Today we had another of our hackathon/sprint days, this time involving James, Scott Mills (one of my fellow Sciencemob founders) and myself. Our aim was to see if we could hack together a cheap platform for aerial photography, using some easy-to-get gear.
Our shopping list was:
1 Parrot AR Drone - http://ardrone.parrot.com/
1 GoPro Hero HD camera - http://www.launchhelmetcams.com.au/gopro-hd-hero-cameras
1 Garmin eTrex10 GPS - http://sites.garmin.com/etrex/
1 roll duct tape
1 kitchen sponge and scissors
Some sheets of A4 scrap paper and a pen
All up, that’s about $650 - $700 worth of kit, depending on where you buy it from. So what did we do? Well, Scott’s video will tell the story.
And here is the end result; an image taken from the video frames georeferenced over a NearMap image, using the paper waypoints as reference. And yes, that's the three of us in the little inset part! You can also see where the cricket pitch lines up on this image - the old image shows the pitch looking a dark green, whereas in our image it was much more of a tan colour.
There are, of course, limits to what you can do in a day (or half day, as we processed this and put out a blog on the same day). But here’s our take on this in terms of pros and cons...
Stability and battery of the AR Drone – as a platform, for testing, this was great (especially for cost), but the stability of the AR Drone in wind is not great (even unmodified), and the battery life is quite short. The autohover also seems to over-correct, and well, we did crack the casing underneath when it fell from a great height (although it still works, as does the GoPro it landed on, thanks to the kitchen sponge!).
Wireless video – the onboard video is streamed to the device over the wireless connection, meaning the quality of the onboard video is not great (and why we strapped a GoPro to it in the first place). Given the camera is 320x240 resolution, you don’t really have much to lose…
Reading a mobile device in direct sunlight – it’s still very hard to read devices in direct sunlight. We had an iPhone, a Galaxy Tab and an Acer Iconia, and all were very difficult to watch the video on for trying to steer and control the AR Drone. So James, our pilot, instead was steering it visually rather than via the video stream.
Cost and simplicity – the system was put together with no real hacking, and a roll of duct tape and a piece of kitchen sponge (which meant it survived a pretty significant fall that seriously cracked the casing), in about an hour.
It Works – we got a solution out of the system.
So what did we learn? Well, we had fun, and thanks to Scott we have a good record of what we did and how much fun we had playing with it. This sort of thing leads us down all sorts of interesting routes for field work – such as us heading out into the field with one of these – or rather, something more stable - and collecting aerial photography in remote areas while we’re in the field. Imagine getting out of a vehicle in the middle of nowhere, setting up some control points, and flying this around for a few minutes, then having aerial views of your site in less than an hour. Combine that with our PDA field data collection solutions... and hopefully we've got some good value propositions for the field from this.
This is something I’ve talked about in my presentations for some time; cheap, portable aerial photos, and technology supporting field workers. With today's "hack", we’ve proved it’s possible on a shoestring. Of course, we have loads of questions and further places to take this but for now, I’m satisfied that we really are on the right track, and this is something that is worth more time and effort.
A big thanks has to go to James, who spent loads of his own time on this project practicing how to fly this beast, and to Scott, who gave up a day of his “holiday” here in Perth to come and contribute and film the day.