This engineering note is the first of two parts:
Part 1 Design and Simulation.
Part 2 Real-World System Realization. (Being written)
It aims at demonstrating how you may use Matlab/Simulink together with Rapid STM32 blockset and ARM Cortex-M3 processors (STM32) to develop digital signal processing systems; using a tilt sensor as a case study.
It covers the development process from design, simulation, hardware-in-the-loop testing, and creating a stand-alone embedded system. The content is supposed to be as simple/introductory as possible.
In this first part:
1. The motivation for using Simulink for embedded system development is explained.
2. A simplified model of tilt sensor system is developed.
3. Kalman filtering and Unscented Kalman filtering (UKF) theory is summarized.
4. Graphical instructions are then provided to guide you through the whole process of implementing a Simulink model to design, simulate, and evaluate the performance of an UKF for a tilt sensor system.
Note: Source code is also provided to perform Monte Carlo simulation based on Simulink model to evaluate UKF performance using covariance analysis.
In the second part, graphical instructions will be provided to guide you through the process of transferring your design from Simulink model to real-world stand-alone tilt sensor system based on Rapid STM32 - R1 Stamp board.
Visit www.rapidstm32.com for more information.
Thank you for your kind comment.
I have not the time to finish part 2 - the real-world implementation example.
However, please visit our site at www.aimagin.com to learn about our hardware and software code generation tools for microcontrollers.
Specifically, this is the tutorial for getting started: http://aimagin.com/blog/waijung-tutorials/.
Thank you very very much, you are the best one who explains and simplifies KALMAN filter in this clear way. Could you please send me the part 2 upon been ready. it will be very helpful also, thanks again.
Thanks your model helps me a lot ...if you have extended document please e-mail to...
Add seed1 and seed 2 declarations to PreLoadFcn callback so the model can run stand-alone.
Change the Title and add a link to another introductory note on Kalman filtering at