Donate to Remove ads

Got a credit card? use our Credit Card & Finance Calculators

Thanks to johnstevens77,Bhoddhisatva,scotia,Anonymous,Cornytiv34, for Donating to support the site

Musk endeavours

The Big Picture Place
onthemove
Lemon Slice
Posts: 540
Joined: June 24th, 2017, 4:03 pm
Has thanked: 722 times
Been thanked: 471 times

Re: Musk endeavours

#208157

Postby onthemove » March 17th, 2019, 10:29 am



Interesting article. This bit also stood out to me...

Worse still for Tesla is the growing number of experts calling out its approach to autonomy as fundamentally flawed due to a lack of LIDAR. As we have discussed previously, regulations will almost certainly mandate the use of LIDAR in the event of the legalization of autonomous vehicles. Tesla, which has eschewed that technology, could well find its technology made wholly obsolete before it even has a viable product.


I'd fully agree with that. I cannot see any regulator allowing self driving cars with cameras only. Even though in theory, computer vision (deep learning CNNs) can outperform humans at recognition tasks - and that's what's needed to identify roads, signs, people, cyclists, other vehicles and so on - I believe there is an expectation from the general public that autonomous cars will be a step up in safety compared to human drivers.

And certainly while the technology is still finding its feet, it seems pretty much a given that regulators are going to demand the kind of backup / redundancy from lidar, etc, complementing vision, particularly when such technologies already exist, are not massively expensive and have been proven to add value by others.

Correct me if I'm wrong, but I believe of the few fatalities so far in cars with either auto pilot, or fully autonomous, all were cars which didn't have lidar (or similar) to backup their cameras.

Lidar can't identify what an object is - that's why cameras with deep learning CNNs are the primary 'engine' of autonomous driving, but lidar can tell you if you are on course to hit a solid (but unidentified) object that the cameras didn't spot. Either a pedestrian walking their bike across the highway, or a lorry in an adjacent lane you're about to pull into, or a safety barrier you're about to collide with, etc.

With the cost of lidar so relatively cheap, and the clear additional safety benefits it provides, I wouldn't want to be sharing the road with autonomous cars that didn't have it. And with all the fear of the unknown / suspicion around self driving cars, I think the regulators are going to take a similar view.

tjh290633
Lemon Half
Posts: 8208
Joined: November 4th, 2016, 11:20 am
Has thanked: 913 times
Been thanked: 4096 times

Re: Musk endeavours

#208159

Postby tjh290633 » March 17th, 2019, 10:31 am

Don't aircraft require triple redundancy for automated systems?

TJH

dspp
Lemon Half
Posts: 5884
Joined: November 4th, 2016, 10:53 am
Has thanked: 5825 times
Been thanked: 2127 times

Re: Musk endeavours

#208162

Postby dspp » March 17th, 2019, 10:38 am

tjh290633 wrote:Don't aircraft require triple redundancy for automated systems?

TJH


No.

It depends what the basis of the certification is/was. For example the Boeing 737 has, at best, dual redundancy in many areas. The likely design flaw (tbd) that has caused all the 737-MAX (which is basically the third revision of the 737 design) is arguably only single redundant in the MCAS system (tbd), or at best dual redundant. The rudder failures in early 737 were if I recall correctly single redundant as well, but that got improved.

https://www.pprune.org/rumours-news/619 ... frica.html

In contrast the Airbus tend to be triple redundant as a minimum on primary systems, but one does need to read the smallprint.

I think the jury is still out on the necessity (or not) of Lidar in automated driving systems. Logic suggests it is not vital but practicalities may make it the most cost-effective pathway. Lidar certainly is not cheap at present. The combination of radar + wide-spectrum cameras, plus (slow) ultrasonics is better than us mere pink jobs have as sensors.

regards, dspp

odysseus2000
Lemon Half
Posts: 6364
Joined: November 8th, 2016, 11:33 pm
Has thanked: 1530 times
Been thanked: 959 times

Re: Musk endeavours

#208177

Postby odysseus2000 » March 17th, 2019, 12:05 pm

For those interested this is an account of some of the differences between Boeing & Airbus systems:

https://pdfs.semanticscholar.org/d4ac/1 ... 7117a9.pdf

It will be hugely indicative of future AI if non Lidar systems can not work better than humans. In the UK humans kill or seriously injure about 10 people per day with cars, massively more than airplane accidents & so the prize for anyone who can make a low cost non Lidar system is colossal.

If non Lidar systems can not be made to work it will be a huge tell for the future of AI & will likely send off the advent of mass robotic driving long into the future as Lidar systems at wavelengths that do not damage human eyes are expensive & if one needs Lidar one presumably needs redundant Lidar. It is not clear to me that Lidar systems as used by some robotic cars can be made low cost in mass production, but happy to be corrected if this is wrong.

Regards,

dspp
Lemon Half
Posts: 5884
Joined: November 4th, 2016, 10:53 am
Has thanked: 5825 times
Been thanked: 2127 times

Re: Musk endeavours

#208185

Postby dspp » March 17th, 2019, 12:38 pm

o2000,

That article describes the more recent Boeing designs, and all the Airbus designs. What it skips over is that the Boeing 737 series is completely different and is not FBW in the sense described in the article. In my opinion the 737 is a series of kludges laid one on the other that go way back to the heritage as the 707 and 727 - it is truly an ancient 1950s design. If you presented the 737 for 'new' certification today it would likely not get approved. This is an aside for Tesla purposes, but is important to understand if one is a Boeing investor. For a variety of reasons as SLF my preference is always A320 series over 737 given a choice.

A key difference between aircraft and automotive is that one can stop the car and get out. You can't do that in a plane. So the dgree of control system redundancy required could reasonably be different. I have not seen public domain info on how redundant the full Tesla FSD implementation is intending to be, but in HW3 there appear to be at least two fully independent processors/computers, and I think that there is dual camera coverage over the entire 360. There may be more than this but in principle it looks to me as if they have at least a dual redundant system in mind to be working towards*. In the event of fault conditions requiring reversion to human control the FSD implementation appears to be "stop in lane" in the event of an unresponsive human. But I think we are a long way from seeing level 4 or 5 demo'd in a Tesla. Anyway it is worth differentiating between AI for FSD and the sort of FBW used in Airbus/Beoing as they are not at all the same thing.

regards, dspp

* they may have quad in mind, there are some teases on the web.

odysseus2000
Lemon Half
Posts: 6364
Joined: November 8th, 2016, 11:33 pm
Has thanked: 1530 times
Been thanked: 959 times

Re: Musk endeavours

#208193

Postby odysseus2000 » March 17th, 2019, 1:47 pm

Hi dspp,

Thank you, some very useful information.

Regards,

dspp
Lemon Half
Posts: 5884
Joined: November 4th, 2016, 10:53 am
Has thanked: 5825 times
Been thanked: 2127 times

Re: Musk endeavours

#208194

Postby dspp » March 17th, 2019, 1:51 pm

If you look at the Tesla 3 and the Tesla Y specs

https://en.wikipedia.org/wiki/Tesla_Model_Y
https://en.wikipedia.org/wiki/Tesla_Model_3

knowing that the overall vehicle floorpan dimensions are the same

Wheelbase 113.2 in (2,880 mm)
Length 184.8 in (4,690 mm)
Width 76.1 in (1,930 mm)
Height = 1600mm (for the Y)

and compare them with the VW ID hatch and the VW ID Crozz

https://en.wikipedia.org/wiki/Volkswagen_I.D.
https://en.wikipedia.org/wiki/Volkswagen_I.D._Crozz

(anyone know what is wheelbase ?)
Length 4,625 mm (182 in)
Width 1,891 mm (74 in)
Height 1,609 mm (63 in) for the Crozz

It is notable how similar they are. Likewise the vehicle ranges being touted are similar.

Y: Range 230 mi, 370 km 300 mi, 483 km 280 mi, 451 km

The base I.D. will have a 48 kWh battery with a WLTP range of 330 km and limited performance. The mid-spec model will offer a range close to 450 km. There will also be a range-topping model

So too are the price points, when you compare the prices in the country of manufacture. Y base = $39,000, Y perf = $60,000. 3 base = $35,000, 3perf = $58,000.

Volkswagen has confirmed the car will have a price of around £26,000 for base, You can expect the price to rise to closer to £40,000 for a top-spec ID with a 350-mile range.

(remember that EU pricing includes a typical 20% VAT, and imports into EU incur 10% tariff from USA. So best to compare country of manufacture to country of manufacture for an apples to apples comparison.)

Release dates are
3 = available now
Y = Fall 2020

First deliveries of VW’s standalone electric car aren’t expected to be made until 2020.

It is looking very interesting. It is practically head-to-head competition.

regards, dspp

BobbyD
Lemon Half
Posts: 7814
Joined: January 22nd, 2017, 2:29 pm
Has thanked: 665 times
Been thanked: 1289 times

Re: Musk endeavours

#208200

Postby BobbyD » March 17th, 2019, 2:14 pm

dspp wrote:The combination of radar + wide-spectrum cameras, plus (slow) ultrasonics is better than us mere pink jobs have as sensors.


Attach them to a human brain and we will be laughing.

I don't want to unnecessarily recycle old arguments, but meatware and silicon solutions aren't directly comparable because of the fundamental difference in the processing units. That isn't to say that AD's can't outperform humans, but those AD's are not going to be human drivers rendered in silicon and algorithms, making direct comparisons misleading.

odysseus2000 wrote:In the UK humans kill or seriously injure about 10 people per day with cars, massively more than airplane accidents & so the prize for anyone who can make a low cost non Lidar system is colossal.


As long as LIDAR remains costly. Remember the radar which came with a free $100,000 Jaguar? The same radar now routinely found in found in sub $20,000 cars?

onthemove
Lemon Slice
Posts: 540
Joined: June 24th, 2017, 4:03 pm
Has thanked: 722 times
Been thanked: 471 times

Re: Musk endeavours

#208202

Postby onthemove » March 17th, 2019, 2:19 pm

odysseus2000 wrote:It will be hugely indicative of future AI if non Lidar systems can not work better than humans. In the UK humans kill or seriously injure about 10 people per day with cars, massively more than airplane accidents & so the prize for anyone who can make a low cost non Lidar system is colossal.

If non Lidar systems can not be made to work it will be a huge tell for the future of AI & will likely send off the advent of mass robotic driving long into the future as Lidar systems at wavelengths that do not damage human eyes are expensive & if one needs Lidar one presumably needs redundant Lidar. It is not clear to me that Lidar systems as used by some robotic cars can be made low cost in mass production, but happy to be corrected if this is wrong.


I think you're way off of the mark in understanding where AI comes into the game.

None of the serious self driving contenders will be using "AI" as a single black box. The won't just feed a load of camera inputs into a deep learning CNN and then drive around expecting it to extract every aspect of driving! e.g. learning that it needs to recognise road signs, learning what speed limits are and how to recognise them, learning the rules of junctions, etc.. that would be very naive. (A nice experimental project for a research institute, but definitely not how any serious real world effort would approach developing a self driving car!)

So lidar or not will not tell you anything about the future of AI!

As I mentioned in another post a long while ago on this board (different thread), the modern revolution in AI basically provides engineers with new building blocks for which there was previously no equivalent. It's true that you couldn't make a self driving car without these building blocks - there's no alternative non-AI solution (e.g. to assimilating and interpreting visual information). But how you assemble these building blocks into a working self driving system is standard engineering rather than AI.

The AI in modern self driving cars is (I would hope) modularised into a more traditionally engineered system. Different components responsible for different aspects... one or more components for recognising and reading road signs, like speed limits, etc.. one or more for simply looking where physically it would be possible to drive (independent of human rules)... others looking for the human rules that constrain where and how to drive...

I would expect any real-world self driving car attempt to have inside the software/computer an explicit model of the environment around the car. Something that can be presented to the user, or to a crash investigator, or even just to the original developers as they debug their system.

This may sound "well durr!"... but deep learning CNNs don't of themselves provide that!

One of the criticisms that's always been levelled at artificial neural networks, is that you can't 'unpick' the 'weights', you can' t 'see' what it's 'thinking'. And while there have been a few research attempts to get around that (you've probably seen the psychadelic dreamlike pictures that google put out - https://www.ibtimes.co.uk/google-deepdr ... ne-1509518), they are nowhere near useful for engineering purposes

This all comes back round to - you shouldn't think of self driving cars as an "AI" (singular). They are still a system engineered from components. Some of the compenents may be revolutionary - like the ability to identify and read speed limit signs in real time, to identify traffic lights, or zebra crossings, etc. But these will still feed into other higher level controllers in a traditionally engineered way.

There is not going to be a single 'AI' in a self driving car; rather a collection of specialised modules - some AI some not, all combining to provide the self driving system.

Some AI modules will perform their AI task better than others. I can fully imagine in a few years that people will talk about adding particular AI modules, or talking about what AI modules they have, etc, like people talk today about what type of tyres they have, etc.

The non-lidar aspects clearly can be made to 'work' .. as in do the driving task equivalent of a human being, but we know humans make mistakes.

Lidar on self driving cars is really akin to automatic emergency braking on non-self driving cars. It's just there to cover anticipated exceptional shortfalls.

We accept humans driving without automatic emergency braking, but now the technology exists, I can see that like ABS, etc, it will in the near future become compulsory for none self driving cars.

And as such there's no reason why such systems shouldn't also be a compulsory complement on self driving cars as well.

It's not a 'redundant' system in that it can't actually take over from the AI self driving aspect. It's just a safety fall back for exceptional circumstances.

As long as the lidar has self diagnostics that can establish whether it's working correctly or not, there isn't then a need for dual redundancy or anything like that.

If the lidar indicates that it isn't working, then the non-lidar controller is robust enough to either just pull over to the side of the road, or even potentially make it to a nearby garage a few miles away.

The car isn't going to fall out of the sky if the lidar stops working!

As for the issues you have with lidar, only you seem to be particularly worried by them. I haven't seen any serious sources raise concerns over damaging human eyes - and there are plenty of lidar systems out there on the roads on prototypes which would not be permitted if there was any risk of damage.

The race is currently on to get lidar cheap - and progress is rapid. I watched a (industry) video a few months ago which seemed to imply that it's now a sprint to the finish with lidar. The cost is already getting reasonably low - would already be viable on top end cars. Now it's just the standard engineering processes of getting it standardised, volumes up and therefore cheaper.

In response to other comments on the thread...

As for the general question of redundancy (where did the comparison with airliners come from)... I mean, cars are on the ground, and only carry typically 4 to 6 people. The risk assessment is entirely different.

The risk with a self driving car and failures, is going to be more towards identifying the failure, rather than having redundancy to take over. A car can simply put on a self-driving equivalent of hazard lights (which I'm sure will be developed) that basically says "Hey, I've detected a fault and will be stopping immediately" warning everyone around. And then it just simply applies the brakes and stops.

There doesn't need to be another replacement system there to take over. All that self driving cars need is the appropriate self diagnostics to check that they are functioning as per spec.

Things like checking cameras are functional, and providing the appropriate coverage (pointing the right way), not obstructed, etc. And that integrated lidar is appropriately aligned, etc. And that all electrical circuits are functioning correctly.

There doesn't need to be 'redundancy' of the kind used in aircraft ... in self driving cars, it just needs to be able to detect a fault then stop.

onthemove
Lemon Slice
Posts: 540
Joined: June 24th, 2017, 4:03 pm
Has thanked: 722 times
Been thanked: 471 times

Re: Musk endeavours

#208204

Postby onthemove » March 17th, 2019, 2:37 pm

odysseus2000 wrote:It is not clear to me that Lidar systems as used by some robotic cars can be made low cost in mass production, but happy to be corrected if this is wrong.



https://www.bloomberg.com/news/articles ... iving-cars
Waymo Starts Selling Sensors to Lower Cost of Self-Driving Cars ...the technology has already found its way into the wild to help collect trash, sweep streets and pilot drones. Retailers such as Walmart Inc. and grocery store chain Stop & Shop Inc. are using lidar-equipped robots to help keep shelves stocked. A security company called Knightscope uses lidar to help patrol malls and parking lots... The development curve is similar to Moore’s Law in computer chips: Every 18 months, lidar sensor resolution doubles and the price drops by half. “That whole cost curve and drive of volume is what’s opening new industries,” ....


https://techcrunch.com/2019/03/06/waymo ... 8j9y3G-P8w
Insiders say those costs have fallen further thanks to continuous advances by the team. And considering that this short-range LiDAR is cheaper than the top-of-range product, the price is likely under $5,000 a unit.


According to the first article...

"The closest comparable sensor made by Veloydne—the recently unveiled VelaDome short-range lidar—is priced at roughly $4,000 for a single sensor and half that when purchased in bulk. Velodyne said it will sell more than 10,000 lidar units this year."

So they're selling around 10,000 units per year and can do it for $2000 in bulk.

Now for example take Telsa's production figures / targets of 5000 cars per week.

Imagine if those cars had say 4 or 5 sensors... within 2 days, you'd have used as many sensors as Velodyne expect to sell in a year at $2000 in bulk.

At car industry rates of production / volume, there are some clear economies of scale to be had!

I think it's safe to say that low cost lidar, effective for self driving cars being produce en masse, in the very near future, is pretty well a given.

onthemove
Lemon Slice
Posts: 540
Joined: June 24th, 2017, 4:03 pm
Has thanked: 722 times
Been thanked: 471 times

Re: Musk endeavours

#208207

Postby onthemove » March 17th, 2019, 3:01 pm

odysseus2000 wrote:It is not clear to me that Lidar systems as used by some robotic cars can be made low cost in mass production, but happy to be corrected if this is wrong.


https://www.businesswire.com/news/home/ ... d-Low-Cost
"...today [2016] announced a groundbreaking design for a solid-state LiDAR sensor that can deliver a subsystem cost of under $50 U.S. when sold in high-volume manufacturing scale. The technology will impact the proliferation of LiDAR sensors in multiple industry sectors, including autonomous vehicles ..."


Though I've not been able to find any updates on where this development has got to over the past couple of years.

dspp
Lemon Half
Posts: 5884
Joined: November 4th, 2016, 10:53 am
Has thanked: 5825 times
Been thanked: 2127 times

Re: Musk endeavours

#208210

Postby dspp » March 17th, 2019, 3:14 pm

otm,

It is possible that lidar will be like fuel cells. Lots of good progress, but other things progress faster (BEV), and so ultimately it gets set aside as a mass-produced technology (much to my surprise btw). If they get FSD working on camera + radar only then lidar investment will stop, and lidar volumes will crash, and lidar will not ever get down the technology cost reduction curve. Lidar is costly to get down the last part of the deployment curve.

It is also possible that other radar approaches may deliver the outcomes that lidar is seeking. Millimetric SAR is interesting https://en.wikipedia.org/wiki/Synthetic-aperture_radar

I'm not quite sure of the lingo these days, but what I might call "physics model level analysis" and "behaviour level analysis" will ultimately get applied via NN to image (as video) interpretation. Examples:
- A bag blowing in the wind across a highway is not necessarily something to avoid. That same bag pinned to the highway (roadway, the green safe-to-drive envelope) could conceal a human body, or a hard object (a brick, a bottle). We humans watch the way the bag behaves to tag it with attributes "bag, empty" or "rags on child".
- a child running across the street is easy to detect as frame by frame objects to avoid. A child running behind a parked car causes a human to project forwards and evaluate where the child might run out, and pre-emptively avoid. Those are obvious cases, as are the cases of the ball bouncing across the street (where is the child!), or the dog (that child again).
- etc. (a lot of work is going on in these areas, and a lot of this work is not driven by automotive). Excuse me if I am not up to date on the lingo, it has been a long time since I worked & studied these areas.

The point is you don't necessarily need more sensors to solve these problems, indeed more (or different) sensors can very easily make things worse (as anyone who has worked on CEC can tell you, https://en.wikipedia.org/wiki/Cooperati ... Capability). It all depends.

I watch with interest. Especially as Tesla are doing it in-house whereas VW has been a purchaser. So for Tesla it is a part of the mkt cap evaluation.

regards, dspp

onthemove
Lemon Slice
Posts: 540
Joined: June 24th, 2017, 4:03 pm
Has thanked: 722 times
Been thanked: 471 times

Re: Musk endeavours

#208219

Postby onthemove » March 17th, 2019, 4:16 pm

dspp wrote:The point is you don't necessarily need more sensors to solve these problems, indeed more (or different) sensors can very easily make things worse (as anyone who has worked on CEC can tell you, https://en.wikipedia.org/wiki/Cooperati ... Capability). It all depends.


I dont' disagree with what you say in general.

But when it comes to lidar, I view it more of a final fail safe, rather than a core part of the self driving (although some projects are using the lidar in with the cameras as part of the core self driving part, and also potentially mixing it in with the core self driving part could also help ascertain system integrity - does what the lidar see correlate with what the cameras see, etc)

But like humans have been driving for many decades without automatic emergency braking, now the technology exists, it is now being made compulsory...

https://www.autocar.co.uk/car-news/indu ... ndatory-eu

Likewise, I don't doubt that self driving cars can be made to work well without things like automatic emergency braking. It's just there are different levels of 'well'. If humans after driving for decades now need it, why shouldn't autonomous cars which are primarily using similar visual mechanisms to drive.

The CNNs used on visual images are likely to be prone to many of the same things as the human eye - camouflage and optical illusions.

A CNN/ANN could misinterpret the size of someone - is that a child close up, or an adult further away? If the ground is leading directly away from the camera, it might not be easy to distinguish.

A CNN/ANN could be misled by an unusual sized object nearby. I've seen one student project online where they had a (toy) car being driven from a single (monocular, not stereo) camera that had been trained to infer distances based on the semantic information. That's to say it knows what sort of size a tree typically is, what sort of size a car typically is, and so on, and from that it estimates the distances to each pixel in the image - effectively estimating the sort of results that might come back from lidar (in fact, iirc,I think they trained it using lidar). But if you were to put a bonsai tree near, or a toy car, chances are it might get a little confused.

And that's the key thing - the core AI - the convolutional neural networks (CNNs), etc, are 'fuzzy'. They are fuzzy by design.

There are always going to be very exceptional cases where they could be misled or simply make a mistake.

And that's why I feel that it's very likely that regulations will demand some kind of fail safe fall back - just like automatic emergency braking for human drivers, there's every reason to believe that regulators (and the public) will demand the same for self driving cars.

Whether that's specifically lidar, radar, ultrasound... that's a separate discussion - I think the key point is that regulators will demand that the fall back technology will be a physical sensing system that is able to actually 'sense' the presence of something even through mist, fog, smoke, etc, and not be reliant upon a camera for which items could be hidden through camouflage or other visual trickery.

BobbyD
Lemon Half
Posts: 7814
Joined: January 22nd, 2017, 2:29 pm
Has thanked: 665 times
Been thanked: 1289 times

Re: Musk endeavours

#208223

Postby BobbyD » March 17th, 2019, 4:31 pm

dspp wrote:I watch with interest. Especially as Tesla are doing it in-house whereas VW has been a purchaser. So for Tesla it is a part of the mkt cap evaluation.


VW do have in house AD programmes, notably Autonomous Intelligent Driving GMBH which is a wholly owned subsidiary of Audi, what they don't have is the need to hype products which aren't currently deliverable.

tjh290633
Lemon Half
Posts: 8208
Joined: November 4th, 2016, 11:20 am
Has thanked: 913 times
Been thanked: 4096 times

Re: Musk endeavours

#208234

Postby tjh290633 » March 17th, 2019, 5:38 pm

onthemove wrote:As for the general question of redundancy (where did the comparison with airliners come from)... I mean, cars are on the ground, and only carry typically 4 to 6 people. The risk assessment is entirely different.


It came from me, and I was thinking of "Autoland" systems which use ILS and other radar aids to work. The pilot has to take over once the aircraft is on the ground, but it was used to allow operation in very low visibility.

In a crowded urban street there are probably more people than in an airliner. I can see the advantage of detecting movement all around the vehicle, and being able to negotiate narrow roads, but you can't just switch on your hazard lights and hope that everyone else can avoid you. I live in a narrow lane, in which lorries often find they cannot get through and have to reverse out. One such on Friday was trying to get back out with a tail of cars behind him. Maybe they should have a device which stops them if they pass a sign showing a 6'6" width restriction.

TJH

BobbyD
Lemon Half
Posts: 7814
Joined: January 22nd, 2017, 2:29 pm
Has thanked: 665 times
Been thanked: 1289 times

Re: Musk endeavours

#208241

Postby BobbyD » March 17th, 2019, 6:08 pm

I'd imagine you will be looking at quite a degree of overlap between sensors and systems for anything which is passed for autonomous use. From memory Delphi used to claim 480 degree LIDAR coverage from 4 corner mounted LIDARs.

This is how the system looked at the beginning of last year:

Image

odysseus2000
Lemon Half
Posts: 6364
Joined: November 8th, 2016, 11:33 pm
Has thanked: 1530 times
Been thanked: 959 times

Re: Musk endeavours

#208250

Postby odysseus2000 » March 17th, 2019, 6:48 pm

This articles cover some of the issues facing German industry and there is a link that I couldn't copy that describes the troubles at VW:

https://www.bloomberg.com/news/articles ... =applenews

Regards,

odysseus2000
Lemon Half
Posts: 6364
Joined: November 8th, 2016, 11:33 pm
Has thanked: 1530 times
Been thanked: 959 times

Re: Musk endeavours

#208255

Postby odysseus2000 » March 17th, 2019, 7:01 pm

The Lidar systems that I know of which have been used in cars operate in the IR at 900 nm and more recent ones at 1500 nm.

The problem with 900 nm is that one can only operate at low power as otherwise you damage human eyes, so its limited in its applicability for long range driving where you must be able to 'see' for at least 100 m which is the approx braking distance at 70 mph.

1500 nm don't hurt human eyes and can operate at more power, but they do damage cameras.

Range has a huge influence on radar.

If you consider the most basic isotropic system the power out goes as 1/( distance squared), the reflected signal also goes as 1/(distance squared) so the power levels between1 m and 100 m need to be increased by 10,000. In practice lasers are focused but still must swing over a big arc and the reflections are not 100%, so there is something like 10,000 difference needed.

The lasers, as I understand it, being sold by Waymo et.al are short range so the power levels are low. Please correct if wrong.

Also the idea that robotic cars that have faults can just stop isn't on, not only for TJH's example, but imagine cars and wagons traveling at 70 mph on a 3 lane motor way, if the lidar on one car breaks it can't just stop, it might be able to signal its trouble to all other vehicles that would help it to get to the side, but it has to have some way of getting safely out of the stream of cars.

Regards,

odysseus2000
Lemon Half
Posts: 6364
Joined: November 8th, 2016, 11:33 pm
Has thanked: 1530 times
Been thanked: 959 times

Re: Musk endeavours

#208259

Postby odysseus2000 » March 17th, 2019, 7:06 pm

This gives some information on the potential dangers of lidar:

https://news.ycombinator.com/item?id=18886795

If you are sure its wrong I would like to be corrected.

Regards,

odysseus2000
Lemon Half
Posts: 6364
Joined: November 8th, 2016, 11:33 pm
Has thanked: 1530 times
Been thanked: 959 times

Re: Musk endeavours

#208260

Postby odysseus2000 » March 17th, 2019, 7:11 pm

onthemove
So lidar or not will not tell you anything about the future of AI!

As I mentioned in another post a long while ago on this board (different thread), the modern revolution in AI basically provides engineers with new building blocks for which there was previously no equivalent. It's true that you couldn't make a self driving car without these building blocks - there's no alternative non-AI solution (e.g. to assimilating and interpreting visual information). But how you assemble these building blocks into a working self driving system is standard engineering rather than AI.

The AI in modern self driving cars is (I would hope) modularised into a more traditionally engineered system. Different components responsible for different aspects... one or more components for recognising and reading road signs, like speed limits, etc.. one or more for simply looking where physically it would be possible to drive (independent of human rules)... others looking for the human rules that constrain where and how to drive...



Yes, I understand what you are saying, but I think the composite of systems will feed into a "decision engine" that will decide what to do. I imagine that this will be called an AI system and if useful will be able to drive without any human in the vehicle and do so more safely than if it was human driven.

Regards,


Return to “Macro and Global Topics”

Who is online

Users browsing this forum: No registered users and 11 guests