Slashdot Mirror
Tiny LIDAR Chip Could Add Cheap 3D Sensing to Cellphones and Tablets
There are expensive dedicated devices that do 3D scanning (like the high-end tablet in Google's Project Tango), and versatile but bulky add-ons, like the Sense from 3D Systems, but it's not a capability built into the typical cellphone or tablet. That could change, thanks to a microsensor being prototyped now (at low resolution) at CalTech.
From The Verge's coverage:
The tiny chip, called a nanophotonic coherent imager, uses a form of LIDAR (Light Detection And Ranging) technology to capture height, width, and depth information from each pixel. LIDAR, which shines a laser on the target and then analyzes the light waves that are reflected back to the sensor, are best known for their use in precision-guided missile systems and self-driving cars.
While LIDAR itself isn't new, [project lead Ali] Hajimiri explains that "by having an array of tiny LIDARs on our coherent imager, we can simultaneously image different parts of an object or a scene without the need for any mechanical movements within the imager." Each "pixel" on the new sensor can individually analyze the phase, frequency, and intensity of the reflected waves, producing a single piece of 3D data. The data from all of the pixels combined can produce a full 3D scan. In addition, the researchers' implementation allows for an incredibly tiny and low-cost scanner, all while maintaining accuracy. According to the researchers, the chip can produce scans that are within microns of the original.
While LIDAR itself isn't new, [project lead Ali] Hajimiri explains that "by having an array of tiny LIDARs on our coherent imager, we can simultaneously image different parts of an object or a scene without the need for any mechanical movements within the imager." Each "pixel" on the new sensor can individually analyze the phase, frequency, and intensity of the reflected waves, producing a single piece of 3D data. The data from all of the pixels combined can produce a full 3D scan. In addition, the researchers' implementation allows for an incredibly tiny and low-cost scanner, all while maintaining accuracy. According to the researchers, the chip can produce scans that are within microns of the original.
You can bet the porn industry is the first to take advantage of this
In phones, maybe. But automated cars, obvious application.
John McAfee 'It was like that time I hired that Bangkok prostitute; to do my taxes, while I fucked my accountant'
This is the imager only, the sensor, not the part that sends out the lasers. You still need laser emitters to use with the sensor.
Honestly, it's insane how ludicrously powerful phones are getting. I meant take this one article for example - 4K 120fps slow-motion video recording is coming to smartphones in 2016.
4K videoing at 120fps?! From a smartphone?
Isn't progress wonderful.
Why OpalCalc is the best Windows calc
No small amount. 123d Catch is usually broken and buggy, and being able to do terrain scans with my phone is important for me.
Plus, I dream of the day when we can take up full 3d video on our cell phones. Not silly stereoscopy, actual 3d, with the ability to choose any perspective. With good enough quality on the 3d imager to understand transparency and reflection (which are painfully common in real life), and with a good enough software stack to assume "that which I couldn't see between times T1 and T2 is a reasonable interpolation of how it looked and where it was positioned between these two timepoints". Plus the ability to knit together multiple people's recorded 3d data into a single scene.
Recording full 3d data may sound like a huge recording, but actually when you think about it it's not really, the geometry and texture of scenes doesn't change that much between frames in the real world; most objects are static. And the macroscopic changes should be readily subject to compression. Within a decade or two of 3d video algorithm refinement I could envision the results being smaller than a 2d recording of the same scene.
But anyway that's all "gee whiz wouldn't that be neat" stuff... right now, I just want to be able to scan landscapes on my phone better than 123D Catch can do. :P
Trump's plan to get rid of Mueller appears to be 'be so guilty of so many things that Mueller works himself to death.'
When you have a team of PhDs working on a project essentially for free (paid for by the government, not CalTech), in a subsidized (nearly free) clean room, on a device where yield doesn't really matter, "cost" tends to not be realistically estimated.
It is not more realistic to estimate the cost by looking at the actual money spent by all sources on the project. That's likely a couple hundred thousand dollars on this one (or so) chip, but most of that is NRE.
When someone tries to fit this into a commercial process and figures out what custom processes are required, we'll find out what the real cost is. It may be $1000/chip, and that may still be very marketable.
Lidar isn't an acronym, its a compound word from light and radar (radar isn't an acronym anymore either).
I use one of those iPhone apps that lets me scan documents with the phone, emitting a PDF that's just as good as the ones I used to need a flatbed scanner to get. When I need to scan a book chapter at the library or surreptitiously copy someone else's document, no more having to arrange to bring it home to be scanned.
LIDAR would extend this concept to 3-D output. Most of us don't have 3-D printers at home, and would have no use for one that just works in cheesy plastic. But imagine being able to email a 3-D spec file you scanned in the field to a service that would print, say, a replica of the gold ring you liked at the jewelry store. Not cheap, but a lot better than paying retail markup.
Just imagine, there could be a phone app that displays an arrow to show the user which way to walk. Using the Lidar to detect obstacles, the app could enable a phone zombie to become almost self-driving, avoiding obstacles and other people. Almost like a real person.
For a smartphones, I'd rather expect so-called "time of flight" cameras to catch-up before LIDARs. Basically, you have an array of LEDs which illuminate the scene using sine or square wave intensity modulation. The imager works at a high framerate (or uses other windowing techniques) to extract the phase shift in each pixel, which gives you 2D ranging information. Of course, there is still the problem of phase unwrapping.
So in this kind of system, you trade off dynamic range for accuracy and cost. As most measurements with smartphones will probably be performed at short distance, this system seems more suitable than regular LIDARs.
Firstly: Try catching a concorde flight today.
Secondly: All new-design contemporary turbofan jet aircraft travel slightly SLOWER than their turbojet predecessors (0.85–0.855 for a 747-400 instead of 0.89-0.91 for turbojet designs such as the Convair 900 and Boeing 707 - the 747 has always been a turbofan engine aircraft). Going faster than this has severe fuel penalties as it causes major issue with the fixed-pitch fans (and the larger the fan the lower the top speed).
There's a push to make civil transports slightly slower still for fuel economy purposes (bigger fans == greater efficiency/thrust but lower optimal speed as noted above). Having said that, the journey times are slightly faster because higher thrust levels mean that newer aircraft (or re-engined versions of old designs such as the 747) spend less time accelerating to cruise speed and altitude. In addition the higher thrust levels mean that the aircraft can carry far more passengers than original designs at the same or lower overall fuel consumption (same latency, bigger pipes)
Believe it or not, the big thing currently keeping speeds high is wing sweep. The "traditional" sweep on a jet transport was "set in stone" in 707-727 days and is optimised for near-transonic speeds. It's inefficient at the actual high-subsonic 0.8-0.85 speeds travelled at today - with inefficiency increasing as speed goes down (and of course, swept wings have nasty low speed stall characteristics plus they're a major contributor to the height limitations of current transport aircraft for the same reason - stall characteristics in thin air/high speed - aka "coffin corner", exacerbated by speed limits imposed by the use of fans for thrust instead of pure turbojet)
Reegnineering the 747 or other classic designs for a lesser sweep is a no-go area. There's been a lot of resistance within the industry to "de-sweep" wings as it's felt that passengers will associate this with turboprops, but this is a concept which _will_ happen with increasing emphasis on fuel consumption.
Several industry magazines have postulated that the ideal cruisng speed for future civil transports is likely to be in the 0.78-0.80 range, although there's a possibility that higher altitudes will be involved too (going above ~45,000 feet brings ozone into the cabin. This needs mitigation) To get there _will_ require much straighter wings than we currently see. Whether passengers like it will probably not matter in the end.
For a smartphones, I'd rather expect so-called "time of flight" cameras to catch-up before LIDARs. Basically, you have an array of LEDs which illuminate the scene using sine or square wave intensity modulation. .
Unfortunately, emitted IR signals outside get too corrupted for ranges farther then 10m or so.
I'm not sure about indoors but I can't find any ToF system that can go farther then 10m.
Light Field cameras almost seem useful but they have their own limitations.
Lidar is the only reasonable way to obtain depth information over long distances. And it's accurate too.
ToF does work ok for short distances though, AFAIK