Drogon’s not dead! While it’s been over a year since my last post on the topic, I have been working away. Although I went most of the year without touching it, the pace has picked back up over the last few months and I’ve worked through a number of changes and improvements.
The only hardware change was swapping out the frame for a new Q450 V3 Fiberglass Quadcopter Frame (450mm) from Hobby King. It’s a bit smaller but lighter. So far I really like it. All the electronics are the same.
First, I had attempted to ditch the transmitter/receiver a bit too early in development, relying completely on wifi and the Raspberry Pi. This required a lot of support software that took away from the more important, and fun, flight control software. It ended up being too much of a distraction, so back on the receiver went. I re-integrated the receiver code and back on track I was. On the plus side I now have streaming video to an Android app from the Raspberry Pi’s camera. I also swapped out the Java code on the Pi to a new Python app (It’s all in my Github repos). All it’s really doing now is capturing logs from the Arduino and providing video streaming.
Next, I developed a self-tuning addition to the PID algorithm. This should hopefully be saving time and producing a better tuned PI. It still requires manual tuning to get it started, which I’m working through now, but progress is being made. I also re-calibrated the accelerometer-gyroscope relationship and how the data is pulled into a single position, so I am getting much better and cleaner data. I also incorporated rotational correction. This uses the Gyroscopes Z-axis by increasing opposite motors and decreasing the other opposite motors to control and take advantage of the motor/propeller torque.
The last major area was in test rigs. I’ve gone through a few iterations of test rigs. The first test rig I used, which is in my first videos, was tethering to the floor. This had several issues in not giving the quadcopter enough freedom of motion and significantly impacted flight. The second test rig I used was hanging from the ceiling. This was much better at providing more range of motion, but still had issues in giving it too much freedom while still having too much effect on flight. My current test rig is now a balance beam. It is basically two 2x4s perpendicular to each other, with one raised slightly. This works well with my current frame. When I get the PID tuning working well, I may graduate to the hanging rig (hanging, with pulleys to small weight to pull slack).
Here is a video of the testing I’m working on: