Aerial Survey and Photogrammetry with X5 & X7

We have been working to develop a well oiled solution for Aerial Survey and Photogrammetry over the last few moths. While there are a number of off the shelf solutions out there which produce great results, we are looking for something that will cover larger areas at a higher resolution. We started with the X5 as it could be hand launched and was a good platform that flew at higher airspeeds and could progress us into the X7. Here is some of our progress;

 

The X5 is running on APM 2.6 and has prove extremely reliable and stable. The camera used is a canon IXUS 132 running CHDK and shooting on an intervalometer script. The main purpose of the X5 was to learn about vibration dampenning and apature and ISO settigs to produce clear images. This took a significant amount of effort, however we are at the stage where 2 out of 3 images are photogrammetry quailty.

 

Our test baby has had a few hard landings (things were laid out a little neater when we started) . We are very excited about the recent interest and work being done on laser altimetry in recent months. We are also currently testing one of these devices as flying wings are extremely hard to land manually (especially when they are loaded up with gear). You can see the APM in the middle of the plane, 5.8GHz live video and 900MHz telemetry transmitters, and the IXUS 132.

 

We found the X5 reasonably easy to hand launch WITH A WELDING GLOVE (it will cut your hand to shreds - not if...when). You can see the camera poking out of the front of the plane. Images are good, however, to get the plane to navigate the survey grid well, it becomes a little twitchy which means that a high number of the images are not taken pointing straight down. Navigation accuracy (Nav Constants and Pitch and Roll constants)  v's image quality is a trade off. Others may have found the perfect solution..we haven't. (Yes we are investigating some of those gimbals). We found the ideal cruise speed with a 4S 5000mAh battery was about 12m/s and we are getting about 30 mins of flight time. 

We have been using Agisoft Photoscan which works really well and appears to be more flexible than Pix4U. We have no surveyed ground control in the images, and we have no geotags in the photos as we were running an intervalometer script. However, it was great practice flying and processing, changing settings and trying again. You cans see that you lose images when the camera is not pointing down and loose some of the survey area. We have found that it is better to not use bad images and have gaps, rather than use them and distort the good data.

Once we had some confidence in what we were doing, we unleashed the BEAST!

First of all, the X7 is more streamlined than the X8, but doesn't have anywhere near the same room for hardware. If we were to do another flying wing, we would try the X8. We are running the X7 on Pixhawk and have a nikon J2 taking photos. The J2 is triggered from the Pixhawk via an infrared LED shutter controller. 

The Nikon is snug in expanding foam with its lens fitting nicely into the camera hole in the X7.  With decent batteries and a big camera, you simply can't launch the X7 by hand, so we have built our own heavy duty catapult powered by 4 spear gun rubbers;

It took a while to settle down the nav and flight constants for the X7 but it flies beautifully once tuned. It cruises at around 15m/s and is far less twitchy than the X5. Photos are higher resolution and a lot sharper. We love it as an aerial photogrammetry platform, however landing is still an issue. As well, as we are only running 4s batteries, it's always touch and go as it comes off the end of the catapult. We are still playing with battery/motor/prop combos to get the optimised take off and duration. 

Again, using Agisoft Photoscan, (with Geotags, but no ground control), the results were extremely pleasing;

We are still playing and tuning, but our focus at the moment is;

1) ground control -we are looking at a network gps system where you can drop a bunch of transceivers around the survey area and have them log back to a central PC where only one of the transceivers is located in a surveyed location.

2) AUTO LANDING....WOULDN'T IT BE NICE ! 

3) We have purchased a 757 Ranger to try slower speeds, but significantly longer endurance to see if it improves the sharpness of the images. 

Anyway, this is what we are up to. Would love to hear comments from anyone else out there chasing photogrammetry.

Views: 9035


3D Robotics
Comment by Chris Anderson on July 5, 2014 at 9:57pm

Great post -- impressive results. Good tip about the welding glove! Autolanding works very well with our X5s and X7s.  Have you tried it yet? 

Comment by ausdroid on July 5, 2014 at 10:09pm

Chris, yes, we have tried autolanding on the X5 repeatedly. The problem we find is that after 20 - 30 minutes in the air, the zero altitude pressure has changed enough to stuff up the landing. We have also found that if you don't bring the flying wings in fast, they stall and roll. I would be very interested in any advice. However, more interested in what is happening with the lasers. We are playing with the SF02 and would be happy to test any code that developers come up with.


3D Robotics
Comment by Chris Anderson on July 5, 2014 at 10:21pm

Ah. We generally don't have them up more than five minutes or so. Agree on the need for speed on flying wing landing -- those tip stalls at slow speed aren't pretty. 

LIDAR-Lite support should be coming in a few weeks. That's the best solution we've found for accurate ground detection.

Comment by Antonie Kruger on July 5, 2014 at 10:32pm

HI Guys, nice post - thanks for sharing.

Is the plan with the ground control network to be able to post process the exact position of the camera at exposure time?

Would it not be simpler to place control on A3 size targets on the ground and coordinate these before you fly with a RTK GPS?

Antonie


Moderator
Comment by Gary Mortimer on July 5, 2014 at 11:16pm

I found the X5 at altitude and speed to be a horrible platform. Its ok for a fun wing but put any payload in them and they are horrid. Gene had issues with the X8 and we concluded between us that the aerodynamics of these wings are sub optimal. Having much better success with the XUAV Talon in so much as I will say its my favourite platform ever! It can easily fly 250 hectares a flight. It is easy to hand launch.

I never autoland and autotake off. I like to feel that the platform is flying correctly when I take off before I commit it to auto flight. I often only have tracks to land on so I need to be in practice hand flying. In fact one in five flights are only circuits so I keep my thumbs tuned. Can't help thinking that many crashes are caused by peoples lack of situation saving hand flying practice.

Do you really need to carry all that FPV weight as well?

 

Comment by Martin on July 5, 2014 at 11:25pm

I noticed that on the second picture the magnetometer was mounted directly above the motor...

Comment by Simon Wunderlin on July 6, 2014 at 12:51am
Why hasn't anybody tried to attach the same barometric sensor to the ground station and then sync pressure before landing over mavlink?
Comment by ausdroid on July 6, 2014 at 1:29am

Antonie,

The plan is to have A3 size targets with solar powered GPS which are hacked to pull the raw data. These GPS units will all talk to a base GPS unit located at a surveyed position. Knowing the error at the surveyed position, we can then apply that error to the mesh units so that we have lots of ground control from only one surveyed position. We need to do this as we can not get access to some of the areas we will be surveying. The plan (and it is only a plan at this stage) is to drop these units (or place them) with a multi rotor.

Comment by ausdroid on July 6, 2014 at 1:30am

Martin,

Yes it's above the motor, but it's on a 6 inch mast.

Comment by ausdroid on July 6, 2014 at 1:32am

Simon,

Good Question. We have also wondered about this. I guess you would have to calibrate the ground unit against the air unit. However, the laser altimeters will be brilliant once they are sorted out.

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