Drones move fast

We started building our first drone in 2011. I’d like to say we were the ~20 teams in India to build one from scratch. It was clunky and the frame had an imbalanced weight distribution. We used cheap kitchen utensils for parts of the frame! DJI still wasn’t selling any ready to fly drones – in fact, they had just demoed a new attitude hold mode for their Naza controllers. You could get a drone controller in the form of a board known as the KK Board, but using that wouldn’t be fun, would it? We built ours using an Arduino and a custom interface shield that connected to a 9DOF sensor board from Sparkfun.

Kitchen utensils used to build the frame.

We spent about 3 months trying to get air time after our first build – it never happened. Then we changed the algorithm to not read the accelerometers (which were too noisy). Another problem was having a pilot that knew how to fly a drone. Here’s a video of the first time we caught air, complete with gasps and celebration.

Another video of our tests. The drone sure loved eating plants.

Here’s a video of DJI testing their “new” algorithm at the same time.

Fast forward to 2018, and we have seen an explosion in the number of drones. It always amazes me how improvements in other segments of technology have assisted this, for example MEMS sensors. The noise ceiling for the accelerometers was too high in early sensors to allow for reasonable algorithms and systems. The problem had to be solved at a sensor manufacturer level!

In a similar way, our first drone (pictured above) had weight imbalance issues that the controller could not compensate for, and they had to be solved during the frame building process.

I watch some of the videos from DJI now, and fondly remember our first foray into this world. Some photos from 2012 below: