EEVblog Electronics Community Forum
General => General Technical Chat => Topic started by: BU508A on February 18, 2020, 11:39:17 am
-
Hi,
looks like, that the automobile industries can learn one or two things from Tesla.
Money quote from the article linked below:
"What stands out most is Tesla's integrated central control unit, or "full self-driving computer." Also known as Hardware 3, this little piece of tech is the company's biggest weapon in the burgeoning EV market. It could end the auto industry supply chain as we know it."
https://asia.nikkei.com/Business/Automobiles/Tesla-teardown-finds-electronics-6-years-ahead-of-Toyota-and-VW2 (https://asia.nikkei.com/Business/Automobiles/Tesla-teardown-finds-electronics-6-years-ahead-of-Toyota-and-VW2)
Interesting times are ahead, I think.
-
Are the AI chips Tesla designed ASICs or some off the shelf chip that has been remarked to obscure its identity?
-
Are the AI chips Tesla designed ASICs or some off the shelf chip that has been remarked to obscure its identity?
It's a completely custom chip designed from the ground up.
-
This custom hardware is one of the most interesting things about Tesla.
-
(https://cleantechnica.com/files/2019/04/tesla-vision-neural-net-hardware-fsd-ap-autopilot-full-self-driving.png)
-
Are the AI chips Tesla designed ASICs or some off the shelf chip that has been remarked to obscure its identity?
ASIC designed by an in-house Tesla team, diffused in Austin, TX by Samsung Semiconductor.
It's the largest neural-net processor out there in volume production (Intel is due to launch a general purpose AI accelerator next year.) Essentially each device is about 5x faster than the GTX1060 of the prior generation.
There are two independent ASICs per car, with independent power supply, clock, communication and camera feeds, which provides redundancy for full self driving (if/when that is achieved.)
-
While it is impressive, I'm not sure that this is 100% necessary or adds a lot of actual value.
If you look at what aomma.ai can do with a single camera and essentially a cell phone processor, I'm not sure all that dedicated AI hardware is all that necessary.
-
While it is impressive, I'm not sure that this is 100% necessary or adds a lot of actual value.
If you look at what aomma.ai can do with a single camera and essentially a cell phone processor, I'm not sure all that dedicated AI hardware is all that necessary.
As I understand it, the Tesla solution uses a lot less power than something implemented on general purpose hardware would do.
-
But who cares? That general purpose processor uses minuscule amount of power compared to motors and probably even headlights. Sure, saving power gives you an edge, but investing money into custom silicon is probably not the greatest idea.
Unless you know that car business is not sustainable and you are really in the business of selling those chips. Which is fine, other car vendors would just buy them, the same way they buy processors now.
-
But who cares? That general purpose processor uses minuscule amount of power compared to motors and probably even headlights. Sure, saving power gives you an edge, but investing money into custom silicon is probably not the greatest idea.
Unless you know that car business is not sustainable and you are really in the business of selling those chips. Which is fine, other car vendors would just buy them, the same way they buy processors now.
I can't remember which Tesla presentation I saw it in, but the power saving is significant. Even the efficient Tesla processor uses several hundred watts when "thinking hard". It makes a big difference on an electric car, apparently.
-
comma.ai uses a cell phone. It does not consume that much that you would notice. So I'm not sure what it is thinking hard about.
-
I think the reasoning behind designing their own processor is covered in this presentation:
https://www.youtube.com/watch?v=Ucp0TTmvqOE (https://www.youtube.com/watch?v=Ucp0TTmvqOE)
-
If you look at what aomma.ai can do with a single camera and essentially a cell phone processor, I'm not sure all that dedicated AI hardware is all that necessary.
Except it often tries to drive you off the road. Driving capability and doing it reliably are 2 very different things. Also don't forget this hardware is meant for FULL self driving.
-
Honnestly? Not sure I care... Self driving capability is not on my list of things I look for in a car. Especially not with an "agile" approach to SW development..
-
It's meant for full self driving but so was the last hardware. I really think it depends on whether it's needed though. Depending on how many cameras they rely on it may be best to go overboard and leave some power for overhead/future sensing capabilities.
-
That's a very incompetent article.
The traditional Carmakers usually never design electronics by themselves (any more), they define the architecture in the car, and order the electronics from big suppliers, like Denso, Bosch, Valeo, Continental, and so on.
Their processing units (µP) in the different modules are of course automotive specific designs to fulfill AEC Q100, for example, and are tailored to the required features, performance and price. Very rarely commercial off-the-shelf processors are used.
There's a transformation already going on in the Automotive Industry, one part being to merge electronics into Silver Boxes.
So why should one assume, that the traditional suppliers won't be able also to design such electronics, or not already offering that?
It depends only on what the OEMs want to have.
Merging too many functions into one box only is a bad idea, like Infotainment and Autonomous Driving, because you have too many tasks in parallel, which will endanger the safety functions, and such a central system for everything wiould get too complex, SW and HW wise.
And also TESLA did not reach AD level 5, as these many deadly accidents of TESLA cars prove.. so why this worshiping of this PCB, but ignoring the lack of sufficient AI power?
Frank
-
comma.ai uses a cell phone. It does not consume that much that you would notice. So I'm not sure what it is thinking hard about.
Comma.ai is not a redundant setup. It uses a single camera, and relies on the phone's software not crashing and the non-ECC RAM functioning correctly in all circumstances.
This is fine while you have a human in a loop but can never achieve full self driving (with a distracted or completely absent human) because it could not handle a software/hardware crash or non-ECC memory with a bit flip, etc.
-
Comma.ai is not a redundant setup.
Yes, sure. But it shows just how much is possible using a single processor and a simple camera. Putting 5 of those processors in a car is cheaper and probably better than designing a crazy AI ASIC.
-
Merging too many functions into one box only is a bad idea, like Infotainment and Autonomous Driving, because you have too many tasks in parallel, which will endanger the safety functions, and such a central system for everything wiould get too complex, SW and HW wise.
Modern SoCs support virtualisation. Look at NXP's iMX8 for example which also targets automotive markets. Using virtualisation it can -for example- drive both the instrument cluster and infotainment system as if it is two entirely seperate systems.
Comma.ai is not a redundant setup.
Yes, sure. But it shows just how much is possible using a single processor and a simple camera. Putting 5 of those processors in a car is cheaper and probably better than designing a crazy AI ASIC.
I have a feeling that getting the few % extra right for a self driving vehicle (especially poor visibility conditions) needs an exponential amount of extra processing power. I don't think that a simple setup using a smartphone is very representative.
-
The traditional Carmakers usually never design electronics by themselves (any more), they define the architecture in the car, and order the electronics from big suppliers, like Denso, Bosch, Valeo, Continental, and so on.
Which is exactly their weak point. Too many layers. You can't easily reuse talent or knowledge because you don't own it.
EaaS (engineering as a service) is a lifetime subscription, not building much value over time.
Merging too many functions into one box only is a bad idea, like Infotainment and Autonomous Driving, because you have too many tasks in parallel, which will endanger the safety functions, and such a central system for everything wiould get too complex, SW and HW wise.
Based on the eMMC bug their infotainment is some other system or at least processor or card. It may be sandwiched against this. That I do not know.
-
If you look at what aomma.ai can do with a single camera and essentially a cell phone processor, I'm not sure all that dedicated AI hardware is all that necessary.
Except it often tries to drive you off the road. Driving capability and doing it reliably are 2 very different things. Also don't forget this hardware is meant for FULL self driving.
The telsa hardware likes to drive you into the back of emergency vehicles.
I don't really see the point of gushing over "FSD" hardware that doesn't seem anywhere near being ready for FSD.
-
The traditional Carmakers usually never design electronics by themselves (any more), they define the architecture in the car, and order the electronics from big suppliers, like Denso, Bosch, Valeo, Continental, and so on.
Which is exactly their weak point. Too many layers. You can't easily reuse talent or knowledge because you don't own it.
EaaS (engineering as a service) is a lifetime subscription, not building much value over time.
Denso once belonged to Toyota, Visteon to FORD and Delphi to GM.
It is and has always been a make-or-buy decision.
Obviously, in a long process, these OEMs decided that it's more cost efficient to outsource the engineering, losing competency and maybe profit margin and also being more dependent on suppliers.
But anyhow, all these traditional car makers are profitable, but TESLA is not (yet). TESLA for sure is only at the beginning of such a decision process.
Maybe this situation might change for the others also, as the classical combustion engine - powertrain might be obsolete in the future, replaced by even more electronic / electrical components.
Frank
-
But who cares? That general purpose processor uses minuscule amount of power compared to motors and probably even headlights. Sure, saving power gives you an edge, but investing money into custom silicon is probably not the greatest idea.
It's not the power consumption they're afraid of, it's the power dissipation that goes with it. Automotive electronics need to work under very high temperatures. They're fighting for every Watt they can avoid to not increase their problem.
-
I have no experience with Tesla cars, but what is wrong with old designs? Just because a design is newer, doesn’t make it better. Look at the 3458A for example; if it ain’t broke don’t fix it.
I would have liked to see actual criticism of specifically what makes the Tesla system better, not just “Tesla future, Toyota old and smelly”. I still have no idea which is better after reading that article :-//
-
I have no experience with Tesla cars, but what is wrong with old designs? Just because a design is newer, doesn’t make it better. Look at the 3458A for example; if it ain’t broke don’t fix it.
I would have liked to see actual criticism of specifically what makes the Tesla system better, not just “Tesla future, Toyota old and smelly”. I still have no idea which is better after reading that article :-//
My understanding of the article is not, which is better, Tesla or Toyota / VW.
My understanding is: what makes this to the supply chain for the companies which produces automobiles and their suppliers. In Germany, lots of jobs are in the supplier sector for the automobile branch. If those suppliers aren't needed any more, this would have a huge impact to the economy, not alone in Germany.
-
Yes, I agree with your assessment of the article. I guess my confusion is better stated as: if the German way (or Japanese way) produces a more reliable system, then why shouldn’t we stick with that? I think the author was hoping that people would just assume that having more people involved is a less efficient way to produce an ECU, but unless you actually compare the new ECUs against the existing ones, how can we actually know this is the case?
-
I think, since these ECUs are very complex systems, there is only one way to find out: it must be done and see what happens. Not sure, if I'm very happy with this way.
-
I have no experience with Tesla cars, but what is wrong with old designs? Just because a design is newer, doesn’t make it better. Look at the 3458A for example; if it ain’t broke don’t fix it.
I would have liked to see actual criticism of specifically what makes the Tesla system better, not just “Tesla future, Toyota old and smelly”. I still have no idea which is better after reading that article :-//
My understanding of the article is not, which is better, Tesla or Toyota / VW.
My understanding is: what makes this to the supply chain for the companies which produces automobiles and their suppliers. In Germany, lots of jobs are in the supplier sector for the automobile branch. If those suppliers aren't needed any more, this would have a huge impact to the economy, not alone in Germany.
I think you have to look at the supply chain differently. In the end the supply chain has to deliver what the car manufacturers require. You also have to ask yourself: how much development needs a car manufacturer do themselves and what can they outsource? If they outsource they can spread the development costs of common components which don't add any value to the brand (like ABS pumps, windshield motors, radiators, etc) over much more cars than the manufacturer makes themselve. And it is not like the suppliers are sitting on their asses. They are actively developing new technologies too.
-
Comma.ai is not a redundant setup.
Yes, sure. But it shows just how much is possible using a single processor and a simple camera. Putting 5 of those processors in a car is cheaper and probably better than designing a crazy AI ASIC.
I'm 100.00% sure whatever "ai" self-driving system with a single camera and mobile phone CPU will fail miserably to provide anything else than technological demonstrations.
There is a long, long way from a demo or even test drives, to a reliable and safe self-driving car. The difference between good conditions and bad conditions is easily 3-4 orders of magnitude in complexity, it's a massive showstopper for most startups on this field.
In more demanding conditions, even human brain with 90 billion neurons and 1000 megapixel stereo vision sometimes have severe difficulties processing where the road actually is. It needs a combination of very sophisticated sensing and very complex processing, including "AI-like" parts like neural networks, but also static algorithms.
I strongly suspect even Tesla, with their custom processing core and a dozen of cameras, will still be a long way from reaching reliable fully self-driving system. They are far from having "too much" processing power, likely still the opposite! I think going custom ASIC for processing is likely exactly the right direction. There will be practical limits in power consumption and parallelization, no one's going to put 5kW worth of server PCs or graphics cards in their cars then spend time synchronizing parallel processing. If Tesla can achieve the same processing power in a few hundred watts, they are years ahead on the processing hardware side. This "version 1" likely isn't what actually brings us fully autonomous cars, it could be version 2 or version 3; maybe it's in 2030's; but if no one else is solving the processing problem, then Tesla is likely the one who gets the fully working system first.
This said, processing hardware isn't everything. I also strongly suspect Tesla, at some point, will admit they need to throw some actual distance-measurement hardware, lidar, even radar, at the problem, even if they currently seem to give the impression they are only working with standard 2D multicameras. Stereo cameras can go a long way, but augmenting it with actual distance measurements will provide a lot of confidence in difficult corner case conditions, and for Tesla, developing a low-cost solid-state LIDAR in co-operation with some innovative player in photonics field is not unlikely at all, even though if they don't publicly talk about it now.
For example, a stereo camera vision will fail miserably to provide any sense of 3D shape with fresh, smooth snow, in uniformly lit cloudy day. I have actually witnessed a human drive directly in a 1.5 meter deep ditch, nearly at walking speed, simply because everything in our 2D vision is pure smooth white with no texture; human brain has hard time figuring out how to stereo image it, I'm sure Tesla will have hard time as well. OTOH, in similar conditions, a 3DTOF camera worth of $20 components does great job producing an accurate (within 10cm) point cloud from the whole site.
Of course, even a LIDAR wouldn't see what's inside/below the snow... Which is again why processing/AI is so important. And you can't do it with a few hundred neurons. An ant cannot drive a car, nor can your dog. We need some human-level intelligence here, in worst cases.
-
That's a very incompetent article.
Its what passes for journalism in 2020.
The traditional Carmakers usually never design electronics by themselves (any more), they define the architecture in the car, and order the electronics from big suppliers, like Denso, Bosch, Valeo, Continental, and so on.
That's not really true. Some of the biggest car makers are more integrated than appears at first sight. Does Toyota make its own gearboxes? Kinda, because they are the majority shareholder in Aisin. Does Toyota make its own electronics? Kinda, because they own a big chunk of Denso.
Their processing units (µP) in the different modules are of course automotive specific designs to fulfill AEC Q100, for example, and are tailored to the required features, performance and price. Very rarely commercial off-the-shelf processors are used.
Most car makers partition their electronics so the most difficult things to get AEC Q100 qualification for are outside the safety related areas, and don't need it. AEC Q100 is a difficult market to be in. It demands a lot from the vendor, yet prices are very low. Excessive integration works against the car makers.
-
I have no experience with Tesla cars, but what is wrong with old designs? Just because a design is newer, doesn’t make it better. Look at the 3458A for example; if it ain’t broke don’t fix it.
I would have liked to see actual criticism of specifically what makes the Tesla system better, not just “Tesla future, Toyota old and smelly”. I still have no idea which is better after reading that article :-//
My understanding of the article is not, which is better, Tesla or Toyota / VW.
My understanding is: what makes this to the supply chain for the companies which produces automobiles and their suppliers. In Germany, lots of jobs are in the supplier sector for the automobile branch. If those suppliers aren't needed any more, this would have a huge impact to the economy, not alone in Germany.
The work that needs to be done won't disappear, though. The workers would simply be working for another company.
For commodity products, the shared supplier model can be the most efficient. For example, Ford and GM share a transmission (gearbox) design these days... it is cheaper for both companies and they are not competing on gearboxes.
-
The main difference between Tesla and the comma.ai solution is the ability to drive in NOT IDEAL CONDITIONS. Poor visibility and any type of obstruction can affect the performance of OpenPilot. With just a front camera and maybe the backup camera, it cannot make line changes unless the driver confirms it is perfectly safe. OpenPilot is nothing more than a smart cruise control and it will never be able to self-drive. So $999 for a bunch of wires and the Panda device (STM32 chip) is a very expensive solution for a cruise control.
-
The work that needs to be done won't disappear, though. The workers would simply be working for another company.
Not the same workers, I'm afraid. Maybe not even workers in the same country, or continent.
German automobile industry definitely needs to keep pace with global technology trends in order to be profitable.
-
Tesla has Level 2 self driving, and with current hardware they probably could get level 3.
Audi A8 has certified level 3 self driving, and they are working on level 4. Level 4 has geofencing, most of the time it can do everything and you can take over if you feel like.
The hardware costs are like a used VW Golf in the Tesla. I guess most car makers are not going to invest this much on experimental tech on production cars, because customers want something affordable. It is not the question of "can I do this", more like "does it make sense to do?"
-
[...] It is not the question of "can I do this", more like "does it make sense to do?"
A lot of research into advanced technologies may only become financially viable many years from now...
-
Audi A8 has certified level 3 self driving, and they are working on level 4. Level 4 has geofencing, most of the time it can do everything and you can take over if you feel like.
And it works in their yard, maybe. Since they never bothered to release it to the public. And even a few videos where it is demonstrated by their employees show it's a joke.
-
I agree totally with Siwastaja above.
In a presentation on electric vehicles I went to recently, the guy (who is big in this field) said that nobody in the industry has any idea how to make a self driving vehicle which can do what human drivers can do in good conditions.
Even if they had 1000x the CPU power they would not know how to do it.
And then you get less than favourable conditions, bad light, snow, etc. Any skiers here? Try skiing in the shadow of a hill, if the snow is smooth. It takes a lot of cues. Even parking in a supermarket car park is orders of magnitude beyond today's software expertise.
-
Audi A8 has certified level 3 self driving, and they are working on level 4. Level 4 has geofencing, most of the time it can do everything and you can take over if you feel like.
And it works in their yard, maybe. Since they never bothered to release it to the public. And even a few videos where it is demonstrated by their employees show it's a joke.
Heh, Teslas level 2 self driving demonstrably doesn't work either and it's on public roads. :-DD
-
We may argue here until hell freezes over, but the fact is that AI, specifically AI applied to the automotive sector, will be the next technology differentiator which separates the prosperous from the struggling countries.
With the German auto sector -for the time being- still considered the world's premiere, and such an important part of the German economy, it would be darn foolish not to invest in AI research focused to the automotive sector.
There was even an article on the Deutsche Welle discussing this topic. It focuses on AI as a whole, not only automotive AI.
https://www.dw.com/en/skeptical-germany-lags-behind-on-artificial-intelligence/a-51828604 (https://www.dw.com/en/skeptical-germany-lags-behind-on-artificial-intelligence/a-51828604)
Whoever develops an automotive grade AI "standard", will lead the world's industry for the next decades.
As an example of automotive technological leadership, just look at CANBUS.
-
Audi A8 has certified level 3 self driving, and they are working on level 4. Level 4 has geofencing, most of the time it can do everything and you can take over if you feel like.
And it works in their yard, maybe. Since they never bothered to release it to the public. And even a few videos where it is demonstrated by their employees show it's a joke.
Heh, Teslas level 2 self driving demonstrably doesn't work either and it's on public roads. :-DD
https://www.youtube.com/watch?v=oK5_lRTUato&feature=emb_title (https://www.youtube.com/watch?v=oK5_lRTUato&feature=emb_title)
-
We may argue here until hell freezes over, but the fact is that AI, specifically AI applied to the automotive sector, will be the next technology differentiator which separates the prosperous from the struggling countries.
With the German auto sector -for the time being- still considered the world's premiere, and such an important part of the German economy, it would be darn foolish not to invest in AI research focused to the automotive sector.
But what problem should AI solve? It reminds me of an example someone told me years ago. Researchers spend decades on developing a robotic hand which can lift an egg without crushing it. Meanwhile in the egg-packaging plants they use robots with vacuum suction cups to lift the eggs and put them in boxes at an incredible speed. Most self driving systems seems to be aimed at using visual information to move around in traffic. There are much better approaches to make self driving cars. The more general direction the car industry is going is to make vehicles communicate between each other and with the infrastructure. If cars can simply tell each other where they are and traffic lights tell cars it is red or green (and this technology is already being deployed!) the need for processing visual information through AI is much less necessary. In the end all a car would need is to be able to detect non-communicating objects. But not to a high degree because information about such objects is also shared between the cars.
-
If cars can simply tell each other where they are and traffic lights tell cars it is red or green (and this technology is already being deployed!) the need for processing visual information through AI is much less necessary. In the end all a car would need is to be able to detect non-communicating objects. But not to a high degree because information about such objects is also shared between the cars.
And you just need to replace all infrastructure and keep 100% of pedestrians and usual cars away from the roads. Not to say have 100% accurate mapping everywhere.
-
Has it "learned" not to crash into firetrucks yet?
-
But what problem should AI solve? It reminds me of an example someone told me years ago. Researchers spend decades on developing a robotic hand which can lift an egg without crushing it. Meanwhile in the egg-packaging plants they use robots with vacuum suction cups to lift the eggs and put them in boxes at an incredible speed. Most self driving systems seems to be aimed at using visual information to move around in traffic. There are much better approaches to make self driving cars. The more general direction the car industry is going is to make vehicles communicate between each other and with the infrastructure. If cars can simply tell each other where they are and traffic lights tell cars it is red or green (and this technology is already being deployed!) the need for processing visual information through AI is much less necessary. In the end all a car would need is to be able to detect non-communicating objects. But not to a high degree because information about such objects is also shared between the cars.
I think, the AI is aimed to solve the last 0.1% situations, which weren't thought of. The unexpected things which are going to happen and
which can be easily solved by a human with all his knowledge about the world and the rulesets he is following in a flexible way and his experience.
This is a very simple example imho:
https://www.youtube.com/watch?v=jV51BGIzkwU (https://www.youtube.com/watch?v=jV51BGIzkwU)
-
If cars can simply tell each other where they are and traffic lights tell cars it is red or green (and this technology is already being deployed!) the need for processing visual information through AI is much less necessary. In the end all a car would need is to be able to detect non-communicating objects. But not to a high degree because information about such objects is also shared between the cars.
And you just need to replace all infrastructure and keep 100% of pedestrians away from the roads. Not to say have 100% accurate mapping everywhere.
You'd be surprised how well governments keep track on sizes and location of roads. How else can they be managed? Currently GPS is useless for self driving cars but there are terrestrial based location systems at the horizon. For sure you can always find an uncharted dirt road but on such roads there is very little added value of having a self driving car anyway.
-
We may argue here until hell freezes over, but the fact is that AI, specifically AI applied to the automotive sector, will be the next technology differentiator which separates the prosperous from the struggling countries.
It sells Teslas for the moment and other companies will try to follow their example, but the emperor has no clothes. At some point there will be some government employee in between the next stopped car and the Tesla crashing into it and then it will all come tumbling down.
If you let the auto drive without human intervention for 99.9% of the time, the human won't be paying attention for the 0.1% ... it's as simple as that.
-
Audi A8 has certified level 3 self driving, and they are working on level 4. Level 4 has geofencing, most of the time it can do everything and you can take over if you feel like.
And it works in their yard, maybe. Since they never bothered to release it to the public. And even a few videos where it is demonstrated by their employees show it's a joke.
Heh, Teslas level 2 self driving demonstrably doesn't work either and it's on public roads. :-DD
https://www.youtube.com/watch?v=oK5_lRTUato&feature=emb_title (https://www.youtube.com/watch?v=oK5_lRTUato&feature=emb_title)
That guy is seriously delusional. He keeps telling the car to change lanes and change speeds, including a key lane change to get it prepared to simplify an automated exit from the highway, and then tells us he's not giving the car any input at all. The car does many stupid things during that journey, especially the multiple unprovoked lane changes in the latter part. If you've never seen a car do any automated driving its cool to watch, but its so far away from safe automated driving that Tesla's claims to have automation feature complete by now (which is what they claimed a year ago) are a joke.
-
especially the multiple unprovoked lane changes in the latter part.
I guess you missed that he set lane change to "Mad Max" mode.
-
especially the multiple unprovoked lane changes in the latter part.
I guess you missed that he set lane change to "Mad Max" mode.
Yes, but the car changed lanes with no reason, with no cars around and nothing slowing it down that would prompt it to overtake...
-
especially the multiple unprovoked lane changes in the latter part.
I guess you missed that he set lane change to "Mad Max" mode.
Yes, but the car changed lanes with no reason, with no cars around and nothing slowing it down that would prompt it to overtake...
That's exactly what Mad Max would do... :-DD
-
If you let the auto drive without human intervention for 99.9% of the time, the human won't be paying attention for the 0.1% ... it's as simple as that.
Yep.
The only way fully automated vehicles, without any human driver, will become a full replacement and even achieve better safety figures overall than human-driven vehicles (which is what they promise us), IMO, is if ALL the infrastructures become exactly adapted to them and there is no human-driven vehicle (or pedestrian, or ...) on the roads anymore. Not a single one. (I think it's been said already in some above posts.) Even though bugs are always possible (and even relatively common), machines do operate much better with other machines than with humans or anything unpredictable when it comes to safety.
Once we get there, most of the "AI" that we're currently working on for this will become pointless. If the only vehicles in circulation are all interconnected and all with compatible routing characteristics, avoidance will be a much simpler task.
And I kind of think at this point that this is exactly where people/companies working on and advocating autonomous vehicles are leading us to. A world, in fact, where only machines will be allowed, in the end. We will be just passengers, until we are not even needed anymore (that last part is a bit sci-fi, but the fictitious character is debatable.) The sophisticated obstacle avoidance AI stuff (and dealing with uncertainty in general) is entertaining a lot of people these days, but I'm not sure this has any future as is.
-
I've been saying this for a long time. The concept of a self driving car that requires occasional human intervention is fundamentally flawed because as soon as you take away the requirement to focus the brain finds other things to focus on.
Heck if I were king I'd ban automatic transmissions, can't stand them. I think if everyone had to actually drive the car we'd have a lot more attentive and engaged drivers. Additionally it would virtually eliminate all these "driver mistook the gas for the brake and plowed into a building" accidents. With a proper manual gearbox if you mash any two pedals the car *will* eventually stop and it's virtually impossible to unintentionally accelerate from a stop.
We've already had vehicles for decades that you don't have to drive yourself, they're called buses and taxis. If I can't even drive it then I see little appeal in owning a private car.
-
I've been saying this for a long time. The concept of a self driving car that requires occasional human intervention is fundamentally flawed because as soon as you take away the requirement to focus the brain finds other things to focus on.
Its not just a matter of attention. If people hardly ever drive their driving skills will degenerate (or for young drivers never develop). If you really automate driving (rather than just having the current driver assistance systems) you need to automate it entirely, because you won't be able to rely on a competent driver being present.
-
Yes that too.
What we need is *less* automation and more involvement, not more. All these driver assistance technologies are crutches that enable people to do things other than drive while they are driving a car. This creates positive feedback and people become worse, less attentive and less practiced drivers.
Driving is a privilege and it's a task that demands 100% focus of the person doing it. If a person doesn't want to drive or can't pay attention they should ride the bus, it's a far more environmentally friendly mode of transit than a private car anyway. At least this assistance nonsense hasn't spread to motorcycles and bicycles yet although the roads are ever more treacherous for those of us who choose those vehicles.
-
Yes that too.
What we need is *less* automation and more involvement, not more. All these driver assistance technologies are crutches that enable people to do things other than drive while they are driving a car. This creates positive feedback and people become worse, less attentive and less practiced drivers.
Driving is a privilege and it's a task that demands 100% focus of the person doing it. If a person doesn't want to drive or can't pay attention they should ride the bus, it's a far more environmentally friendly mode of transit than a private car anyway. At least this assistance nonsense hasn't spread to motorcycles and bicycles yet although the roads are ever more treacherous for those of us who choose those vehicles.
That is our view, as drivers who have come accustomed to owning a car and mastering it. But the landscape is changing. For many young people owning a car is no longer a desirable goal, especially in the larger cities it's more of a burden than an advantage. Together with more and more people striving to live in cities this creates a trend automobile companies are reacting to.
Also, there are countries where driving a car is a most boring activity and people engage in all kinds of distractions because there's just nothing else to do. If assistance technologies are keeping them and others safe I'm all for it. Myself as father of two small kids, I really appreciate automatic cruise control keeping a safe distance from the vehicle in front while the kids in the back are having a shouting match or creating whatever form of distraction they fancy this second.
PS: I hear automatic gearboxes are the new thing with motorcycles now.
-
I don't think there's a problem with assist. Drifting lanes? Sound an alarm and only then try to nudge the car back in lane. Trying to switch lanes with rear or side traffic incoming? Sound an alarm and try to nudge the car back in lane. Etcetera, etcetera.
If you make the alarm loud and annoying enough, people will not try to make the assist a standard feature of their driving style.
PS. thinkfat, doppler radar doesn't see stopped cars ... you're accepting the increased chance of killing or maiming yourself, your kids AND others by watching away. It's a dangerous gamble you are making for convenience.
-
I recommend driving on Madeira. Straight flat road is unknown there. I'd like to know how Tesla handles those roads.
-
doppler radar doesn't see stopped cars
Wouldn’t you get a Doppler shift on the carrier as your own car would be moving? I mean yeah, if your car is stopped and the other car is stopped there will be nothing, but I don’t think that’s a dangerous situation...
-
I don't think there's a problem with assist. Drifting lanes? Sound an alarm and only then try to nudge the car back in lane. Trying to switch lanes with rear or side traffic incoming? Sound an alarm and try to nudge the car back in lane. Etcetera, etcetera.
If you make the alarm loud and annoying enough, people will not try to make the assist a standard feature of their driving style.
PS. thinkfat, doppler radar doesn't see stopped cars ... you're accepting the increased chance of killing or maiming yourself, your kids AND others by watching away. It's a dangerous gamble you are making for convenience.
Doppler radar will "see" stopped cars because your own car is a moving reference system. The automatic cruise control will possibly choose to ignore it because it's an implausible signal, but the collision avoidance system will not. You'll get a collision warning that will definitely draw your attention back to the street. Also, it's not just Doppler, AFAIK there is a ranging radar as well. I remember seeing a teardown on TheSignalPath which involved a vehicle radar from an undisclosed manufacturer.
Also, it's not about voluntarily watching away. Distractions happen, they're unavoidable. We're all just humans, our attention span and ability to focus on boring things is very limited.
-
doppler radar doesn't see stopped cars
Wouldn’t you get a Doppler shift on the carrier as your own car would be moving? I mean yeah, if your car is stopped and the other car is stopped there will be nothing, but I don’t think that’s a dangerous situation...
A drawback with the wide beam radars on current cars is they can't accurately determine the resolved velocity of detected objects. With a doppler sensing radar its easy to correct for the motion of the vehicle down the boresight of the radar. However, when the beam is wide the edges of the beam need a very different correction from the boresight.
-
The resolution needed to ignore the road and all other static objects at long range is well beyond the current state of the art of car radar systems. Couple firetrucks and a police car disagree with the collision avoidance being much use at highway speeds on Teslas. Also the concrete road divider.
When people look away from the road they need a little adrenaline when they look back and they drifted lanes or see an upcoming obstacle ... the near misses keep you honest. That's why every automated intervention needs to give that same Pavlovian training. It can not be allowed to be convenient, it needs to be annoying. At least until they learn to not run into firetrucks.
-
If you look at what aomma.ai can do with a single camera and essentially a cell phone processor, I'm not sure all that dedicated AI hardware is all that necessary.
Custom hardware might be required to address long term availability. High performance cell phone processors are particularly bad about this.
Tesla might also want to avoid leaving their supply chain vulnerable to competitors. The major car companies have a history of quashing competition by buying their supply chain.
-
I thought there would at least be a little beamforming and directing involved, with a phased antenna array. I can't believe it'd work otherwise. Imagine overtaking a truck while the road turns left. If the radar was always looking straight ahead, you'd get a collision warning. Yet, that is not what is happening. The radar seem to look along the road.
-
I thought there would at least be a little beamforming and directing involved, with a phased antenna array. I can't believe it'd work otherwise. Imagine overtaking a truck while the road turns left. If the radar was always looking straight ahead, you'd get a collision warning. Yet, that is not what is happening. The radar seem to look along the road.
If you look at the specs for most of the car radars they have two transmissions, working concurrently in different bands, with different beam widths. A fairly wide beam, and a very wide beam. How would you expect to build a narrow beam scanning radar at consumer prices?
-
How would you expect to build a narrow beam scanning radar at consumer prices?
How about we just give up radar and time of flight? Make 60 GHz cameras with a rectenna array and a metamaterial lens, plus a continuous 60 GHz beam for illumination.
Naive ToF systems are sensitive to interference and complicated systems are hard to do at high resolution. A camera though has a simple structure and with stereoscopy can see depth just fine.
-
How would you expect to build a narrow beam scanning radar at consumer prices?
How about we just give up radar and time of flight? Make 60 GHz cameras with a rectenna array and a metamaterial lens, plus a continuous 60 GHz beam for illumination.
Naive ToF systems are sensitive to interference and complicated systems are hard to do at high resolution. A camera though has a simple structure and with stereoscopy can see depth just fine.
I don't know if they use ToF or some variant of FM. My go to approach would be to transmit fairly wide band noise, and correlate the transmitted and received signals.
-
I don't know if they use ToF or some variant of FM. My go to approach would be to transmit fairly wide band noise, and correlate the transmitted and received signals.
My point was that to do that for a ton of pixels is hard ... the only alternative is to do it very fast and scan, but it's hard to get a tight beam with phased arrays and mechanical scanning is expensive and fragile.
A 60 GHz camera has a comparatively simple structure.
-
I thought there would at least be a little beamforming and directing involved, with a phased antenna array. I can't believe it'd work otherwise. Imagine overtaking a truck while the road turns left. If the radar was always looking straight ahead, you'd get a collision warning. Yet, that is not what is happening. The radar seem to look along the road.
If you look at the specs for most of the car radars they have two transmissions, working concurrently in different bands, with different beam widths. A fairly wide beam, and a very wide beam. How would you expect to build a narrow beam scanning radar at consumer prices?
Since the antennas are anyway a phased array (no space for something else), I can imagine that at least skewing the beam to "look" left and right, or/and the receiver pattern doing the same is possible. Anyway, you'd not have to create a narrow, scanning beam. It's enough to vaguely look in the direction of the turn to suppress unwanted reflections from targets not in you lane.