Especially because coding Kalman filters isn't that hard in C++. I coded a general Kalman filter templates at work starting with my own vector and matrix library.
Trick: Compile-time matrix and vector sizes through templates. Make everything inline such the compiler can optimize all the inner objects away.
Posted Feb 7, 2013 10:36 UTC (Thu) by dragontas (guest, #89225)
[Link]
It is not a simple Kalman filter. The presented Kalman Navigation filter calculates the position from delta ranges and GPS raw data. The tightly coupling with sensor data allows a precise calculation of the GPS clock error and error drift. With this technique it is possible calculate a navigation solution even if only 3 satelites are visible.
On the other side nonlinear control techniques with neural networks are used (dynamic inversion) to control the airplane.
All this high level algorithms are usually programmed and evaluated in matlab by engeneering scientists. This is the reason for using Matlab/Simulink Coder with RT-Linux hardware to create a one click solution for testing the algorithms in flight.