Hacker News Reader

Comment by IgorPartola | Hacker News Reader

Here is my proposal for self driving cars: forget trying to make them drive on local roads at first. Add transponders to all interstate tarmac that can be used to detect location and lane position. Then use a much cheaper LIDAR just to map where the other cars are. Boom: you have made long haul driving autonomous. You can then slowly expand this to smaller roads. But why try to recreate human drivers when machines can use much better sensors than us, but can process visual info much slower, and gather it with worse fidelity?

View on HN

Your proposal is exactly how self-driving cars were created:

"Back in 1995 the goals for self-driving cars were more modest than they are today. They weren’t called autonomous, but self-driving. And there was no plan to have cars drive themselves on city streets, just on freeways and highways — on the Interstate. The plan was to bury cables in the pavement over which all the cars would drive and communicate with each other and with the road, itself. The goal was to fill the road with cars driving at the speed limit, spaced precisely one meter apart" [0]

[0] https://www.cringely.com/2016/08/25/self-driving-car-old-eno...

This goes out the window as soon as someone places a single traffic cone on the roadway. Lanes are not a fixed property of the road.

This was last year, but I live off of San Antonio Road and the one day that the Los Altos municipal maintenance staff placed the cones on the median instead of in the lane (I think in prep for the workers to trim the plants in the median), a Waymo self driving car in front of me stopped immediately upon encountering the coned region - there were no cars in front of it or pedestrians or workers within sight and we were going probably 35mph. I'm sure they changed the programming/data for that situation but there are several erosion drop offs on highway one along the coast that are coned off in unique ways, and everyone in a car is going by at 55-75 mph.

Yes, because Waymo doesn't require special detectable hardware as proposed by GP. And this is exactly the argument the parent tried to make: Your self driving car needs to react to unpredicted changes in the environment. Sensing/visual information (incl LIDAR) thus seems unavoidable to achieve safe self driving. But getting to that point and not having your tesla drive into a wall or your uber into a pedestrian seems like the hard part and from there it is probably not that hard to reach level 4 / 5. Safety is pretty much all we are complaining about in those threads, after all.

I was trying to be supportive of GP with an example but did not say that explicitly.

That means they would need to use new cones with transponders. Yes, it's more expensive, but also way more reliable.

The detection of the cones isn't the problem. The software properly identified that there were cones near the roadway, that's why it stopped. The really hard problem is programming decision-making logic that will properly ignore cones that aren't in the roadway, 100 times out of 100.

I used to do QA for the Waymo cars, and boy do they get worked into a tizzy if there's any kind of construction going on. It was the most common situation for remote assistance (read: people sitting in a room sending pathfinding suggestions) to get involved.

Judging from some of these comments, the problem is still very difficult.

So when a police officer puts down a cone behind a broken down car in the freeway--??

The cone that transmits to every vehicle within listening range: "don't hit me"?

Stealing that transponder code would be hilarious - place invisible virtual cones wherever you like!

(Seriously, cones really ought to be dealt with by image recognition? Easiest possible case, an object specifically designed to be seen?)

I can see the 007 getaway now. With a press of the button, a bunch of tiny transponders are thrown on the road. The bad guy's cars automatically brake because they can't find a way to get around the new obstacles.

Another press of the button dispenses a martini.

And when all the cones' fancy electronics break or aren't charged? Is the cop just supposed to not stop? What about poorer jurisdictions that can't afford it?

And if the officer puts down a regular cone and the driver is texting and doesn't see it? Like, yes this would require changes to how we do things. Police officers will have to be careful to use the new style cones correctly. Is this any different than when they first started having use cones/flares? There is a price for every technology. I mean one of the alternatives is that we rely on the LIDAR being so good as to not hit the car and the officer standing next to it because it recognizes that situation. Which it will. Most of the time. Until it doesn't. Seems to be that having the transceiver and LIDAR backup is better than fancy expensive LIDAR.

What if it's just a regular cone, like lots of people carry in their trunk?

And less likely to be stolen when the cones can be tracked.

Not just any transponders: transponders that never fail or run out of battery juice, and are never subject to RF interference from anything.

Because yours is a ludicrous proposal. It'd be well more expensive to modify infrastructure than to build a computerized driver. Not only is the modification expensive, you have to convince the local authority to foot the bill, do it right, maintain it, etc. Never going to happen.

Your proposal also lacks reliability. What if one of these "transponders" fails? We can build redundancy into a self-driving car quite easily (a second computer, never relying on just one sensor). Providing redundancy to road infrastructure planet-wide is ... a much larger problem.

You could easily convert HOV lanes into Automated Driving lanes. And if we have the ability to fund HOV lanes, we could also fund Automated Driving lanes.

Self driving cars are already dependent on standard infrastructure markings such as lane lines, and whenever those markings are confusing or faded it has lead to fatal crashes, such as the fatal Tesla crash on 401.

I predict self driving will not fully succeed until we build infrastructure to support it into the roads.

> Self driving cars are already dependent on standard infrastructure markings such as lane lines, and whenever those markings are confusing or faded it has lead to fatal crashes, such as the fatal Tesla crash on 401.

A Tesla is not a self-driving car and they won't become one with that approach.

Waymo vehicles have virtual maps that include the lines (along with a LOT of other data) so they are not affected by fading lines as they always know where the lines are supposed to be.

Basically Waymo is building the infrastructure you all want, but in a virtual sense instead of the extremely expensive and impossible physical one.

> Waymo vehicles have virtual maps ...

You only need one reality <-> virtual desync to cause a crash.

Well no, it's not like the map is the only thing used.

How well does the Waymo approach hold up to a GPS failure?

GPS failure is the entire reason the LIDAR map exists. The map is used to give centimeter accurate location without any GPS availability.

The problem isn't difficulty, the problem is cost.

We could've had self-highway-driving cars in the 60's if we wanted to tear up the roads and install guide cables under every lane. https://en.wikipedia.org/wiki/History_of_self-driving_cars#1...

”Providing redundancy to road infrastructure planet-wide is ”

IgorPartola doesn’t talk about “planet-wide.

”Not only is the modification expensive, you have to convince the local authority to foot the bill, do it right, maintain it, etc. Never going to happen.”

Not in the USA, where government solutions are “socialism”, and voters rather pay for individualized solutions (example: not allocating money to maintain roads, but letting individuals bear the cost of repairs caused by potholes (https://www.wired.com/2015/07/heres-much-citys-crappy-roads-...)

Other countries may act differently.

”What if one of these "transponders" fails?”

You use passive transponders that rarely fail, and put them at every x meter, where cars work fine if they’re at every 2x or 5x meters, and maintain your roads.

I don’t see why that would have to be more expensive than the existing practice of adding, e.g. https://en.wikipedia.org/wiki/Cat%27s_eye_(road) and various lane markers to roads (cat’s eyes probably could be sufficient for relatively good lane keeping)

Even if it is, it may still be a net win if it means lanes can be made less wide, or distances between cars can go down, or even if it ‘just’ means fewer people getting killed and maimed in road accidents.

Roads are modified and maintained all the freaking time, new lanes, new lights, new cameras, new surfaces, etc. This is in the end the cheapest and most realistic way to automated transport.

Roads are modified and maintained, but only in a completely ad-hoc manner. So some localities have good roads and maintenance, and others have terrible roads and maintenance and even road design. You can't expect any kind of consistency, because roads are mostly handled at the local level in the US, and many localities are completely incompetent, or refuse to spend any money on it. What you're describing would require consistency, and that's a completely impossible expectation in America.

Roads are extremely expensive, adding sensors to new roads should actually be a rounding error.

what about the old roads?

And what about when you repave those roads? It's introduced an infrastructure burden that's there forever.

The interstate itself is an infrastructure burden that's there forever. Somehow we find a way to pay for it, no? Painting lane markers costs $X/mile, right? This would take the cost to $(X + x)/mile where x is the cost of the transponders. Nobody said self-driving cars would be cheaper, just safer.

No, we don't find a way to pay for it. https://www.ced.org/reports/fixing-americas-roads-bridges

at this point I have to ask why are self driving cars needed when we're talking about the tax payer eating up the cost and are we sure we don't have other things in which the money would be better invested.

sure, self-driving cars would be nice and we could even rationalize how this could make sense, but at the end of the day I would rather see everyone get 1) high-speed internet 2) clean water 3) reduce the dependency on fossil fuel 4) ensure access to quality food 5) move towards environmental friendly materials.

Why does taxpayers paying for it make it an issue for you? I don't really get why that factors into it, especially given that taxpayers already pay for the roads.

All the other things you mention are good, and could probably be a priority, I am not saying they shouldn't be. But if we are talking about a technical solution to the problem, here's the thing:

a. You can either have cheap self-driving cars and an upfront infrastructure cost + additional ongoing cost, or expensive self-driving cars capable of driving on shitty roads. My conjecture is that the latter won't 100% solve the problem anyways, so why bother? Let's not let perfect be the enemy of good.

b. Why do roads get lower priority for you than high-speed internet? And why do you prioritize high speed internet above reducing reliance on fossil fuels?

c. Maintaining roads is already a thing we do, it is already done via public funding, and again that makes your existing cars cheaper. You'd need a damn tank to roll through unpaved streets at 40 mph, let alone at 80 mph like everyone currently drives in their little Civics with 15 inch wheels.

Again, I don't know which thing should be a priority, not a problem I set out to discuss here. I am saying that IMO the cheapest and best way to get to self-driving cars is as I described. Let's upgrade the roads so we can all have nice things. This also lets us upgrade the rest of the roads to make them self-driving capable later. Or we could try to make a computer that's as good at driving in Montana blizzards as a human, and not see a single consumer self-driving car for the next 50 years.

not saying roads/infrastructure is not important. I’m saying that I would not want to pay extra for something that maybe is going to guide the cars. we already have something that does that: it’s called a railroad.

my bigger point is that we’ve fallen in love with the idea of having self driving cars. do we need self driving cars? who should shoulder the cost?

the things I gave as examples are not ordered. truly high-speed internet everywhere would be great as i believe can actually enable telecommuting and eliminate traffic. it can also enable access to education and drive us, as a society forward.

i believe (and someone correct me if wrong) that driving on the interstate (and self driving cars/trucks) is a somewhat [way] easier problem than driving on any road. you’re not going to get a lot of value from what you’re proposing, but it’s going to be super expensive.

In theory, theory and practice are the same. In practice, not so much. Sure, we may see prototypes of self-driving cars going on the interstate in California on sunny days. But I have yet to see a self-driving car make it through a blizzard in Montana. You can't see lane markings with a LIDAR if you can't see the road covered in an inch of hard packed snow. But radio signals of correct frequencies have no problem with that.

Think about it this way: the roads themselves are very expensive. But they make each car much cheaper. Most cars don't have to be equipped to drive over rocks and roots (we have tarmac), or lakes/ponds/rivers (we have bridges). We already spend a ton of taxpayer money to make it so that a 15 inch wheel with a mostly smooth tire and 5" of suspension travel is all a car needs. Why not install transceivers to make cars not require to have a super complicated LIDAR system?

LiDAR is just one small sensor fusion component in self-driving cars. There are up to 6 LiDARs, up to 4 radars, many cameras and sonic sensors that provide a stream of data for SLAM. The algorithms we have for perception right now are sufficient for 99.9% situations already, although not in real-time with the current technology. Once we surpass 200Hz, we should be fine. On the other hand, path planning is still in the stone age and there is not much hope it'd get better anytime soon (full of NP-Hard problems).

Last I checked, the limiting factor was the fact that they couldn't build a complex enough LIDAR that was both small and cheap.

That has no effect on perception quality, just on economy & production. The problem is "solved" research-wise, it's just about miniaturization, speeding it up etc. in evolutionary way as semiconductors and materials improve. Path planning is on another orbit, not solved even in theory...

I've been saying this for a couple years now. Naysayers don't fully understand the problem. There is no safe way to do self driving on our current roads in all climates without at least some infrastructure enhancements. Pretty much any of those enhancements you can imagine are orders of magnitude cheaper than building dedicated roads, which is the only other sufficiently safe option. The moment we start trying to take the driver out of the equation for real en masse, this will become obvious to everyone.

That is the "boil the ocean" solution to the problem.

Basically, "update all roads on the planet".

$$$. If you put the cost on taxpayers it will never be approved. Needs to be privately funded.

Yes, let's have the Ford highway that only Ford vehicle can drive on because the transponders are proprietary. We don't have privately painted lane markings, why would we have privately installed radio lane markings?

Would rather not have a privately nor publicly funded and owned physical panopticon. Too much room for control and abuse.

Maybe even separate long distance roads that only autonomous cars are allowed on.

Given a LiDAR of a sufficient range, freeway driving is a solved problem. Google did it in 2013.

And yet I don't see a single self driving car for sale.

Given sufficient energy we can create anything because E = mc^2, but we have yet to actually produce enough energy to create whatever matter we want.

Ice and snow are real conditions outside of the Bay Area.

Edit: Downvoters, please explain yourselves. If you have evidence that ice and snow have been solved for level 5 I’d love to see it.

LIDAR is a dead end

I think Elon Musk is basically correct when he made that statement. It probably wouldn't change the way current cars drive.

He was also right because, I believe the LIDAR units on say a waymo car probably cost more than an entire tesla vehicle.

But I couldn't help thinking if one of these LIDAR startups comes out with an "automotive pricing cheap" chip, it would be a great sensor fusion thing to add.

Reply on news.ycombinator.com