Casadi Maiden Flight on PX4 Autopilot

[James Goppert]

As part of my PhD work, I'm using Casadi to test new algorithms for estimation on the PX4 autopilot (video link below). I will be releasing an open source library written in python for estimation and control soon on github that employs casadi. Contact me if you are interested in getting more details or for possible collaboration.

Casadi has shown promise on the PX4 autopilot by reducing the amount of time to compute sparse filtering equations vs matlab symbolic toolbox (from 3 hours to 30 seconds). Also the cpu usage is very low, I'm running an rk4 covariance and state prediction at 200 Hz and corrections at 50 Hz for mag, 200 Hz for accelerometer and still the cpu usage is at 8%. Also, being open source, people will no-longer need a matlab license to work on the filtering equations in the PX4 autopilot, which has reduced the amount of developer contributions in this area. I want to thank the Casadi development team for great work!

This is the maiden flight, we have already made even more performance improvements:

https://drive.google.com/open?id=1DcRDlnfx2Ls66Vdvx-EObBOsMZ0zVt2Y