On the back burner for the moment mate, started a new job and haven’t had the time to work on it.
Small bits of progress, mainly needs the software side sorting
1 Like
@Bobbysmile this looks like a fantastic project! If you have a moment it would be great to pick your brain on some questions.
For background my interest is in mapping for orienteering maps. At the moment I’m filling in holes in 10+ year old LIDAR data where erosion has made it out of date, so fortunately there is limited veg and photogrammetry works (hence my asking about RTK on a DJI mini on your other thread). Down the track I’d like to investigate a LIDAR solution, and am wondering if with my technical skills I can create something within a vaguely affordable hobby budget.
The key questions I have are:
- With a sensor like yours (or I was also looking at an livox Avia Avia LiDAR sensor - Livox ) are the results going to be good enough for my purpose – in particular, would you expect to be able to see through forest canopy sufficiently to detect terrain hight? I have limited background in this and my benchmark is the 10-15 year old LIDAR available for my area which was conducted with planes.
- What sort of IMU are using/is needed? I worked on autonomous underwater vehicles 15 years ago and back then they were incredibly expensive…
- What sort of gimbal are you using?
- I am only newly back into flying (after a 15 year hiatus) and have been gobsmacked by the reliability and ease of the DJI mini. Has ardupilot etc. come far enough in that time that you can reliably use one for this sort of application with a reasonable expectation of getting the gear back? Or would you be better off searching for second hand commercial drones (e.g. I’ve seen some secondhand Matrice drones come up.)?
Hi,
-
The Velodyne one is dual return, it works OK through foliage, more modern sensors are triple (like the livox) or quad return which will perform better. It’s about adjusting expectations though, you’ll get the odd point, but don’t expect to see a clear ground plane under the shadow of dense trees with anywhere near the same density as light foliage, but you have a chance of getting some data through the gaps. The images on Livox’s website seem fairly representative. With repeat scanning you can get really impressive results though, look at some of the stuff from the L2, its crazy where the tech has come to.
-
The IMU is what killed this project for me, I had an old SBGSystems Ellipse (from long before they called them that) but it had major comms issues I could never resolve. Within the ‘hobby’ realm for good data you’re looking at one of those, or one of the XSens units, both of which are thousands of pounds. You’ll want dual antenna heading to get usable results, this is what the matrice and other commercial systems are using, I can’t speak to their IMU.
-
I never got to the stage of gimbal mounting it. I did one test hard mounted, saw the IMU wasn’t up to scratch and sadly had to shelve the project for lack of time.
-
Honestly, ardupilot was a battle, it’s seriously impressive in what it can do, but it’s a steep learning curve and just one wrong move away from going belly up. It gives you a lot of flexibility though. If I went back to the beginning, I think I’d still use ardupilot, but work on a larger octocopter platform, it makes life so much easier.
Thanks a lot for the detailed response @Bobbysmile! My slow reply is only because I’ve been thinking carefully about all the various aspects!
There are quite a lot of aspects of this project that could be showstoppers on my budget. So I think the first step will be to put together a reliable drone platform (I have enough components on the shelf other than a modern flight controller/autopilot) and use it extensively for photogrammetry as a proof of concept before pulling the trigger on a LIDAR sensor.
Regarding the IMU, do you think something like a CUAV X7 Pro plus https://www.cuav.net/en/x7-pro-plus/ might actually be accurate enough to avoid needing a separate IMU? Although I guess that would only work if the LIDAR sensor were hard mounted.
Which poses another question: is a gimbal for the LIDAR sensor an absolute necessity? I assume so, but it would be much simpler without it!