Jul 14, 2016
Gilzo Interesting...
Elon Musk on Twitter�
Jul 14, 2016
Austin Powers Guess this is the big thing in 8.0.�
Jul 14, 2016
malcolm Bit of background
Lidar - Wikipedia, the free encyclopedia
Interesting possibilities for AP.
Let's hope it's a successful R&D which ends up giving drivers a much clearer representation of just how far/what the system can "see".
I think the problem with AP is anthropomorphism; we tend to infer that it has a field of view and visual acuity that is similar to our own.
I am sure that drivers would benefit from a clearer visualisation/representation of the system's true visual field/range/resolution.
In the meantime, just remember the Autopilot song:
"All together now, 1, 2, 3,
Keep your mind on your driving,
Keep your hands on the wheel,
And keep your shifty eyes on the road ahead..."�
Jul 14, 2016
Garlan Garner Absolutely fantastic.�
Jul 14, 2016
Austin Powers Says no changes to hardware. Interesting take.�
Jul 14, 2016
Garlan Garner I thought that was interesting myself. I wonder if there will be no changes to future hardware. I certainly hope that when I get my M3 that it will be on even newer hardware. ( I'm sure it will be )�
Jul 14, 2016
cronosx No.. or partly
He said that he can improve the current radar to an extent, while a new HW will get even better results
It's seems he really don't like lidar, and it's commited to prove his point
�
Jul 14, 2016
Austin Powers Model 3 will get watered down S and X hardware (if they want to make $$$). Also, there will be a premium for such features (e.g. fee to access superchargers)�
Jul 14, 2016
hiroshiy So,... what is a coarse point cloud?�
Jul 14, 2016
ShockOnT It means the radar returns a series of points in space (a cloud of points).
It's low resolution (coarse), so you might not see how many fingers someone's holding up, but you can see there's a person with arms and legs (for example).�
Jul 14, 2016
Gilzo Follow up tweet from Elon:
Good thing about radar is that, unlike lidar (which is visible wavelength), it can see through rain, snow, fog and dust.
Elon Musk on Twitter�
Jul 15, 2016
W84M3 SpaceX uses LIDAR so he does not hate it. He said it's overkill and way to expensive for cars.�
Jul 15, 2016
cronosx I never said hate.. i said don't like, you are right, i didn't specified "in the automotive".
and i didn't say it aloud, but i'm partly for his side, altought i would like more redundancy
but i'm a software guy, and since this could be a security issue, i would like the more redundancy i can.. but then, he need to make the car at an accessible price, and of course, estetical appetible, and lidars while can be small they are a little more hard to hide while manteining the usefullness�
Jul 15, 2016
wdolson That's a very good point people should keep in mind. People think, if they can see something, the car should too, but even most animals have a very different view of the world than we do. Cats for example have a different mix of receptors in their eyes, so they see things differently than we do. They do see color, but much more washed out than we do, and not only do they have eyes tuned for better night vision, but their vision is also optimized for seeing moving things. We see stationary objects very well, but stationary objects for cats are kind of blurry, but moving objects are sharp and clear. They sometimes move their head to see things better.
The sensors on the car see the world differently than we do. Much of the time the computer can come to the same conclusion a human would and do it faster, but other times the system just doesn't see some types objects or misinterprets objects. Pulling into the garage the car sometimes things my tool chest is another car, which is harmless in that circumstance, but the car didn't come to the same conclusion about the object I would.
It is interesting that it looks like Tesla has figured out how to get more information about the world using the existing sensor suite. It may be a while before it is in our cars though. I doubt it will be in version 8.0 of the software, but who knows, they may have been working on it for some time.�
Jul 15, 2016
andrerodpt I'm just wondering if this is part of the Master Plan v2.0.
Full autonomy in X years is nothing new and it will be a major blow on expectations (and stock) if the plan only includes that.
Is it just me or this week went a lot slower since Elon announced the revelation of Master Plan for the end of the week?�
Jul 15, 2016
Matias Because Tesla's radar is not phased array radar, I'm really curious how this would work.�
Jul 15, 2016
Rowsby By using the motion of the vehicle, they time sample and smooth the point cloud data to approximate what a spinning LIDAR system does. Wonder what the processing overhead on that technique is? Has to be high.�
Jul 15, 2016
Matias To my understanding, Tesla's radar has no angular resolution. So I don't understand, how it could distinguish between a truck that has stopped before a T-junction or is right at the junction.
Temporal sampling tells, that there is something new somewhere near the junction, but not exactly where it is.�
Jul 15, 2016
Matias On the second though, if this new radar return is straight ahead, distance to it gets shorter faster than if it is at some angle relative to a car. But is this difference clear enough to make AEB?�
Jul 15, 2016
sillydriver I'll offer my wild guess about this. I also assume that Tesla's radar hardware gives no angular information at all, just distance at a point in time. So as you say, the problem is distinguishing between a truck stopped ahead of you and a truck stopped off to one side.
I think you can distinguish between the two if you make some assumptions. if you are going v = 100 kph (just to use round numbers) and the truck is straight ahead of you, the echo will be closing at a steady 100 kph. However let's say the truck is stopped x = 100 meters ahead of you and a = 10 meters off to the side. It's line of sight distance is r = Sqrt(x^2 + a^2) = Sqrt(100^2+10^2) = 100.499 meters. Its rate of closing would be v * d/dx (r) = v (x / (Sqrt(x^2 + a^2)) = 99.5 kph. Assuming the truck was really stopped and was a fixed distance a to the side, the rate of closing would diminish with time as the angle between your car's direction of travel and the truck increased, until you finally pass the truck, where the closing rate would be zero. The change in the rate of closing would follow a particular smooth curve if the truck was really stopped at a fixed distance to the side. Tracking the change in closing velocity (calculated from distance) and fitting the curve would indicate that the radar echo was not an obstruction ahead, or if the closing speed was steady it would indicate it was an obstruction. Of course if the truck was slowly moving in just the right way it would throw off the conclusion.
EDIT - I see that you came to the same conclusion and posted while I was writing my post.�
Jul 15, 2016
Matias Yes. We both came to the conclusion, that this is possible in theory, but is it possible in practise?
There is very small difference between a truck that is right on the side of the road and which is straight ahead. Difference can be only one meter.
But it is great, if this difference is bigh enough to avoid false positive AEB.�
Jul 15, 2016
electracity The tweet is actually a negative, as it suggests that the next generation hardware hasn't solved some basic problems. A coarse lidar replacement still won't be able to distinguish human gestures and body position like lidar. Google car car read if a bicycle riders foot in on the ground. Musk thinks is Tesla's safety is good elsewhere that his less sophisticated hardware will be good enough.
Two of the three recent autopilot accidents show complete inattention by the drivers. Possibly even sleeping. This is the fundamental argument against Tesla approach to autonomous driving. The better the system, the more drivers can get away with being inattentive.�
Jul 15, 2016
Garlan Garner I absolutely disagree. Elon does not think that way. The M3 will get the latest batteries and technologies.�
Jul 15, 2016
msnow Agree. I have no idea why the poster said that.�
Jul 15, 2016
Austin Powers It's Economics 101. This is akin to asking for BMW 7x features on a 3x
Here is an example: Autopilot costs about $2500 right now. It will continue to cost that much even on the 3 if they were to use the same tech. If they lower the price for that tech, then they will need to bring the higher end model prices down or they will need to innovate further to keep the premium on the X and S. Net-net, the lower end models will not be able to get the latest and greatest tech at a lower price point.
Full disclosure: I have a couple of Model 3 reservations so nothing against that demographic�
Jul 15, 2016
Az_Rael Isn't Mercedes introducing their latest and greatest "autopilot" on their E class and not the S class? And I think Audi has introduced brand new features on their A3 line first in the past before rolling them into the rest of the line. So it isn't unheard of for manufacturers to put features in their lower models that the top line doesn't have yet. They seem to be driven more by which model is up for redesign than maintaining some sort of hierarchy. Especially for stuff that will be introduced fleet-wide eventually.�
Jul 15, 2016
Garlan Garner I still don't buy it. The Model S and X features are primarily software based. It costs absolutely NOTHING to copy software. The M3 has been promised:
Ludicruis - MS and MX have it
200+ miles per charge - MS and MX have it
All future AP hardware standard - Not all MS's and MX's have it.
I'm not sure what other features you could be refereeing to.....that would be vacant on the M3. Air shocks? Who cares?
Elon is certainly forward thinking. Whatever Tesla can fit into the $35K chassis - they are going to get in there. They have competition to beat. They aren't going to blow away the Chevy Bolt by making a car like the Chevy Bolt.
Last thing I recall ELON saying is the word "Cowbells". Even after the promises......"Cowbells".
We will see.�
Jul 15, 2016
Garlan Garner Are you folks listening to the words that Tesla is saying?
ALL M3's will have AP hardware. Period. What else do you want to hear?
So lets add it up. The MS will have AP, The MX will have AP and the M3 will have AP. You can buy the software on all 3 models if you want.�
Jul 15, 2016
RDoc I'm pretty sure that's incorrect. My understanding is that they use Bosch Automotive radars that steer the beam electronically, both vertically and horizontally, and use a continuous wave "chirp" technique to determine both distance and relative speed.
I presume that what Elon is talking about is more a software change than sensor hardware. The data coming back from the radar sweeps would be integrated together to map out a 3D set of points that the software then puts together into a set of objects. It would be very cool if that was assisted by what the camera(s) saw to build up the model of what's in front of the car.
From what I've seen of how TACC/AP works, it doesn't do any model building currently, it just reacts to what it sees instant by instant and drives the car based on that instantaneous perception. Having never written that kind of software, I feel perfectly free to say that it would be a lot smarter and less error prone if it continuously built and updated a smart model of the car's surroundings, then drove the car through that model. I suspect the main hardware change needed would be a stronger computer.�
Jul 15, 2016
Garlan Garner Tesla is where it is today because it ignored Economics 101. Economics 101 would not allow a new startup company in the automotive industry.
Elon paid payroll out of his own bank account for the 2 months prior to the MS unveiling - to the tune of 1.1 million dollars. There is no Economics class that would have a CEO do that.
Tesla is where it is today because its not following "standards". My company is where it is today because its not following "Classroom Standards".
Elon needs to start teaching a new class of Economics titled. "By Any Means Necessary".�
Jul 15, 2016
J1mbo Bosch supply the radar and ultrasonics in the current gen system. I believe Tesla are using the MRR gen 1 radar hardware, and Elon is describing the "micro-doppler" capability of the gen 2 hardware.
This document from the NTSB Safety Forum details how it all hangs together. The focus is on pedestrian detection but the same principles can be applied to any object.�
Jul 15, 2016
Cosmacelf AP sensors are cheap, computing power just a little more expensive. The real cost is in software, which costs nothing to replicate. So no reason why model 3 shouldn't have full suite. I will be disappointed if the next gen AP hardware doesn't have at least three cameras, better radar, and better processing.�
Jul 15, 2016
AmpedRealtor Couldn't disagree more. The sensors in Model S are not anywhere as advanced as sensors in, say, a much less expensive Mercedes E Class. The cost of such technology falls precipitously over time. Of course Model 3 will get improved technology over what is in the S and X. Model S and Model X have nothing special or premium in the sensor category.
I can't wait to hear all the Model S and Model X owners cry foul when the $40,000 Model 3 ships with improved technology, better software, and a more advanced Autopilot system. You'll be falling over yourselves.�
Jul 15, 2016
Matias I'm happy to be wrong in this, but I was under the impression, that phased array radars are just coming to cars and are not yet in use. Do you happen to have some link to this Boch radar?�
Jul 15, 2016
RDoc Pretty much all current automotive radars work in the 77 - 79 GHz band and are electronically steered. Just look up 77GHz automotive radar.
The radial resolution is around .5 m and angular resolution around .5 degrees horizontally, 2-5 degrees vertically. Newer systems are better.�
Jul 15, 2016
Matias I'm sorry, but I could not find any system with beam steering. Could you please give a direct link.
I'm talking about phased array beam steering
Phased array - Wikipedia, the free encyclopedia�
Jul 15, 2016
golfingBuddha I agree with everything here. Well said. Although. they do some holistic path projection/planning but do not (as far as I know) build a 3d model of surroundings. Agreed this would require some pretty intense computing power if that is what you are describing!
So far it seems their theory is, use a camera and a radar to mimic a human driving by using augmented computer vision. The thing seriously lacking in this concept which I believe you are describing is the lack of situational awareness.
Certainly a tough problem to solve.�
Jul 15, 2016
alseTrick So you're saying Tesla should knowingly put in less-than-optimal Autopilot hardware on a car that has more reservations than total cars Tesla has produced up to this point (ie, hundreds of thousands of vehicles with "watered down" Autopilot)?
You're saying that's a good idea? With all of the media backlash over Autopilot right now? You want a worse hardware system to be installed?
By every account the same hardware that's on the S and X will be on the 3 (including future hardware upgrades). And just like the S and the X, it will require an upgrade fee to activate that option.�
Jul 15, 2016
Garlan Garner Its Beta - You don't have to use it. I think its a good idea. Tesla is gleaning real world results.
Its Beta.
DON'T TURN IT ON....if you think something is wrong with it. Tesla isn't forcing it on "anyone". Come on adults.�
Jul 15, 2016
alseTrick Relevance?�
Jul 15, 2016
Garlan Garner Really?
"Less than optimal" = Beta = What you may call watered down = don't pay attention to it if you don't like it.�
Jul 15, 2016
FlatSix911 This is very good news for those of us who have the current gen AP hardware
�
Jul 15, 2016
wdolson Tesla isn't the only start up to do something creative with their accounting to stay alive. When FedEx was in it's first year or two, business was growing, but they didn't have enough cash to make payroll and the CEO knew if he couldn't come up with the money overnight, the company was finished. He flew to Las Vegas with the last funds the company had, won enough to make payroll, and the company never looked back.�
Jul 15, 2016
3Victoria My expectation is that the next release of S/X, probably in the fall, will contain a new sensor suite. The software will be developed with that suite in mind. I then expect that suite will be included in the 3. (After all, we were told that new stuff would come to S/X first). The next gen MobilEye would be sweet if it makes it in time. There was a very good lecture on building a triangular mesh for path determination. Each triangle is marked go/no-go. The car can then proceed along a path through the go-triangles.�
Jul 15, 2016
msnow I don't see how your comments relate to the post you quoted.
Edit: never mind I see there's some drama between you and the poster.�
Jul 15, 2016
RDoc The phrase generally used in these systems is "beamforming". It's a type of phased array that uses digital signal processing to generate the phase shifts.
Beamforming - Wikipedia, the free encyclopedia
Bosch�s MRR Sensor For Driver Assistance Systems�
Jul 15, 2016
calisnow 
Meanwhile, Consumer Reports is working on finding a subscriber base.�
Jul 15, 2016
Matias Beam forming != beam steering.
Beam forming can focus beam to certain direction, but it can't steer it.�
Jul 15, 2016
RDoc You might consider looking at the reference I posted.�
Jul 15, 2016
eloder People are silly.
The Model 3 is still well over a year away. The Model 3 can still have superior features to the MS/MX now, and the MS/MX's built in a year can still have superior features to the Model 3.
A year is a long time for a Tesla not to change. We'll see updated autopilot hardware in the S and X well before we'll see the very same hardware come out on the Model 3. The S and X will still be technology leaders.�
Jul 15, 2016
Matias Could you please give spesific quote where BOCH says is system uses beam steering.
To my understanding, "the front-facing version has a +/-45� aperture angle" only means, that radar's field of vision is that. Not that the radar can be steered within that.
Again, I'm happy to be wrong.�
Jul 15, 2016
CarlK Elon did say S/X will always get new technology before 3 in the last ER. That makes sense too. It's not saying 3 will not get the sensor suit as what's in S/X but likely will not be the next generation one before S/X gets it first.�
Jul 15, 2016
mspohr Even without phased beam steering, the radar sensor will return multiple reflections for different objects at different locations. (Close objects and more distant objects)... As the car moves, these different reflections will change based on their locations. This could give a simulation of "stereo" vision. Perhaps this is the "point cloud" to which he is referring.�
Jul 15, 2016
SR22pilot What is being done is slick but limited compared to what you can get with phased array radar. Considering recent steep drops in LIDAR pricing I wonder if Elon will later change his mind on LIDAR.�
Jul 15, 2016
RDoc Here's some stuff on Delphi radars which are competitive with Bosch. They call it Electronically Steering Radar.
http://delphi.com/docs/default-source/old-delphi-files/b87cda8b-468d-4f8e-a7a8-836c370fc2c2-pdf
and this is a characterization of the system for robotic use
http://www.araa.asn.au/acra/acra2015/papers/pap167.pdf
If you're interested in finding out how this stuff works, I'd suggest doing some of your own searches.�
Jul 15, 2016
alseTrick You're not the only one.�
Jul 15, 2016
alseTrick I think most sane people would agree that new tech will come to the S/X before the 3/Y.
While the 3 is still at least a year away, that gives time for Tesla to roll out a new Autopilot this winter and give the S/X basically a whole year with it before the 3. Perhaps the 3/Y will always be one Autopilot update behind the S/X. Perhaps not. But gadgets like self-presenting handles and doors aren't going to be on the 3. And obviously a full-sized "hatch" won't be either. Something like "3rd Gen Seats" would probably be released on the more premium models first, too.
But I also don't think Tesla is going to put an Autopilot system on the 3 that has "watered down hardware" from what is currently on the S/X. That just wouldn't make sense at all (not that you said that). And in regards to Autopilot specifically, I think Tesla will want to upgrade to new versions on all of their vehicle models at roughly the same time. It will speed up the "learning" of the system, since there will be so many 3's on the road in the next 5 years. Why would Tesla want to delay? It could cost them in the race to full autonomy.
That's my thinking at least.�
Jul 15, 2016
xav- Isn't radar limited to 18 feet though?�
Jul 15, 2016
oneday No. It's the ultrasonics that are range limited�
Jul 16, 2016
Matias Thank you. This technology apparently is not traditional phased array, but produces similar results.�
Jul 16, 2016
RDoc Looking at these collision avoidance systems, it appears that they are interfaced to the car via CAN bus and do the data processing internally using ASICs with the system just reporting the 3D position, size, and velocity of the targets it sees.
Elon's comment about "temporal smoothing" and "point cloud" sounds like Tesla is going to go back into the raw detection data before the ASICs have completely worked it over. If that's true, retrofitting it to existing vehicles might be a major undertaking. I doubt these can be reprogrammed.�
Jul 16, 2016
brec (I have nothing to contribute technically, but here's what I read on Twitter...)
@elonmusk: Working on using existing Tesla radar by itself (decoupled from camera) w temporal smoothing to create a coarse point cloud, like lidar
@ihavetweets: So SW driven big advances coming with AP and no new hardware needed for existing AP fleet?
@elonmusk: @erik_smith1 @ihavetweets certainly moderate and maybe big advances w no incremental hardware�
Jul 16, 2016
RDoc "No new hardwar for existing AP fleet" sounds GREAT to me. I hope it happens and will be fascinated to hear how they do it.
Looking at the technical details of these systems, as much as is released, is really fascinating. Some very clever people have been working on them.�
Jul 16, 2016
ZeroDarkSilver Have you all considered the possibility that he tweeted this to counter the Osbourne effect on current-generation AP cars?
If people are led to believe that existsing AP hardware can be nearly as good as whatever is hypothesized to be in AP 2.0, then they'll be less likely to defer a purchase now.
Elon's tweet is pretty vague and definitely non-committal. We should not be surprised if we never hear about this again.�
Jul 16, 2016
xav- He's saying its possible to improve current system. It doesn't mean they won't make any update to existing hardware�
Jul 16, 2016
hybridbear This is my thought too. I almost canceled my order to wait until AP 2.0 hardware is being built into cars. If I had another vehicle to drive I definitely would've done that, but since my lease is ending & my Focus Electric has to go away I decided to keep my order.�
Jul 16, 2016
ZeroDarkSilver We've seen a similar play before with the battery swap concept. Dangle a cool future technology to allay a temporary perception problem. In that case they actually went to the trouble of building one at Harris Ranch, but of course the whole idea died after that. But mission accomplished, I think everyone is past the range anxiety boogeyman now.
This may be Elon fighting the LIDAR-or-nothing boogeyman. Or the LIDAR-is-just-around-the-corner boogeyman.�
Jul 16, 2016
IngTH Are you sure about that? Musk said in 2014 they would use Bosch, but i have doubts about that.
Does anybody know exactly which brand/model the radar is?�
Jul 16, 2016
omarsultan One of the things Tesla does well is simplify supply chain. It keeps costs down and simplifies SW development. We see this already with battery packs, on-board chargers, center displays, AP HW and drive units. I would expect the S, X and 3 to all have the same AP hardware simply because the more HW variants you have the more complicated you SW development and regression testing gets (exponentially). Much like the battery packs, I think they will then allow you to enable whatever features you want (or bundle them in to the price of the higher end cars (like the used to do with Supercharging).
I do not think you will hear a peep about AP 2.0 this year because they don't want the new product overhang deferring orders and they are already at risk of missing this year's goals as it is. However, I would expect AP 2.0 to hit the S and X early next year so it can be thoroughly tested before it the Model 3s start hitting the streets.�
Jul 16, 2016
ItsNotAboutTheMoney Well, swapping _now_ was about CARB credits, but between the demo and Harris Ranch came tank mode. Tank mode made swapping more complicated and may possibly be the big obstacle to it.
If Elon's talking about enhancing AP1 as reassurance, it'd be because of the threat of it being crippled by the NHTSA, not because of the AP2 rumor�
Jul 16, 2016
scottf200 I would have never paid for the ludicrous upgrade but I would be happy to fork over more money for additional/better hardware to have even better AP. I use it a lot ... in a trust but verify mode with my hand(s) on the wheel!!�
Jul 16, 2016
ZeroDarkSilver You are joking right. The Model X has probably the most complex and challenging supply chain in the history of automobiles. Elon went so far as to call it "hubris". They appear to be taking those lessons to heart with the Model 3 design, but Tesla has a long way to go with improving their manufacturing capability.
Another reason to make it seem like AP1.0 hardware can be stretched to near-LIDAR capabilities. It's not clear at all whether this phased radar point cloud concept is actually practical or confers real-world benefits. All Elon said is they are "working on it". No guarantee it turns into anything that would actually ship.
Don't get me wrong it would be great if this produces some advances, but I'm a natural skeptic. On the other hand, Elon has pulled off some pretty hard things () so...�
Jul 16, 2016
Garlan Garner Very nice.
Everyone chant with me:
Zeeee - Row.
Zeeee - Row
Zeeee - Row
Zeeee - Row.
..........�
Jul 16, 2016
ZeroDarkSilver wha?�
Jul 16, 2016
Garlan Garner I was encouraging people to cheer you on to hit the target of your personal rocket.
Awww. Its to late now.... I had to explain it.�
Jul 16, 2016
omarsultan You are talking about (and Elon was referring to) design and manufacturing--different than parts supply chain. I believe Musk's quote was "the most difficult car in the world to build."�
Jul 16, 2016
ZeroDarkSilver All those falcon wing door parts come from suppliers, no?�
Jul 16, 2016
callmesam Or built in-house.�
Jul 17, 2016
Todd Burch Tesla's cars are probably more vertically-integrated (meaning parts built in-house) than any other car in the world.�
Jul 17, 2016
JohnSnowNW ![[?IMG]](https://lh3.googleusercontent.com/blogger_img_proxy/AEn0k_tyNqsfQ9GPe_Th_9_jD4kydrV5hpIaOp1Mr0pDh0FvvMDfA8NS6qw91jkDgMwTOCfDwcv4wS3ukZ_zXurDUaVKkJvH=s0-d)
�
Jul 17, 2016
IngTH Should be a Bosch front MRR sensor then?�
Jul 17, 2016
Garlan Garner Great News! That means that it should be available to deliver in 18 months or so.�
Jul 17, 2016
Matias From earlier announcements (e.g.) Chademo adapter, we know, that "soon" is 18 months. So I think 18 months is too optimistic.�
Jul 17, 2016
TaoJones An October arrival of v8.x with improved AP1.0 functionality would suffice.�
Jul 18, 2016
Hota Yes, absolutely. I think Tesla is very concerned about total dropoff in Model S sales and that people are either waiting for AP 2.0 or the Model 3 so the company is trying to counter. Look at all the recent prices drops and the referral program.
Now I do think that they are legitimately looking into improving the existing AP but only as a counter to prevent further sales drops as people are holding out.�
Jul 18, 2016
Garlan Garner Geeessshhhh. I just need to be happy for a minute. lol�
Không có nhận xét nào:
Đăng nhận xét