Pioneer Looks To Laserdisc Tech For Low-Cost LIDAR 52
itwbennett writes: Pioneer is developing a 3D LIDAR (light detection and ranging) sensor for use in autonomous vehicles that could be a fraction of the cost of current systems (the company envisions a price point under $83). Key to this is technology related to optical pickups once used in laserdisc players, which Pioneer made for 30 years. From the ITWorld story: "The system would detect objects dozens of meters ahead, measure their distance and width and identify them based on their shape. Pioneer, which makes GPS navigation systems, is working on getting the LIDAR to automatically produce high-precision digital maps while using a minimum of data compared to the amount used for standard maps for car navigation."
Re: (Score:2)
Re: (Score:2)
You are off by one post.
Re: (Score:1)
Gay Man Walking!
I just saw one for $115 (Score:1)
https://www.sparkfun.com/products/13680
The operational description is pretty cool too.
Re: (Score:2)
I am still thinking about getting a v2 in the next month or so
Re: (Score:2)
That looks more like a laser range finder than LIDAR. That is there is no scanning mechanism to build up a 3D map of the environment. You could add one, but as it stands there isn't one so calling it LIDAR is really rather dishonest.
High Accuracy Point Cloud Data is scary.. (Score:4, Insightful)
GPS location information has revolutionized mapping. Google and Apple get high quality position data for cars traveling. Any non-mobile platform system have many barriers to entry (cars traveling around taking photos), verifying road information. Google gets to validate incredibly accurate maps under the guise of providing traffic, location services and so on the the mobile phone users.
LIDAR, particularly those paired with accurate color mapping information for those point clouds are going to be creating high accuracy, full color 3d meshes of the pretty much anywhere that cars go near. The small snippet of video just after https://www.youtube.com/watch?... [youtube.com] (from the previous slashdot article about autonomous golf carts) shows some of the power of LIDAR. Just after the car flyby it shows the point map being updated by every pass through an area. Incrementally correcting the 3D view of the area. The example is just a few passes by the same car.
Imagine 1000s or 100,000s of automonous vehicles with LIDAR pushing data to a vendor for "navigation experience", but still building a 100% model of everything line of site to the LIDAR. And then imagine 100s of drones flying overhead, doing virtually the same thing.
Throw in some AR like Magic Leap being able to deal with that 3d Mesh. Pretty amazing things that our kids will grow up, pretty scary for everyone else..
Re: (Score:2)
So basically Google Streetview in 3D. Somehow I fail to see the big problem, it's not like the view from public roads is a big secret, unless it's considerably above human height and peeking into walled gardens.
Re: (Score:2)
I'll give an analogy.
Street view is like someone snapping a photo of your house as they drive by every few months (or years).
The LIDAR information coming together now, is every few minutes (when we have lots of autonomous cars) having high accuracy 3d streetview being pushed. If you move a brick, google* will know. If you cut back a tree, google* will know. If you have a new car, google* will know. And google* will know within a few hours of you having changed your visible frontage.
Although it could
Re:High Accuracy Point Cloud Data is scary.. (Score:5, Informative)
Point clouds are enormous data sets. It is not uncommon in LIDAR mapping to have the results be several gigs in file size. To send that data back to a central mapping location over cellular is not likely. Also the computation required to make a useful mesh off of a point cloud is equally significant. Then the CPU/GPU/RAM requirements to render and handle that 3D map and have it appear on a low res in screen dashboard is not insignificant. I agree with what you are saying about the data collection, but currently there is no simple way to send and receive the huge amounts of data and have them processed. More likely in the near future autonomous vehicles which return to a spot for physical uploading of TB's of data are more cost effective. Then a hugely decimated mesh is generated and it will slowly begin in appearing in vehicles. Google once again is far ahead in this effort. The alternative: photogrammetry is less data intensive but is better suited to UAV's.
Re: (Score:2)
I agree, the point data is available for processing. Simple ways of reducing the data load (assuming that wireless stays slower than broadband), is having the existing mesh, and identifying outliers from a confidence interval. Those outliers are rejected as either transients (cars, people, etc) or changes in the environment. A lightweight protocol could allow vehicles to identify candidate areas and upload as needed. We have early dot.com era enterprise class processors in our pockets. The huge amo
Re: (Score:2)
I concur. WiFi etc will make the data transmission a non -thing. The outlier differential concept does work well. I already use it in cleaning up point clouds before meshing. Very accurate (RTK) GPS would be a big benefit as well. it would be interesting in your scenario if the map itself over time evolves. Colors when leaves change or fall. Watch a street sign adjust to new grafitti etc. The time lapse/delta would be a useful big data set to mine for many sorts of information.
Re: (Score:2)
Exactly. The many sorts of information is where there is a spectrum of
- Cool (accurate fall color maps, tree growth rates, etc), to
- Annoying (local government charging for mundane, but visible property improvements), to
- Scary (complete timeline of when you were at your house, who visited and when).
Cool is cool, Scary is where I pause.
Google knows where you live, it knows where you drive. (I'm not trying to demonize google, but they have all the cards).
Yes but... (Score:2)
Think about the video game implications! All that 3D data! It was everyone's dream to play your FPS in your little home town... imagine a whole virtual world. World organized virtual battles... The Chinese are invading Canada, quick get online and do your part citizen! :)
Re: (Score:2)
And the real-world, in location practice for urban warfare...
- To foreign state, 'please use our autonomous vehicle, they are _really_ good'.
- Years later, turn on intelligence mode for the vehicles.
- To army, stick your VR goggles on and get familiar with the foreign state.
- Drop army in, and they already have ground level intelligence as to what is where...
VR and 3D gaming is definitely a dual-use technology.
Re: (Score:2)
All I want is a feedback system to warn me when I'm driving too close to the car ahead of me, or being followed too closely by the car behind. You need a minimum of 2 seconds lead time to avoid running into the car ahead of you, at whatever speed. 1 second is for you to react and start applying the brakes, and 1 second is for the brakes to stop the car. For rain or fog, add 1 second. For snow, +2. For ice, +4.
Re: (Score:2)
1 second is for you to react and start applying the brakes
Pretty much a worst-case assumption, but okay.
and 1 second is for the brakes to stop the car.
[...] at whatever speed [...]
Uh, what? I'm pretty sure that the time to stop is a function of the speed, and for common cruising speeds significantly above one second.
It doesn't really matter, though, because as so many, your statement is based on the assumption that the car in front will just go from cruising speed to stationary in zero time.
Their requirements are lacking (Score:2)
Unless they just want the small/low speed vehicle market, their requirements are grossly lacking. For a road-going autonomous vehicle, "dozens of meters" is useless at any real speed. Try ~100 meters as a start and go from there. Otherwise you'll always be outrunning your perception and unable to see anything in time to stop.
Re: (Score:1)
For a road-going autonomous vehicle, "dozens of meters" is useless at any real speed.
If you can see for 100 feet, you can stop most vehicles from most non-highway speeds... if you can make snap decisions.
So, how long does it take an autonomous vehicle to decide to brake?
75 meters braking distance to dead stop at 70 MPH (Score:2)
At highway speed, 70 MPH, a car needs 75 meters (dozens of meters) to come to a dead stop. Human reaction time requires further distance, but that doesn't apply to fully autonomous vehicles. It only partially applies to a technology-assisted safety system which augments the human driver.
Re: (Score:3)
Sure, if you assume no processing time is required to characterize what the sensor sees, and want to slam on the brakes at a minimum of 0.66g's of deceleration the moment anything enters the edge of your perception. That doesn't seem like a viable human-carrying vehicle to me.
1 million processing operations = 1.2 inches (Score:3)
> Sure, if you assume no processing time is required
Let's give the system time to do a million processing operations. The GPU and CPU should be running at 1 Ghz or better, so we need 0.001 seconds to run a million operations.
70 MPH is 103 feet per second, or 1236 inches per second. So in the .0001 seconds required to do a million processor ops the car will travel 1.2 inches. You figure it could do more extensive processing, 10 million operations? Okay, that's 1 foot of travel.
Computer processors are re
Re: (Score:2)
A ball bouncing into the road is often followed by a child. On urban/suburban streets a bouncing ball is cause for hard braking.
This is the kind of problem that will require strong AI to resolve. Until then we will have to make do with highway driver assists.
For those who say: 'Simple, the class of toys is finite. Just build an expert system to recognize bouncing balls etc.' Consider how much fun it would be to throw a bouncy ball from an overpass into highway traffic. Your expert system also has to de
Re: (Score:2)
You should learn the difference between a single sensor return and a full perception result. Also, look up sensor noise, integration time, classifier inputs and training. Please tell me how you're planning on dealing with rain and snow.
Yes, if it was as simple as a magic sensor that could tell you instantly at max range 1) that the thing is a car (even though you won't see all of it immediately) 2) what its velocity vector is and 3) what its going to do, then yes, things are as simple as you wish they were.
Re:Their requirements are lacking (Score:5, Insightful)
Most accidents occur at less than 40 mph; if "dozens of meters" equates to about 100 ft, that represents about 1.7 seconds at 40 mph. Assuming a coefficient of friction of 0.8, it is theoretically possible for a car traveling at 40 mph to stop in 67 ft; call it roughly 70 ft. If the system can apply the brakes within 500 ms, that's enough to be useful, although clearly it can't stop you from plowing into a car stopped in the fast lane of the highway.
Speaking of highways, the only reason people can manage to drive on highways is that the things you're most likely to hit are traveling in the same direction; if they were slaloming between stationary obstacles at 60 mph most drivers would be dead, fast. What makes highway driving safe is that the closing speed between vehicles is usually modest; usually on less than ten fifteen miles per hour. So actually the system might have more effect on the highway so long as speed discrepancies are in the normal range.
Re: (Score:2)
"Speaking of highways, the only reason people can manage to drive on highways is that the things you're most likely to hit are traveling in the same direction"
You've obviously never driven on highways where there's no median and the opposite lanes are seperated from you by a magical force field generating white line.
"usually on less than ten fifteen miles per hour."
Try 140mph for the above scenario.
Re: (Score:2)
I have, but I avoid dodging oncoming cars.
Re: (Score:2)
You might not get a choice one day. Watch some of the car crash videos on youtube for examples.
Re: (Score:3)
fuck it, it won't do as much work at a relative speed of 140 mph, lets all go home.
who's idea was it to develop this piece of shit?
have him taken out back and shot.
Re: (Score:2)
The cynic in me says that Pioneer is sitting on warehouses full of 780nm IR laser diodes (used in LaserDisc and Compact Disc players).
Calling them "LIDAR components" will allow Pioneer to mark them up at year end.
I was wondering what, other than the laser, is common to the two applications, given that disk reading occurs at short range over a very narrow angle, and that the pattern being recognized is very simple. My slightly less cynical guess was that the promoters of this project within Pioneer are saying this to give it some superficial appeal to the bean-counters as a worthwhile investment, but I like your theory.
My ship has finally come in! (Score:4, Funny)
I can now build an autonomous vehicle!
WOOHOO!
Re: My ship has finally come in! (Score:2)
well, you've got four motors - that's a start. Assuming the belts aren't rotted.
Re: (Score:2)
Re: (Score:2)
Re: (Score:1)
Re: (Score:2)
"Pioneer, which makes GPS navigation systems"... (Score:2)
really? that is what they are famous for?
I have to google it...ah for cars:
http://www.pioneerelectronics.... [pioneerelectronics.com]
Re: (Score:1)
Re: (Score:2)
Oh god, please don't let this happen. My police laser detector will be completely worthless.
If you can't afford a new V1 every few years, you can't afford to speed. At least, not as a lifestyle.
Some more details? (Score:2)
I find the original article and the one at slashdot to be spectacularly void of information.