Comment by Animats | Hacker News Reader

Comment by Animats | Hacker News Reader

So they drove a bobtail tractor 7 miles on an empty straight road with a human-driven blocking vehicle to prevent anybody from passing. We could do better than that in 2005 with our DARPA Grand Challenge vehicle. Everybody in the DARPA Grand Challenge did better than that. And they call this a "product".

"Working with partners, we will use unmanned trucks to haul freight in 2018." These guys are 5% of the way to where they need to be. This is a business where your bug reports are written in blood.

What they seem to have is an ordinary lane-keeping / auto brake system, like a lot of cars, plus drone-type remote control. All cameras, no LIDAR. Looks like a standard anti-collision radar in the front bumper. The steering motor doesn't seem to be redundant. This is demo-grade technology with too much hype and a tear-jerking video. They put that crap on the road and they're going to kill somebody.

Even Otto was better than this when they did their Budweiser demo on an empty road in Nevada. They at least had a potentially working technology. Years later, now owned by Uber, it still barely works.

You DARPA Grand Challenge guys certainly built some impressive stuff back then, and it's hard to shake a stick at what those teams accomplished.

The problem is, most of the industry has spent the last decade adding neat features and trying to improve their reliability...without much practical thought as to increasing controllability (in terms of accident avoidance) in the inevitability of a component failure.

Controllability has basically been shortcut by keeping a driver behind the wheel. We've done pretty significant testing - regularly hauling freight on the highway while in self-driving, no-one behind the wheel in super unstructured environments, dealing with extreme edge cases w/o intervention.

The big technical accomplishment that we're showing here, is that our system is controllable in the inevitable case of failure...without using a physically-present human as a crutch.

> without much practical thought as to increasing controllability

That's not fair, Waymo and others are using redundant backup systems and obviously try to think of every failure case they can. Besides, aren't the remote-control parts as likely to fail as any other? I sure as hell wouldn't want to drive a car over facetime...

It’s open secret in the SDV industry that all the top players have remote assistants systems, and that it is the to key removing operators from the vehicle. It’s a little disengenus to claim that Starsky is the sole player exploring this space. There are startups that provide full remote vehicle control [0], and GM’s acquisition of Cruise 1.5 years ago was in part driven by an especially good demo of their remote assistance tech.

All that matters in the end is the safety and cost of the resulting system. Remote assistance operators and bandwidth are not cheap, and a system that has a high level of remote faults is most likely operating too far out of operational domain to be truly safe.

[0] - https://spectrum.ieee.org/cars-that-think/transportation/sel...

Either you have a lot of remote drivers, or you get the "your call is important to us" wait.

Yes. I was expecting a real road test on an actual freeway. This was just a truck moving in a straight line on an empty secondary road.

This seems unnecessarily vitriolic. If someone has founded a self-driving truck startup that isn't going anywhere, who cares?

If Tesla's self driving truck division did a crappy demo then fair enough, but cut these guys some slack, at least they aren't in a cubicle at facebook trying to find better ways to sell ads for toilet paper.

It's completely fair and completely necessary. This isn't driving by any stretch. This is like comparing the Emacs Doctor to Alexa. I think it's the first time I clicked flag on a YC submission.

Edit: I think I overreacted a bit and shouldn't have flagged it. But I don't think the commenter you replied to did and can see why that comment is on top.

It's a bit harsh, but not unreasonable. You can buy this level of self-driving at many car dealerships right now. If you take the Udacity self-driving course, you could build this. Doesn't mean you're ready to ship a road-ready product. Remote control helps some, but it won't get you out of trouble. By the time the remote driver is connected and has situational awareness, it's too late.

This isn't webcrap, where it doesn't really matter if it fails now and then. People die when self-driving vehicles screw up. Especially with heavy vehicles. California DMV doesn't allow self-driving heavy vehicles on the road yet, and that's probably a good decision.

If YC is going to fund stuff like this, they need to get some avionics people on the team. Before they kill somebody.

> And they call this a "product".

To be fair, this is an absurdly competitive space. If they achieved something of similar magnitude for a problem dozens of other teams weren't already solving, it would seem much more impressive.

> If they achieved something of similar magnitude for a problem dozens of other teams weren't already solving, it would seem much more impressive.

Are you saying it would be more impressive if dozens of competitors weren't already doing what they're doing? I feel like you're arguing against Starsky's competence by pointing out that they're in a competitive space with noncompetitive technology.

Reply on news.ycombinator.com