If you’re unfamiliar with LIDAR, you might have noticed it sounds a bit like radar. That’s no accident – LIDAR is a backronym standing for “light detection and ranging”, the word having initially been created as a combination of “light” and “radar”. The average person is most likely to have come into contact with LIDAR at the business end of a police speed trap, but it doesn’t have to be that way. Unruly is the open source LIDAR project you’ve been waiting for all along.
Unlike a lot of starter projects, LIDAR isn’t something you get into with a couple of salvaged LEDs and an Arduino Uno. We’re talking about measuring the time it takes light to travel relatively short distances, so plenty of specialised components are required. There’s a pulsed laser diode, and a special hypersensitive avalanche photodiode that operates at up to 130 V. These are combined with precision lenses and filters to ensure operation at the maximum range possible. Given that light can travel 300,000 km in a second, to get any usable resolution, a microcontroller alone simply isn’t fast enough to cut it here. A specialized time-to-digital converter (TDC) is used to time how long it takes the light pulse to return from a distant object. Unruly’s current usable resolution is somewhere in the ballpark of 10 mm – an impressive feat.
It’s a complicated project, requiring the utmost attention to detail to get any results at all. The team behind Unruly have done a great job of both designing and documenting the project. It’s great to see an open source LIDAR package in the wild, giving hackers more options than just the pre-baked commercial modules on the market. We can’t wait to see where the project goes next.
For more on LIDAR, check out last week’s Hackaday podcast – we cover Unruly, as well as a handful of other standout projects in the field.
When I see the words “The laser power, pulse width and eye safety ratings are all under software control and can be set to Class 1, Class 1M or Class 3R to suit the application” along with the words “Make it hack worthy” I see many blind people in this projects future.
We would never be able to get away with 3R purely under software control even if it were done to SIL-4 / Level A safety standards.
Personally, I’m just looking for A LiDAR blinder so I can keep these erratic AV’s away from me while I’m driving.
Why would you do that when human drivers are much more erratic on average? Or do you already have strobe lights for blinding those?
There must be something preventing the use of reflective optics so the optics could be fixed above the xmitter/sensor and rotated by a motor on top of that. I’ve not found that yet but see lots of spinning xmitter/sensor units for more than point LiDAR like this unit.
It works, but at short range the system is blinded by the reflection on the mirror
I really don’t think a class 3R laser that is completely open to the world, with no shutters, interlocks etc is going to fly in a sale-able product. Different story in a research lab, but when you’re working with open Class IV on your optical table the whole lab is the laser enclosure, off limits to untrained persons, with door interlocks and danger signs posted, laser interlocks on the room door etc.
Obviously it’s hard to put those features into a lidar – the laser intrinsically must go out into the world.
Regulatory limits to laser safety, which limit the power, seem to be the main range-limiting factor of civilian lidar (and I don’t think software-only here’s-how-you-overclock-it is good enough).
I can’t find the actual laser diode they’re using and its wavelength documented, but something eye-safe like 1550nm would be better (and probably more legal power) than 800-1000nm, which are short enough for retinal danger.
Thanks to everyone for the comments, this is great feedback for the Unruly project.
As it stands at the moment, the Unruly is a demonstrator of what is possible under Python software control. If we decide to release this product into the wild then the eye-safety classification will be embedded into the hardware with the laser power fixed at Class 1.
Please remember that this is an open source project and we want to be as transparent as possible with information about how LiDARs work. Any comments, criticisms or contributions are most welcome.
How will you handle the custom optics information? Will you release the dimensional/material specs or just an orderable part number? What optics manufacturers have you been using and/or recommend?
I’ve had difficulties finding ones that will do custom jobs at volumes less than 10k pcs.
The lenses in Unruly are plano-convex acrylic with AR coatings. BFL is approximately 23mm. We got tooling made for US$2200 (single cavity) and if there is enough interest we could sell the lenses for about US$3.00 each excluding shipping.
We’ve tried lots of different lens manufacturers from USA to China and they all have their pros and cons for both glass and acrylic types.
Great project – I really enjoy watching the project. On laser safety for consumer use, it brings on a new level of complexity but it seems like a simple Failure Mode Effects Analysis (FMEA) could lead to a few single fault modes that would be easy to add safety checks for. For instance, if the laser drive FET shorted in an on-state then an analog watch-dog timer could disable the TX power supply after a few microseconds. I don’t recall the safety rules for Class 1 at 900nm, but I think 1mW average and ~1uJ integrated are the limits for a small aperture exposure across something like a 10 second exposure time. If the system is moving it becomes easy, but if it is stationary things get more difficult. The general guidance should be to not look into the transmitter and keep the pulse repetition frequency low (like most laser rangefinders do now). Keep up the great work!
Thanks for the comments John. We use a certified software tool and calibrated test equipment to verify the eye safety ratings. The actual formulas are quite complicated and non-linear so you do get some interesting effects. For instance, there is a very high average power limit at 905nm, meaning that you can fire the laser faster without affecting eye safety, but there is a low single pulse energy limit that directly affects the rating.
The Unruly uses a “single failure mode” safe design in that the energy for the laser is stored in a separate capacitor that is discharged when firing. The recharge rate of the capacitor is very low so even if the firing FET shorts into an “on” state the current through the laser is too low to turn it on.