Home        Store        Learn        Blog

Using underwater LIDAR for non-GPS navigation

Hi Colin,

You imagine this type of software : https://youtu.be/6a_zg5s6_QE

We have used it to construct the 3D model of cave in realtime. The sensor used is sonar profilometric 2D, so we need sensors to know orientation (IMU) and position (USBL or DVL) of the robot. With your lidar in 3D version, we can imagine don’t use positioning system to reconstruct the model, it is very interesting.

Benoit

Hi Benoit

Yes this would be very interesting for us.

Our vehicle would be using the IMUs from the Cube Orange device with optical flow aided by a single beam echosounder.

It would be amazing if we could see the 3D structure being created in real time as we have a dual screen display on our control system. The first screen is the QGC software - this would be a great addition for us on the second screen…

We would be happy to send you a prototype LIDAR to see if it can be interfaced in the same way as the sonar.

C

Dear Colin,
I am Lionel Lapierre, working at University of Montpellier, with Luc and Benoit on research project related to karstic exploration with robot. As Luc and Benoit already said, we are very interested and impressed by the underwater LIDAR you are developing. The 2D version would be quite useful for scanning a section of the underground conduit, and the 3D version is opening very interesting SLAM perspectives.
We are currently developing an acoustic SLAM algorithm which could be adapted to your device. Would you be interested if we try to adapt and run our algorithm on some data acquired with your 3D device ?

Cheers

Lionel

Hi LIonel

Yes that would be very interesting to us. We would be happy to send you both systems as soon as we have prototypes ready. Hoping to have the 2D version ready in 2-3 months time and the 3D by end of the year.

For the 3D version do you think it might be possibly to navigate in open seawater by scanning sections of the seabed? This is not something we need for our current project but it would be a huge advantage for future ones…

C

Hi Colin,
well, with the 2D scanning, you can always build a map and check for loops but you won’t be able to correct the current odometer drift (while building the map). Nevertheless, the 3D scan can be used, with sufficient overlapping, between 3D scans and perform a complete SLAM, if the terrain has discriminant features. If you have to some data sets (including simulated ones) we can check if our SLAM approach can be used.

L

Hi Lionel

Sounds great - we will get you some data as soon as we have the systems ready…

C

Have you guys looked at what they are doing with the Skydio2 drone?

Seems similar at least concerning the processing power/ algorithms. Real-time terrain mapping and flight correction.

Looks like an amazing machine Darrell.

I wonder if they are creating point clouds from all those cameras?

C

Hi all.
I’m Atle and I work for NORCE, Norwegian Research Center, on perception systems.
Your work on using subsea LiDAR is really fascinating, but the conversation is now one year old.
Do you have any updates or results you can share?

During my PhD, I designed a segmented LiDAR reflector to reconfigure the FOV of spinning LiDARs (See https://doi.org/10.3390/s20123388). You may find it interesting for the forward-facing LiDAR.

Atle

Hi Atle

Yes, we have had a number of delays, mainly due to concentrating our efforts on our photogrammetery camera system.

We acheived our first 2D scan around 6 weeks ago, and are now testing a prototype 3D scanner. Its a resin printed system with nylon gearing, so not the best quality - but we hope it will be good enoguh for the testing we need to do, before moving to the CNC for the final version.

We are hoping to see the first 3D scans in the next week. I expect them to be fairly rough, but should be good enough to allow us to optimise the horizontal and vertical control.

Your paper is very interesting. In the oil & gas and renewables industries we often have a desire to optimise the LIDAR for a particular object of interest. The ability to increase point cloud density on that area is novel.

C

Great to read that you are making progress.

Is it possible to share some of the intended specifications for the 3D LiDAR? Is it based on triangulation or time of flight and which wavelength did you use?

I’m (among other applications) investigating spinning LiDAR with reflector for aquaculture use.

Atle

Hi Atle

System is a 360 degree horizontal scan and 50 degree vertical (+/- 25 deg adjustable center).

It is a time of flight unit at 420nm wavelength. Target is 72,000 points per second, but we may fall a bit short of that.

It will be approximately the size of a coke can.

Aquaculture is certainly on the list of target markets for us.

C

Hi Colin
I’m looking into underwater navigation for one of our projects using stereo camera and IMU, but are also considering lidar. Do you have any additional information about your progress? How’s the development and testing progressing?