Self-driving cars often use LiDAR — think of it as radar using light beams. One limitation of existing systems is they need some method of scanning the light source around, and that means moving parts. Researchers at the University of Washington have created a laser on a chip that uses acoustic waves to bend the laser, avoiding physically moving parts. The paper is behind a paywall, but the University has a summary poster, and you can also find an overview over on [Geekwire].
The resulting IC uses surface acoustic waves and can image objects more than 100 feet away. We would imagine this could be helpful for other applications like 3D scanning, too. The system weighs less than a conventional setup, too, so that would be valuable in drones and similar applications.
The high-frequency acoustic pulses create phonons that deflect the photons in the laser beam. By varying the frequency of the pulses, the beam will sweep over a 20-degree field of view. The phonons are similar to a diffraction grating. They not only alter the direction of the beam but also change its wavelength.
This effect simplifies the receiver setup. When laser energy returns to the receiver, the measured wavelength informs the receiver of the angle corresponding to the transmitted beam. To be practical, the system needs to have a greater range, and this is due to the system’s poor efficiency of about 5%. To reach 300 meters — a number making it practical for autonomous cars — will require 50% efficiency, something the team thinks they’ll achieve soon.
LiDAR is one way for you to measure your drone’s altitude. It is also crucial for your next knife-throwing machine.
+
I love compliant technology. It’s mind bending.
Hmmm. ‘Acoustic’ means that something must be creating sound waves in air. I’m not sure if there is anything nonmechanical that can accomplish that. So there must be something physically moving, right?
“…and interdigital transducers (IDT) are used to generate acoustic waves.”
And:
“IDTs primary function is to convert electric signals to surface acoustic waves (SAW) by generating periodically distributed mechanical forces via piezoelectric effect (an input transducer).”
Piezoelectric effect. That means mechanical movement, and so, strictly speaking, there are still moving parts.
;)
Don’t get me wrong, this is great technology. And like C wrote, I also love compliant technology!
Came here to say this, thanks :)
yep, lots of people consider piezo and even mems as non mechanical… that always baffled me… like the frore system airjet thing which are vane attached to piezo to create ariflow to replace conventional fans… or in a similar way the “bladeless” dyson fans which are anything but bladeless…
It’s a bit strange that people do that. MEMS stands for “Micro Electronic Mechanical Systems”, and “mechanical” means that it’s something that is moving molecules from one place to another. :)
Actually, it’s possible to move air with a plasma, and we can discuss at length if moving ions around with a modulated high potential difference is a mechanical process. :)
It’s mechanical, but unhinged.
The other article on HAD with plasmaacoustic metalayers seems to indicate a fully non-mechanical system, as in creating electric potentials/voltage to control plasma. Maybe a similar approach would work here, and then be really non-mechanical.
‘Acoustic’ doesn’t mean in air- it’s mechanical waves in any medium, including solids and liquids. In this case the movement is 1.55 um. Note that the movement is just in stretching/squeezing a diffraction grating, causing it to deflect light differently.
So yes, movement is a fundamental part of operation, but like… barely. It’s kind of like saying an LCD has moving parts just because the liquid crystals reorient. I guess technically, sure. But not like a micromirror array.
Hey I thought I would clarify this as the research was done in my lab. I believe it used just regular water as the medium which I assume was used being simple to work with.
And I’d like to add on to this too, the main idea behind it being “non-mechanical” is that what is steering the light isn’t a mechanical device, but acoustic waves. Yes technically the production of the waves is a mems device but as another was saying there are many ways to produce sound and isn’t physically steering the laser or mirror.
I don’t remember what journal it was in but there is a lot of clarification on the research when you look at one of the papers published from this that explains it great.
What’s the proper term for “I don’t need to buy a motor and a slip ring”
I have not read the paper, but I expect this is accomplished through creation of a non-uniform pressure, wave field in the air with the Pyo So electric sensor. This causes density differences in the air which caused the light to bend and turn. You can see this phenomenon in the environment if you look at a piece of pavement on a hot day. you can visibly see the air rising on top of the pavement because of the bending of light through the air when it is very hot. The air around the pavement is lower density because it is hotter.
If it’s MEMS i think that’s moving parts too and wondering about endurance until something breaks off in the chip.
It’s not mems. The beam deflection is caused by reflection off a diffraction grating, and compression changes the angle of reflection. There’s nothing to break off.
I love the quibbling about moving parts.
When you hammer in a nail, there’s actually a compression wave that travels down the nail, at the speed of sound in metal, that gets transduced into the wood. Thus carpentry is a subdiscipline of applied acoustics. :)
Right. At the level all this stuff works it really doesn’t matter. Nothings going to break unless one abuses it.
Agreed. Obviously “moving parts” is shorthand for it not having wear-limited moving parts or anything that needs lubrication like bearings, motors, etc. Equally obviously is that all MEMS devices (SAW filters are definitely MEMS, by the way for the nit picker who objected to that) technically have moving parts and are mechanical in nature, but if it comes in a non-serviceable PCB mount package that never needs servicing and can divide extreme operating conditions it doesn’t really matter, does it?
I mean, a “solid state drive” it also ‘technically’ a ‘moving part’ of I use it to smash a pedant in the face.
You can get the paper from here :
https://www.researchgate.net/publication/371944399_Frequency-angular_resolving_LiDAR_using_chip-scale_acousto-optic_beam_steering
“One limitation of existing systems is they need some method of scanning the light source around, and that means moving parts.” Electro-optic devices exploiting the Kerr effect have been doing this for decades with no moving parts (not even MEMs); modulate an electric field to slightly tweak the index of refraction of a crystal, if you’ve cut/shaped it correctly that modulation will steer the beam left-to-right, add another cell in series for up-to-down and you’ve got yourself a scanner.
The device in the article still looks interesting, however. Field of view seems quite wide, anti-Stoke frequency shift for angle-discernment seems cool, and it’s probably the case that acoustic modulation is less power hungry than the electro-optic modulation approach.
Would this perhaps allow higher repetition rate LiDAR? Is conventional LiDAR rep rate limited by mechanical movement? Not a LiDAR expert but interested if this could be used to track v movement of very fast surfaces—1 to 10 km/second scale speed over small distances of order 1 cm length.
Solid state lidar (ie. no moving parts) have existed for quite a few years now. The novelty with this paper is not that it can avoid moving parts, but rather that it can perform beam steering in a smaller and potentially cheaper package.
:+1
Solid state lidars have been overhyped vaporware for quite a few years – things that work in a lab but never managed to reach production because they’re crap.
Quanergy started pivoting back to rotational-scanning lidars during the last year or two before their bankruptcy, but by then it was too late: They’d already spent too much money on their solid state OPA products that never really reached market (the prototypes they had were actually larger than a SICK TiM781, cost around the same, and had far worse FOV and accuracy. They eventually completely gave up on the solid state OPA products because they never got them to work acceptably well – they only offer rotational-scanning LiDARs now.)
Velodyne has an offering that I haven’t gotten a chance to play around with yet. I think up to 30m so not as long as some of their rotating lasers ones but that’s far enough for a lot of applications. Again, haven’t had a chance to play with it but it’s out there.
On a smaller scale, the newer iPads and iPhones have a mems based lidar and ofc there’s the now discontinued real sense l515. They are both plenty useable for certain limited applications although they are most certainly not a replacement for longer range laser scanners.
Regardless, even if such lidar aren’t successfully commercialized yet, my issue with the article is it is written in a misleading way as though this is the first time anyone has built a lidar system without physically moving parts.
Acousto-Optical Modulators are well known things for a long time. Mainly, they, in the form of PCAOM (polychromatic AOM) used to control the color of beam from mixed-gas “white” laser. 8 channel PCAOM made by company named NEOS was most popular choice at the time. But in late 90s on one of ILDA conferences I saw scanning laser display system using AOMs to deflect laser beam instead of usual galvos. In production and available in stock. But AOM systems didn’t get popularity, because that 20 degrees (actually 10, without distorsion and other problems) angle is too small, and you had to use lenses to widen working angle. And lenses means losses and beam dispersion.
Using AOMs in LIDAR will suffer from same problems. Losses, distorsion, beam dispersion, small angle and so on.
Really, for LIDAR simpliest way to scan environment is a rotating mirror. Brushless motor is cheap, simple, give full control of speed and no wearing parts or any AOM problems. Unfortunately, article is not on Sci-Hub yet, so I didn’t read full text, may be the novelty is in some other things that guys figured out.
A MEMs mirror perhaps.
This is super interesting. A few holes though.
First off, to see 10x further they will need 100x the “efficiency”. The power returned from a point source is proportional to the inverse of distance SQUARED. So even if they turn the 5% efficiency into 50 (which can be difficult), they will only see about 3x further.
You also have to ask what can they see at 30 meters and at what probability of detection. There is a large difference in the return signal you get from a nice white diffuse target and a black car door.
There are also other questions around the laser source being used. From what I could see, they are likely using a massive benchtop laser with an amazing line width and spectral purity. That’s difficult to miniaturise. To see further, they will need to increase the power of the transmission laser source and that makes it even more difficult to handle back reflection. Optics aren’t perfect.
Anyway… Great article and very interesting tech from the University. It’s a very novel way to steer the laser and maybe it can be used for Lidar in the future.