Comment by Krzysztof Bosak on October 25, 2010 at 2:36pm

@TJB
Yes I made a joint-venture with Trigger Composites.
I needed a serious maker of a mature platform, after stumbling into limitations of amateur platforms.
So this way the autipilot has been tuned to the platform, and the platform was chosen to be the best for the task (this includes a mix of price and durability). After all those efforts, it appears that we made a few modifications here or there one couldn't do using off-the-shelf autopilot. It will gives us slight advantage now (pre-programmed missions), it will give us even more edvantage tomorrow. If the pracitce had shown a different fuselage woudl be more practical, we could have done it. So far we are really happy how it works together and happy to have dismissed classical landings in automatic mode (difficult to have high endurance, high lift and short landing). In short, a lot of lucky decisions have been done.

@F PIVOT
The parachute is made by Trigger Composites. We are doing everythin except small parts that could break:
standard equipment includes all propellers, servos, motors, ESC (yet we had to reject 80% of existing sdolutions after compatibility testing).
The parachute itself is not particularly hi-tech, not a single element looks cosmic in the whole UAV, what is difficult is modifying the autopilot to make parachuting operation putting as few stress as necessary, hundred times.
So we had 80% of good guesses based on good engineering practices, the rest had to be ironed out in tests.
Yes even paper parachute could work, most probably, a few times.
Square parachute reduces stress load compared to circular one.

Comment by Eric T on October 25, 2010 at 3:14pm

What program are you using to do the photo-stitching?

Comment by SciFly on October 25, 2010 at 7:20pm

Can you beat the LOS restictions by using a chase vehicle in order to cover more ground?

Comment by Krzysztof Bosak on October 26, 2010 at 1:26am

@Eric T Photo-stitching in this example was done by microsoft ice. But we have sub-contractors that can process the data, also if the user places reference points on the ground or knows landmark positions. This processing is inevitably very costly if you want to gave a few cm precision map. So on Polish territory you can contact TriggerComposites and we can arrange with airspace authorities and image processing company an operation that otherwise would be more costly and complicated with manned technology.

@SciFly LOS restrictions in some countries forig moving of the base station. In other countries you can have many spotters so you can fix them on the ground and they can relay 'John, it's ok' secret message. In Poland we did 19km long 2-way photomapping of motorway construction site by following it on a local road keeping within LOS without problem. We didn't blocked any traffic and ATC was quick to grant a permission because weather was too rainy for others to takeoff in the whole area.

Comment by Andrew Dunlop on October 26, 2010 at 2:50am

Fantastic work Krzysz. You say lucky guesses, but I am sure it is your comprehensive understanding of the subject matter that allows you to make educated guesses about what will work and what wont.

I hope the Pteryx is a BIG success.

Moderator Comment by Gary Mortimer on October 26, 2010 at 3:04am

Yes I will second that, well done, its not as easy as people think.

Comment by Krzysztof Bosak on October 26, 2010 at 5:09am

Thank you everybody for encouragement.
BTW The first screenshot of motorway mission are on the webpage. I will make a movie of it.
It appears beides making autopilot, UAV and arranging photo stitching, we will serve as mission consultants and operators on polish teritory as nobody looks ready for business until everything is known, clear and simple.
Our next step will be simple UAV ATC services, national communication network backup and Global Lost&Found Database ;-)

Comment by brakar on October 26, 2010 at 3:26pm

Amazing system, makes me wish I had a bunch of $$$ !

Comment by Mark Willis on October 27, 2010 at 4:13pm

Krzysztof, really sweet work! Are the stitched results orthogonal and georeferenced?

Comment by Krzysztof Bosak on October 27, 2010 at 6:43pm

@Mark
The motorway results are not georeferenced because this was for making and overview of works to be carried out.
They are not even orthogonal as we have used a prototype without stabilisation, but with extra batteries and 2 cameras. The camera that worked was in continuous shooting mode so the points should be remapped using timestamp, yet nobody took the effort.
The square map is orthogonal but again nobody cares about georeferencing.
Normally we have points from GPS and corresponding photos (roll (should be null on the head), pitch, course+heading estimate, time, lat, lon) that we can provide for processing software.
I havent finished GPS export but my logdecode makes approx photo placing in kml since the beginning.
Georeferencing can be done roughly by inserting and overlay into google earth. All depends on application.
The key idea is that UAV flew exactly where it was requested so the user just took a look at the soil and planned operations along the 18km long section accordingly.

Comment

• ‹ Previous
• 1
• 2
• 3
• 4
• Next ›
• Page