This is the last Sunday before the DRES regulations come into effect at the end of the month, so it’s the last opportunity to fly legally without an Operator ID on the aircraft. Well, actually it isn’t really, because BMFA members have an exemption until the end of the year, but I’ve got mine anyway.
When I got up this morning it was a little bit misty, which then got worse. As it’s always ten times worse when you’re standing in a field in the middle of nowhere, flying didn’t seem like a good idea. In fact, there was a moisture in the air from the mist that was making everything sopping wet. It wasn’t nice this morning, up until lunchtime when we had about an hour of sunshine before it went straight back to grey and murky again.
The picture you see at the top of this post is something that I’ve been working on all week. It’s a 3D printed drone racing gate as I’m attempting to replicate the Drone Racing League and Lockheed Martin’s Alpha Pilot competition, but done my way using micro drones and much cheaper technology. I’ve been all week designing gate bits that fit together in Fusion360 and I have to say that I’m starting to get the hang of CAD a bit more. Maybe I just needed to start with something simpler than a WACO SRE biplane? The gate above is modular, being composed of 12 straight sections and 8 angle joints of 45 degrees. It’s isn’t a regular octagon as I’ve made two sizes of straight section – one 9cm and one 12cm. The 9cm sections are the horizontal and vertical pieces while the 12cm sections are the diagonals. The beauty of this design is that each piece takes about an hour to 3D print and you can make different variants of square and octagonal gates by plugging different sections together. I’ve made a selection of 45 degree (octagon) joints, 90 degree (square) joints and 180 degree (straight) joints for extending the sides to make a bigger shape. There’s a limit to how big you can print on the 3D printer, so the ability to extend sections is very useful.
I found that the gate in the picture was a bit small for a 100mm quadcopter, although I did manage to fly through it without hitting anything on my very first attempt. After adding some more sections, the aperture is now 40cm across, which is about three times the width of the quadcopter. This seems like quite a nice size, so I think I’ll stick with it.
Having built my obstacle, the next challenge is to add some markers and build a computer vision system that can pick the flyable region out of an FPV camera frame. First off, I’m going to add my markers and take lots of shots of the gate in order to build myself a training set. I haven’t quite worked out how best to do this yet, so I’ll have to leave the details until later.
The other part of the plan requires being able to fly the quadcopter from software running on my laptop. This needs a FrSky XJT module connected up to the computer, but this is fairly standard and shouldn’t present any problems. I’ll be ordering the XJT module later this evening, so I’ll have some electronics to do later this week. The thing that’s worrying me is how much latency there is between the video being presented to the computer and it being digitised and processed. Humans have about a 200ms latency and the video that the computer sees is always going to be a bit behind real life, even with a low latency FPV camera. My theory is that the computer can see exactly what a real drone pilot sees through the goggles, so any other latency is down to the video digitisation and processing. I guess I’ll just have to experiment and see how it goes.
Anyway, that’s it for now. I’m off to put Operator ID labels on all my planes ready for December 1st.