DIY LiDAR Thread

[Addison Snyder]

Found this teardown and it gave me the idea of getting a regular IR rangefinder module and swapping out the laser and receiver with a blue 450nm one and a regular photodiode behind a color filter. I'm curious whether it would work.

 

[ntlhui]

If you are spinning a LRF around, then the challenge will be doing the cloud-to-cloud registration to a reasonable degree of accuracy with repeated scans. You'd have to guarantee that during your sweep, the LRF datum remains in the same spot, or at least in a known spot. This is why photogrammetry techniques are nicer, since you can more easily relate a subsequent scan to a previous scan.

Sure, we don't need a pixel perfect cave map, but I do want a straight section of cave to be mapped straight, and not curve to one side or another. Resolution isn't important, but accuracy over distance is.

 

[Addison Snyder]

 

[wetb4igetinthewater]

So a couple things. Previous offerings from Voyis/2G Robotics required a submarine to provide the positioning feed. Submarines are expensive and require a research vessel as shown in that video. Those are hard to come by. In fact, Greece, with all research that is done from May through October every year, only two R/V's exist in the country.

Now if you have a budget like NOAA, not a problem. But I guarantee you that underwater archaeology departments like at Texas A&M and several universities in Florida and other states, do not. It's about half a million dollars to hire a company to do this work for a month (per a phone conversation I had with Fabien Cousteau).

Now the latest Voyis offering, released last summer:

is a game changer. With a navigation system and an ROV from VideoRay, the Voyis laser scanner and skid (for mounting to said ROV), you are talking about a purchase price of under half a million that you have forever (per the quote given to me by Voyis).

And instead of a research vessel, you could use a small boat, as the ROV with the laser scanner mounted is quite small. You need just a boat captain and space for the operator and the cabling required. So theoretically, a Greek archaeologist or marine scientist could take two pelican cases, hop on a ferry, hire a Greek fishing boat captain and acquire the same results as in the YouTube video you provided.


The Defender ROV and Insight laser scanners are rated down to 300 meters. That covers a lot of shipwrecks in the Aegean, I can tell you.

In late August, I'll be heading to Greece along with the founder of Norsk Remote Sensing, to join the Mentor Shipwreck Excavation Programme, to introduce this technology to underwater archaeology in Greece. This technology is the future. We are planning for a small film to be created for the purposes of fundraising to purchase the equipment used and to donate to Greek and American universities. Once those funds are raied and the equipment donated, then there will be more projects. Voyis and VideoRay are in the sales business, not rental, but due to the profile of this project, they will rent us equipment only once. So it is now or never. Failure is not an option for me. Over time, the price of this equipment will fortunately come down, but I want to get it into the hands of as many universities as possible as quickly as possible.

If I'm lucky, there will be an IMAX film made with complete creative control given to Greece's favorite honorary Greek citizen. I hope to make that pitch later this month.

I've been working on this project since September 2018, when I came back from Greece and met one of the oldest surviving spongedivers who reported some of the ancient shipwrecks he discovered during the course of his career. He hasn't reported all of them unfortunately. I'm hoping to search for some of the unreported wrecks he told me about under the direction of a Greek university professor who used to be a government archaeologist.

 

[Addison Snyder]

I'm not looking into buying a pre-built ROV (the same guys quoted me 35k on some passive photogrammetry thing). I'm looking into making (DIYing, one could say ) a couple of cheap parts, smashing them together with an Arduino and/or Rasberry Pi, possibly throwing them in an underwater enclosure, and seeing what happens. I posted that video partly as inspiration for others, but also to show that laser rangefinding underwater using blue/green lasers is possible and happening. Maybe there's some mystical element I'm not grasping right now, but from where I sit it seems entirely possible to pull off. At the moment I'm stuck on getting a blue/green laser rangefinder that has an update frequency of >3hz, but that's in progress.

then the challenge will be doing the cloud-to-cloud registration to a reasonable degree of accuracy with repeated scans.

I've got quotes for various grades of IMUs. So far ~8k would get me one that definitely works well enough for precise/accurate mapping, and ~1k for something that would probably be "good enough" for a while, if not indefinitely. I'm going to prototype using some cheap stuff (~50 bucks), but that technical hurdle has been solved for ~60 years depending on how much money you throw at it. While 10k is a lot of money to me, I'd genuinely consider spotting it down the road if it meant the difference between meh-quality maps and damn-near-perfect maps. As far as repeated scans, I've got an idea where I'd go to mesh those together as well.

 

[wetb4igetinthewater]

I'm not looking into buying a pre-built ROV (the same guys quoted me 35k on some passive photogrammetry thing). I'm looking into making (DIYing, one could say ) a couple of cheap parts, smashing them together with an Arduino and/or Rasberry Pi, possibly throwing them in an underwater enclosure, and seeing what happens. I posted that video partly as inspiration for others, but also to show that laser rangefinding underwater using blue/green lasers is possible and happening. Maybe there's some mystical element I'm not grasping right now, but from where I sit it seems entirely possible to pull off. At the moment I'm stuck on getting a blue/green laser rangefinder that has an update frequency of >3hz, but that's in progress.



I've got quotes for various grades of IMUs. So far ~8k would get me one that definitely works well enough for precise/accurate mapping, and ~1k for something that would probably be "good enough" for a while, if not indefinitely. I'm going to prototype using some cheap stuff (~50 bucks), but that technical hurdle has been solved for ~60 years depending on how much money you throw at it. While 10k is a lot of money to me, I'd genuinely consider spotting it down the road if it meant the difference between meh-quality maps and damn-near-perfect maps. As far as repeated scans, I've got an idea where I'd go to mesh those together as well.

I get that but the link you provided was used to produce this:

You may want to look at the ROVs from BlueROV. Much more affordable and I doubt you care about max depth of 100 meters.

 

[Addison Snyder]

You may want to look at the ROVs from BlueROV

Logistically (and financially) ROVs would not work super well 1000+ft into a cave. For a dozen reasons, it's pretty much out of the question. Navigating flow, lots of rocky protrusions, silt, etc. would make sending an ROV tough - even if you don't include the challenge of tethering it from the surface or limited batteries for propulsion. Possible? Probably, at least from a technical angle. But then you still have to solve the core issues I've brought up in this thread, namely LiDAR and an IMU to tie together the position of the cloudpoints. So I'm focused on those, attaching the resulting module to the front of a DPV, and going from there.

Btw, the ROV might be like 5-10k, but the actual LiDAR used to produce those images is like 35k+, it doesn't come with the ROV.

 

[wetb4igetinthewater]

Logistically (and financially) ROVs would not work super well 1000+ft into a cave. For a dozen reasons, it's pretty much out of the question. Navigating flow, lots of rocky protrusions, silt, etc. would make sending an ROV tough - even if you don't include the challenge of tethering it from the surface or limited batteries for propulsion. Possible? Probably, at least from a technical angle. But then you still have to solve the core issues I've brought up in this thread, namely LiDAR and an IMU to tie together the position of the cloudpoints. So I'm focused on those, attaching the resulting module to the front of a DPV, and going from there.

Btw, the ROV might be like 5-10k, but the actual LiDAR used to produce those images is like 35k+, it doesn't come with the ROV.

So here's the thing about laser scanning. It is dependent on the navigation system. When the data is read, position information needs to be recorded for that data point. Otherwise you just have a bunch of distance measurements and no way to put that back together.

You're really better off with photogammetry for mapping caves. There's a reason why this is so expensive. It isn't an Arduino project.

 

[Addison Snyder]

It is dependent on the navigation system

So, an IMU and then something to log relative cloudpoint by cross-referencing that and time? That's not a hard problem (if you've got reliable equipment). IMU feeds you a position/velocity (presumably into a rasberry pi using one of its many interfaces), the rasberry pi also polls the laser rangefinder and the motor position. Using all three, it would calculate the relative position of the cloudpoint. I want it to do this ~100 times a second as the encoding motor spins. If the encoding motor and the rangefinder don't play nice with each other I'd switch to a step motor, which would be somewhat slower but more precise.

It isn't an Arduino project

The BlueROV is literally built around an arduino. Apples and oranges as far as projects go (one is an ROV, this is LiDAR), but you can do a lot with these embedded/single-board devices.

You're really better off with photogammetry for mapping caves.

Are there any suggestions in that area?

Sidenote, some madlad made his own laser gyroscope: 1-D Laser-Ring Gyroscope
More sidenotes: Fiber Optic Gyroscope Kit - fully assembled

 

[wetb4igetinthewater]

So, an IMU and then something to log relative cloudpoint by cross-referencing that and time? That's not a hard problem (if you've got reliable equipment). IMU feeds you a position/velocity (presumably into a rasberry pi using one of its many interfaces), the rasberry pi also polls the laser rangefinder and the motor position. Using all three, it would calculate the relative position of the cloudpoint. I want it to do this ~100 times a second as the encoding motor spins. If the encoding motor and the rangefinder don't play nice with each other I'd switch to a step motor, which would be somewhat slower but more precise.

The BlueROV is literally built around an arduino. Apples and oranges as far as projects go (one is an ROV, this is LiDAR), but you can do a lot with these embedded/single-board devices.

Arduino even Raspberry Pi's are useful for some projects. I have both, though I am disappointed that my Raspberry Pi 4 with 8 GB of RAM doesn't have the horsepower for online streaming at 1080p. And yes, controlling the motors of an ROV is quite a bit different than sampling a laser and coordinate information.

Are there any suggestions in that area?

For photogrammetry, I can't advise. It is something I want to do just for the sake of finding shipwrecks and reporting to the Hellenic Coast Guard and Ephorate of Underwater Antiquities.

Sidenote, some madlad made his own laser gyroscope: 1-D Laser-Ring Gyroscope
More sidenotes: Fiber Optic Gyroscope Kit - fully assembled

All cool stuff. I wish you luck on your pursuit.