Hello all,
We at Airborne Projects are in the process of testing an integration of the Sony Alpha 5000 with the APM/Pixhawk with very good results so far and we are starting this thread to showcase and get feedback on a possible product.
The capabilities we have so far is to have the Pixhawk connect to a 53x25mm micro computer which then connects to the Sony Alpha (or actually most DSLR cameras). This way we listen for triggers from the Pixhawk and order the camera to take a picture and upload it to our micro computer. The images will be live Geo-referenced.
Upon landing there is a normal USB memory pen connected to this micro computer that has all the images by the APM ready for post processing, there is no need to cross reference the images with the Pixhawk logs.
Another cool trick we pulled is that we can configure the camera in the air or even see the live pictures on the ground. This way you can correctly figure the settings of exposure and ISO without the trouble of going back to the office. We plan to use this features as addons.
What do you think?
You can also find the original pictures that generated the mosaic in this link
Replies
Hello Roger,
Thank you for your thorough answer but I think it misses the point in what we are proposing. I also think that if you can do better you are always free to reach up to the market. No one is diminishing anybodies accomplishments here.
What we "invent" or as I prefer, innovate, is that you can put the pen in the companion compute, fly, and when you come back to the ground you just take out the pen from the companion computer and your images are already geo-tagged. Previously you had get the logs from your APM and cross-reference them with the pictures in the card of the machine. You completely missed this point and this is why I think you missed the point. It is not a problem for you, but this doesn't mean it is not for anybody else.
Your mechanism of triggering the Sony Alpha is just one way of doing things and it does just that. It doesn't give you any feedback if the picture was actually taken or not, while we actually do. The picture may not be taken for many reasons, like the camera not being ready. Plus, again with your mechanism you are back to where everybody is with most cameras: a trigger and then cross reference the logs in the ground.
This might be good for you but believe me that for many people, specially professionals of mapping, having access to the logs or connection to a UAV is not in the list of things they care and want. Value is relative.
The Sony Alpha A6000 is a different camera than A5000, similar but not the same. The stripped down version is mostly for weight. Again that may not be valuable for you but it is for others. Also for the stripped down version we, as a company provide warranty for what we sell.
The price of the strip down is about 800 Euros and the stand alone system of 200 Euros. You compared your system without cameras for 65 Euros with 800 Euros that *includes* the camera. I am sorry but that is indeed bashing and intellectually dishonest.
The companion computer for vision processing and artificial intelligence you have in mind is surely not the one i described. Take a look at the link i posted. Comparing maybe a Tegra or Odroid to a 400Mhz Atmel is again not honest.
Finally the big difference about your setup is that the viewfinder is not the same image that gets taken on the Sony Alphas. So your system of video is only good enough to see where it is pointing. If the pictures are dark or blurry you won't notice until you see the pictures. But this is completely besides the point because what we are proposing and trying to get feedback with this post is for:
The capability for the UAV operator to just hand out the USB pen with the images on site to the mapping provider.
About the ESA incubation it is completely besides the point because it is for a project that is completely unrelated to this.
PS: Your idea of uploading directly to the cloud is a huge idea and is really easy to implement with our technology.
Good flights
Paulo Neves
Alright, fair enough, and sorry if you took it as a bashing as I was afraid, but was happy for myself for my achievements and definetely was understood by you as arrogance.
Once again I apologise and now I promise I'll keep on topic with a more objective critique.
But let me tell you yet first that in this area there aren't bad jobs, just jobs that are ahead of too late in time, thus your work isn't bad at all because it does what it promises, that's a fact!
But going back to the topic.
Firstly, I made confusion with your 800 euros and your 200euros standalone system, for I thought the standalone didn't included the camera, so the client would have to install it on their own. It wasn't clear then, but it is now, but for sure wasn't at all intellectually dishonest. Thank you for the explanation.
Then about the fact that one can simply remove the memory card with the pictures already georeferenced is undoubtly most useful! It'll save time, specially for professionals that are too lazy to do their tasks like you've said, a couple of minutes, in georeferencing it in the office or in a laptop in the field. Nevertheless it's a win for your system without any doubt.
And about the live video feed from the camera? No doubt it helps detecting unacceptable pictures, but there are several problems that make (please don't be offended) 800eur useless. That is, can you make sure the pilot and/or another guy in the team (remember, bigger team, bigger costs) have their full attention to the quality of hundreds or thousands of pictures? (Attention, I'm considering that your quality control is human dependable only, without any processing. If you have other effective means of control then great!) And if you guys detect an unacceptable picture, what will you do? Mark down by hand the coordinates of the picture and then have one of two options, interrupt and sacrifice the programmed flight mission to make again that failed picture or will you make later another flight mission just for the eye detected bad pictures.
But then again, for 800 euros just for buying time, wouldn't a client rather have the images real time in the office ready for processing and quality control of the images (if they are blurred, etc, etc) and also transmiting the position and timing, and having them georeferenced by the time they keep arriving, and having it all READY for data crunching even before the plane lands on the grass. And if there are rejected pictures by someone watching carefully them in a BIG high resolution screen saved already in their harddrive in the office simply makes a new flight mission to target the handful sites that were missed by the bad pics, uploads to the pilot in the field that could reupload the new waypoints perhaps even before landing the plane from the first mission.
Now THAT is something worth paying, and what's funny is that's exactly what I've been doing for years, for 65 eurobucks of hardware that simply works. And if I downgraded to a slower shutter (and therefore a much much much more blurry susceptible) A5000 or A5100 I could even save 10eur by reusing the same cheap led trigger.
It simply works even if I don't have the deep knowledge of the insides of a camera (besides the classic parts that make a camera), because in the end I'm just interested in having the best quality damn pictures as possible for the least cost for me and as fast as possible.
And if you count the time saved by both methods (yours and mine) you'll see which one is not only most cost efficient but effective. It's just a matter of the Market to decide which will it take.
As a matter of fact, because I didn't have thought about this until right now that I'm talking to you (and for that I thank you again), I've learnt SO many things from the comunity that allowed me to have "innovated" so much by my self, that now I feel obliged to give back my innovations.
And as soon as possible I'll release free of charge to the comunity all the documentation of my "gadgets" and workflows and code, so that anybody can benefit from it and have more efficient mapping missions, because the hardware is really that cheap and easy to mount and implement even by newbies provided it's well documented.
I understand that it'll be like a little competition to your product, but what's the best motivator than competition? I hope that my ideias could even help in improving your product.
Next time that I'll go to Coimbra perhaps we could join and talk to share some ideas and perhaps business.
Best regards and bons voos para si também! ;)
Roger Monongo
Great idea,
what about the availability?
If you should need it, I could test it in difficult conditions since I usually work on mountains above 3000 metres ASL and not rarely above 4000 metres.
Hi ActionDrone,
What do you mean by availability?
That is a really interesting proposition. At that altitude do you not have any problems with batteries cooling too much?
Yes, sometimes I had problems with the temperatures, but we are thinking about solutions to this problem, and we actually found something to get things better.
I am trying to do this myself, but I'm unsure how to access the camera files while the pictures are being taken. How fast are you able to trigger the camera? Do you plan this to be integrateable with other Sony cameras? We are using an a5100 right now. We were looking at the a6000 for hotshoe ability, but the a6300 is a possibility too.
How many frames per second? What would your price point be?
Hi Thomas,
The time it takes to trigger is about 100ms and the time between pictures is about 1 frame per second, but it depends on the size of the picture. It is also possible to configure bracketing pictures with this particular camera so you can get burst modes, but we have no tested it as we did not see a need for it. If you are looking for faster FPS a Nikon DSLR is more appropriate.
We have not tested this with other Sony cameras as we do not have them on hand, but we are almost sure they would work similarly. To be specific we actually used the Sony Alpha 5100 not the 5000 ;). We have also got our software working with DSLRs like the Nikon D3200 where we even have live video feed.
We are looking for a 200 Euros price point just with the stand alone micro computer ready to connect to the camera and the APM, although we are working on also selling stripped down 5100's (battery connected to UBEC and no LCD) and the price point will be about 800 Euros.
I will try to post more pictures of our setup. We designed a 3D printable gimbal for the Sony Alpha 5100 if anybody is interested.
Hi Paulo,
Thanks for sharing your progress with the A5100. I would love to see the file for the 3D printable gimbal for the camera. I'm not sure if you would post here or need an email to send the file to.
Thanks so much again! mb
Which libraries do you use on the companion computer?
Do you control everything via the USB port? Or do you need the Wifi too?
Hi Christian,
Unfortunately I cannot yet disclose that details. We have developed upon open source libraries and modified them to match our needs. When we release the product we will publish the modifications we did and it will also become clear which libraries we used. I hope you understand.
Everything is controlled through USB, no WiFi is involved.
-
1
-
2
-
3
-
4
of 4 Next