Dear Vinoth Kumar

There is two issues in this regard:

Hardware:

Depending on the volume of computation (which depends on the sampling time, prediction horizon, etc.), Arduino may not be able to perform the computation within the defined time ,so I suggest you to use more powerful processors.

Software:

Recently, several automatic code generation tools have been developed that produce efficient library-free optimization algorithms for real-time embedded applications. As an example, I suggest you to using these tools:

ACADO : http://acado.github.io/features.html

muAO-MPC: http://ifatwww.et.uni-magdeburg.de/syst/muAO-MPC/

...

In each package, examples are presented to run on microcontrollers or Arduino.

Best regards

That's great. Thanks very much for your valuable answers. Majid Shariat and Michael Short

Majid Shariat and Michael Short Still I am wondering, is there any other way to simulate the real time tank system with MPC. I was using the Arduino Mega 2560 but I got the computationally error. Would you recommend me to use some other arduino boards. Look forward to hearing from you.

Thanks,

Vinoth

Dear Vinoth Kumar

Following the previous answers, I'd like to recommend you another framework that also supports automatic (library-free) code-generation. It is called Casadi (https://web.casadi.org/) and you can use it from MATLAB, Python or C++.

On the other hand, regardind the board, you shoud try with an Arduino Due, Adafruit Feather M0, or any board based on the ARM Cortex-M* architecture. These boards are more (computationally) powerful than an Arduino Mega 2560. You can also try something even more powerful...like a Raspberry Pi or so.

You should also have a look on how are you scripting/defining your optimization problem. Things such as weights, redundant constraints, lack of constraints, lack of regularization, etc, may set an ill-conditioned optimization problem, which may be infeasible or too slow to converge.

Best regards

Dear Vinoth

You can use laguerre functions to make MPC more efficient and make it feasible for real time implementation. For reference you can see the paper " real-time implemntation of MPC on a 16-bit microcontroller for speedc control of a DC motor". Thanks