still to go in autonomous driving, and it’s therefore remarkable that it’s already using these cars to ferry real passengers around Pittsburgh.
This feels like something of a slime ball move on Uber’s part on two fronts: firstly, trying to move the court case with Waymo out of open court and behind closed doors; and secondly, essentially trying to push the case off its back and onto Levandowski’s. I had said previously that the course was going to be fascinating for the details it would provide about how Uber developed technologies and how it would defend against what look like fairly solid allegations, but if it gets its wish here we won’t get to see any of that. But I think it’s the attempt to make this a case about an employee rather than the company that seems particularly sleazy – if the allegations are indeed true, then Uber and not Levandowski benefited the most, and making this seem like a dispute between an employee and former employer feels like a total misrepresentation.
via USA Today
Nvidia Partners with Bosch and Truck Maker Around Autonomy (Mar 16, 2017)
Here are two partnership announcements from Nvidia, the first a deal with automotive component maker Bosch to incorporate Nvidia chips in self-driving solutions, and the second with truck maker PACCAR for self-driving truck technology. Nvidia continues to be one of the biggest names in autonomous driving, and certainly one of the most successful chip vendors (hence Intel’s Mobileye deal). These deals come on top of lots of existing ones, but trucks are a particularly interesting area – it feels like that’s a segment of the market that could actually see real-world adoption of autonomy much sooner than cars.