Robo-Cab Predictions

[apis]

I would like to see a self driving car in the streets of Rome. With kamikaze scooters cutting routinely in front of you in twisting roads.

For a Roman driver, he would simply honk the horn and hurl a pair of delicious Italian expletives, and he would continue driving.
For AI, it would trigger a scram reaction on the brakes.

I don't think what you describe is a big problem. The computer has much better control of the car than an average human driver, so narrow twisty roads are less of a challenge than for humans. They also have much better situational awareness than humans, a self driving car would spot the kamikaze scooters long before a human, predict their intended route and plan accordingly.

[EEVblog]

It's not the twisty crowded roads, it's the rain, hail and snow. The lidar is famously blinded by heavy precipitation.

I remember driving home one night from the city. It was raining hard and Sydney's streets were more than their usual nightmare with nighttime road works, huge flood lighting blinding everything, and contraflows galore that even I had a hard time figuring out. And I remember thinking that an autonomous car wasn't going to be able to handle this within my lifetime.
Maybe if it tailgated the car in front that would be a sensible AI tactic, but on it's own no freaking way.

IMHO you are falling in the same trap as many others: a self driving car can use different sensors so it can get more information than just the visual spectrum humans can detect.

Nope, I was thinking of that as well, a lot actually, and I stand by my statement.

[EEVblog]

I would like to see a self driving car in the streets of Rome. With kamikaze scooters cutting routinely in front of you in twisting roads.

For a Roman driver, he would simply honk the horn and hurl a pair of delicious Italian expletives, and he would continue driving.
For AI, it would trigger a scram reaction on the brakes.

I don't think what you describe is a big problem. The computer has much better control of the car than an average human driver, so narrow twisty roads are less of a challenge than for humans. They also have much better situational awareness than humans, a self driving car would spot the kamikaze scooters long before a human, predict their intended route and plan accordingly.

Now you are falling into the trap of assuming that just because some self driving system can do that kind of stuff in one (or many) situations, that it can do it in all. That's a huge mistake.

[apis]

But the Tesla isn't a self driving car

Perhaps you haven't been reading the news. Telsa say they will have 1 million fully autonomous cars on the road next year.
That's the whole point of this thread.

Yes, and I agree that is complete and utter bullshit. Tesla are giving self driving cars a bad reputation. So far Tesla haven't demonstrated anything near the same level of self driving the others are working on. Elon says every one else is stupid for using lidar and detailed maps, but unlike Tesla the others have actual self driving prototypes on the roads, averaging more than 17 thousand km between each human intervention. If Tesla have a fully autonomous car ready for release they have managed to keep it secret from everyone else.

That is why I keep talking about Waymo, they are the technology leaders and what people should be looking at when talking about self driving cars.

[nctnico]

But the Tesla isn't a self driving car

Perhaps you haven't been reading the news. Telsa say they will have 1 million fully autonomous cars on the road next year.
That's the whole point of this thread.

If the latter is the case then I like to point out Tesla is nearly broke and Musk tries to sell something which isn't there because they want to water down the shares (again) to raise money. In reality Tesla is nowhere near having a self driving car despite what Musk likes to think. If you want to have a discussion about self driving cars then Tesla shouldn't be part of it.

[apis]

If I were to guess, once a significant fraction of vehicles on the road are autonomous, there will be a push for integrating features into roads to help them navigate in a robust manner. Retro-reflectors for lidar or radar, inductive loop sensors or similar. This would make them safer to use during poor visibility conditions than unaided human drivers.

I think so too, if they start providing infrastructure to make it easier for autonomous cars many of the biggest challenges go away. For example, the cars use cameras and computer vision to look for traffic signs just like a human would, but it would be fairly trivial to have a radio beacon to signal the same information. Or you could have whoever is responsible for the traffic signs be required to update publicly available maps whenever they make a change (they probably do that already anyway). You could put iron spikes at regular intervals in the road to mark the lanes so the car can sense where the lane is even if the road is covered in snow. You could have special drop off and pick up points for autonomous vehicles in crowded places, there already exist such special taxi areas outside many airports and train stations.

But so far they have said they want to everything without the need for extra infrastructure.

[EEVblog]

If the latter is the case then I like to point out Tesla is nearly broke and Musk tries to sell something which isn't there because they want to water down the shares (again) to raise money. In reality Tesla is nowhere near having a self driving car despite what Musk likes to think. If you want to have a discussion about self driving cars then Tesla shouldn't be part of it.

But Tesla have cars on the road, lots of them.
No point talking about commercialization of others that don't have commercial cars on the market that people can buy. Sure Waymo are starting to offer a trial commercial service, and they claim it'll be for-profit, but I suspect it will be nowhere near profitable.
Call me when I can buy one at even a remotely affordable price.

[Mr. Scram]

Perhaps you haven't been reading the news. Telsa say they will have 1 million fully autonomous cars on the road next year.
That's the whole point of this thread.

The mistake probably lies in looking at it from an engineering point of view, rather than a marketing one. Musk isn't just responsible for making it happen, but also getting people on board.

[Mr. Scram]

If the latter is the case then I like to point out Tesla is nearly broke and Musk tries to sell something which isn't there because they want to water down the shares (again) to raise money. In reality Tesla is nowhere near having a self driving car despite what Musk likes to think. If you want to have a discussion about self driving cars then Tesla shouldn't be part of it.

How do you know what Musk thinks?

[EEVblog]

I think so too, if they start providing infrastructure to make it easier for autonomous cars many of the biggest challenges go away. For example, the cars use cameras and computer vision to look for traffic signs just like a human would, but it would be fairly trivial to have a radio beacon to signal the same information. Or you could have whoever is responsible for the traffic signs be required to update publicly available maps whenever they make a change (they probably do that already anyway).

LOL
Roads change every day due to maintenance, upgrades, and accidents. It's done by grunts in a reflector vest at a moments notice. No way it's even remotely possible to apply any sort of tech to that reliably.

You could have special drop off and pick up points for autonomous vehicles in crowded places, there already exist such special taxi areas outside many airports and train stations.

Now your talking, because autonomous cars are going to suck at special circumstances for a very long time to come.

But so far they have said they want to everything without the need for extra infrastructure.

Because they know that's the only way it's going to work in practice.

[EEVblog]

If the latter is the case then I like to point out Tesla is nearly broke and Musk tries to sell something which isn't there because they want to water down the shares (again) to raise money. In reality Tesla is nowhere near having a self driving car despite what Musk likes to think. If you want to have a discussion about self driving cars then Tesla shouldn't be part of it.

How do you know what Musk thinks?

Because he tweets a lot.

[Mr. Scram]

Because he tweets a lot.

That's just what he says. See my previous post about engineering versus marketing.

[apis]

Now you are falling into the trap of assuming that just because some self driving system can do that kind of stuff in one (or many) situations, that it can do it in all. That's a huge mistake.

It's my engineering opinion based on what I know and I've seen so far. It's hard to tell how far they have come, they are not exactly sharing data voluntarily. Any video they show where the car does something impressive is marketing and might be just a fluke, or it might be complete theatre. We won't really know until they launch. Waymo doesn't seem to be in a rush (which one can't fault them for, they are prioritising safety over money in this case.)

From a twisty road ten years ago:

[apis]

Because he tweets a lot.

That's just what he says. See my previous post about engineering versus marketing.

Musk talk (tweet) a lot. He also says he is going to terraform Mars and transport people across continents using self landing rockets. Otoh, he now has self landing rockets, something many believed was impossible.

https://youtu.be/xDEKjfnRhqQ?t=44

[Bud]

I have no reason not to believe him. He's been sending tourists to Mars since "by 2018". Right?

[Mr. Scram]

I have no reason not to believe him. He's been sending tourists to Mars since "by 2018". Right?

Correct. We haven't seen any of those self landing rockets or a Tesla in interplanetary space either.

[wilfred]

Waymo doesn't seem to be in a rush (which one can't fault them for, they are prioritising safety over money in this case.)

Now that's marketing. 

In actual fact the driving (yes, shameless pun) motivation for this robot cab is money. They don't want to pay drivers.

But robot or not no-one will like paying to sit in traffic.

People, feel free to list out the reasons autos are the future and where the benefits will come from. I don't follow the development of autos all that closely unless the mass media pick up a story but it's not self evident to me where the longer term benefits are coming from for ordinary people.

Every day I drive in traffic I see some situation and I think how will an auto deal with it. Simple things like today I had the choice to merge into the front of a tram or fall back and merge in behind it. I could have a driver who doesn't want to be behind a tram in the first place (like me) who doesn't want to make it easy and wave me in. It could be an aggressive driver who has already been stuck and definitely won't make it easy.

For those not familiar with trams you can't pass them on the other side of the road and you can't pass them when they are stopped to set down or pick up passengers, And you definitely can't pass them ona busy inner city road where cars are parked and there is no gap big enough in the parked cars for a kilometer. Autos in Melbourne will have to manage it or the fare paying passengers will be unhappy in both time and money. Any driver in Melbourne with any experience with traffic manages to work out the maneuver but I would say it is an advanced skill.

[EEVblog]

In actual fact the driving (yes, shameless pun) motivation for this robot cab is money. They don't want to pay drivers.

Too bad if the LIDAR systems cost more than driver does though 

[EEVblog]

Musk talk (tweet) a lot. He also says he is going to terraform Mars and transport people across continents using self landing rockets. Otoh, he now has self landing rockets, something many believed was impossible.

Err, only the grossly ill-informed believed it wasn't possible, because Space-X wasn't the first to do it. It was done decades before they did it.

BTW, the continent hopping rocket is as big of an impractical joke as Hyperloop is. Not everything Musk says is practical, some of it is demonstrably stupid.

[maginnovision]

I'm still waiting for Musk to announce his own solar roadway.

[magic]

Musk says Tesla will be out of money in 10 months without ‘hardcore’ changes

https://www.theverge.com/2019/5/17/18629166/elon-musk-tesla-money-changes-cfo-employee-expenses

[apis]

In actual fact the driving (yes, shameless pun) motivation for this robot cab is money. They don't want to pay drivers.

The driving motivation behind any company is to make money, that's just how the world is, that is a political issue not a technological. There are lots of people involved in developing autonomous vehicles though, and everyone's motivation might not be as cynical. The reality is that getting people out of the driving equation will save lives. The need for transportation won't go away, and as long as it doesn't there will be a need for drivers. Besides reducing the traffic accident rate you get all the normal benefits you get from automation, lower transportation cost and lots of people get more time over to do other things. If that leads to improvement for "ordinary people" or not (whoever that is) is once again a political issue.

Every day I drive in traffic I see some situation and I think how will an auto deal with it.

People used to say a chess program would never beat a grand champion, turns out you could do it in 1996 with a disappointingly trivial algoritm as long as you had enough computing power. Since then the algorithms got better and we got even more computing power and a chess program that beats the world champion runs on an ordinary desktop computer today. To beat Go you needed neural nets, and today it's trivial (if you know how) to beat any human in perfect information games. Neural nets were developed in the 70's but it's not until now it's practically usable, there just wasn't enough computing power before. There have been an exponential growth in computing power that has made things that where impossible a couple of decades ago feasible. I think even software developers have trouble keeping up, exponential change is just too overwhelming and unintuitive.

Musk talk (tweet) a lot. He also says he is going to terraform Mars and transport people across continents using self landing rockets. Otoh, he now has self landing rockets, something many believed was impossible.

Err, only the grossly ill-informed believed it wasn't possible, because Space-X wasn't the first to do it. It was done decades before they did it.

BTW, the continent hopping rocket is as big of an impractical joke as Hyperloop is. Not everything Musk says is practical, some of it is demonstrably stupid.

Yup, but a lot of people were grossly ill-informed. And yes, completely agree about the continent hopping rockets.

But it's probably Musk's bold optimism that makes people like him, and that is an important part of what makes him successful.

My point was that Musk isn't exactly careful with the truth though, even if he does some cool stuff, so maybe take what he says with a pinch (bucket) of salt.

[nctnico]

Musk says Tesla will be out of money in 10 months without ‘hardcore’ changes

https://www.theverge.com/2019/5/17/18629166/elon-musk-tesla-money-changes-cfo-employee-expenses

What isn't helping is that the tax incentives/credits are capped or being phased out in many important markets in which Tesla operates. This especially hurts sales of the more expensive models. In other words: Tesla needs to be able to make their cars profitable with a price below the $50k mark in order to survive.

[apis]

Musk says Tesla will be out of money in 10 months without ‘hardcore’ changes

https://www.theverge.com/2019/5/17/18629166/elon-musk-tesla-money-changes-cfo-employee-expenses

What isn't helping is that the tax incentives/credits are capped or being phased out in many important markets in which Tesla operates. This especially hurts sales of the more expensive models. In other words: Tesla needs to be able to make their cars profitable with a price below the $50k mark in order to survive.

Or maybe it's because they haven't paid enough attention to the body:
"The wheel well has nine parts"

[soldar]

There is actually a viral video of two people  doing exactly that on a Tesla on autopilot.

Thanks for the heads up! Found it at my favorite porn hub.