SmartPlanes unmanned survey and mapping

Some of you may have heard about SmartPlanes in Sweden. We have been quietly working with survey and mapping from UAV's for about 5 years now and keeping a bit of a low profile. I have just been putting a bit more information on our homepage about the kind of work we do and wanted to share it with you guys here at DiYDrones. I know a lot of you have ambitions of doing something commercial with UAV's in civilian applications and I'd be glad to share our experience.

Our plane was designed to be a portable system for aerial photography that can handle professional use in the field. We use a flying wing made out of covered EPP with vacuum formed polycarbonate fuselage that can be disassembled into 3 pieces for transport. It has an aft mounted electric motor and weighs about 1.1kg with camera and flys for about 35min on a 2500mAH flight pack. Our first version of this plane used a MicroPilot 2028g, now we use paparazzi autopilot. ask me why. It is hand-launch and manual landing. Flight planning is done by defining a single rotated rectangular block with dimensions and one waypoint. The actual navigation path is calculated in the autopilot code.

I think that aerial photography over small areas and single objects is a very good first civilian application for UAV's. We have seen a lot of interest in the technology, but the market is really undeveloped. For most mapping applications though, it is not enough to just take pictures from the air- you have to be able to generate a georeferenced photomosaic that you can use in a geographic information system. This is an area we have put a lot of effort into, within smartplanes and through some partners. We now have some software that allows you to create a good georeferenced mosaic already in the field on a laptop. For the more demanding applications that require true orthophotos and surface height models, we send the data in for processing.

We have found a lot of interest especially in making surface height measurements over small areas, because you can then measure volumes of things like gravel piles, sawdust, peat, rock aggregate, garbage, compost. Whatever people scoop up into big piles, they usually need to measure it and it is expensive to do it with conventional ground survey or aerial photography.

We normally cover an area of about 500x500m in a single flight because aviation regulations in most places require flying within line-of-sight. We can however join several flights into a large block for mapping larger areas. I think the largest block we have done to date had 30 blocks and almost 8000 photos. Very interesting to know it can be done, but at this size, laser scanning becomes more competetive.

Anyway I just want to tell you a little about our work and invite you to have a look at our homepage. Leave a comment here if you have any questions.

Views: 4991

Comment by Sky Monkey on September 16, 2010 at 7:55am
Very nice Steve!

Also, you said "Our first version of this plane used a MicroPilot 2028g, now we use paparazzi autopilot. ask me why. " OK, why?

And finally, how do you generate the height measurements?

Moderator
Comment by John Church on September 16, 2010 at 8:13am
Very nice operation.
I have the same questions- "Why" and also curious how you generate your topographic data.
Great work!

Developer
Comment by Doug Weibel on September 16, 2010 at 8:18am
How about some more info/pictures on your airframe? I'm very interested in the fuselage/wing attachement!
Comment by Steve Joyce on September 16, 2010 at 9:50am
Ok I'll tell you about this migration from MicroPilot to paparazzi. There is of course a pretty major cost difference (the "flyaway" cost for a MP-equipped plane if it goes down in a lake or something is pretty painful) but surprisingly that wasn't the main factor.

We were trying to build a custom solution for UAV mapping and it was really valuable to have complete access to the source code that you get with an open-source solution. In comparison, the MP is a general purpose autopilot with a fairly simple scripting language which didn't really give us the control we wanted over navigation and camera triggering.

It was an interesting experience to set up and tune both the MicroPilot and paparazzi in the same airframe. contrary to what you would expect, there wasn't a big difference in the effort required to setup and tune the two systems. In fact the paparazzi went a bit faster and flew autonomous on the maiden flight. That could have something to do with the order we did things though. When a question or problem did come up, I found it much better to look in the code or the schematics to see what is happening rather than going through tech support. that might be just me (and most people here I expect).

As far as performance goes, we were pleasantly surprised to get better navigation precision with the paparazzi/thermopile compared to MicroPilot/IMU. This is with exactly the same GPS module in the same airframe and the same guy doing the tuning. The big difference was in the turns: in part Paparazzi has a better algorithm for tracking turns, but I think relying on the horizon as an absolute reference works a bit better than an inertial reference in turning attitude. there are of course some weather dependencies with thermal sensors, but the solution has worked well in most weather you would take pictures.

the final factor that came into play was export restrictions on the MP hardware when they came into efffect in ~2007. It wasn't such a problem for us to import the hardware, but we couldn't resell it without going through export permits for each end-user. Something that added cost and above-all delay.

I shouldn't say all this when I am still trying to sell some barely used MP2028's!

/Steve
Comment by Steve Joyce on September 16, 2010 at 10:19am
When I say navigation precision, I mean something like this (flown with paparazzi/thermopiles Jan 2010 at -18C): It really makes data processing go a lot smoother.


Moderator
Comment by Gary Mortimer on September 16, 2010 at 10:38am
Well done Steve good to see you above the hedge again, the accuracy that can be obtained with thermopiles is great. I have had very similar experiences. Looking forward to another T3 entry from you!
Comment by Adrian Eves on September 16, 2010 at 11:31am
Excellent work!

Developer
Comment by Pete Hollands on September 16, 2010 at 11:54am
It would be interesting to know what laser system you are using.
Comment by Steve Joyce on September 16, 2010 at 12:15pm
Hi Pete, No we don't fly a laser sensor. The 3D measurements are done photogrammetrically, ie. derived from all the photos in software during post-processing. I'll try to explain how it works since a few people have asked.
Comment by Eric T on September 16, 2010 at 1:14pm
ya that would be great. Is the software that does the 3d photogrammetry commercial or some crazy program you guys have come up with? Congrats on the great work, looking forward to hearing more. I was looking into doing basically the same thing about 5 years ago but gave up due to the cost of MP and the overall complexity and my knowledge at that time.

Comment

You need to be a member of DIY Drones to add comments!

Join DIY Drones

Groups

Season Two of the Trust Time Trial (T3) Contest 
A list of all T3 contests is here. The current round, the Vertical Horizontal one, is here

© 2019   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service