You said correctly, I can be used self tuning I think it's a part of adaptive methodology. these methods have some advantages and disadvantages. I also have many papers related to design self tuning controllers, You also can follow my papers.
There is not a better control. However, passive schemes avoid leakage of energy. Also, if you need robustness to disturbance effects sliding mode techniques could work, by example integral sliding modes schemes. There exist miscellaneous approaches that combine these two concepts.
Thank's for your answer. You said SMC can help me to have a robust controller but it has some challenges such as chattering or equivalent problem. Do you have a better choice?
I have more than 70 papers related to SMC methodology, you can follow my papers.
There is no a best one, it depends of many parameters. Consider energy, signal control amplitude, actuators speed, admissible sampling rate, etc, etc, etc. You need to do many simulations and mathematical analysis. That works ;)
Flatness-based control (and its advance version : Exactfeedforward linearization). Definitively.
- V Hagenmeyer, E Delaleau (2010) Robustness Analysis with Respect to Exogenous Perturbations for Flatness-Based Exact Feedforward Linearization IEEE Trans. Automat. Control., vol. 55, pp. 727-731
- V Hagenmeyer, E Delaleau (2003) Robustness Analysis of Exact Feedforward Linearization based on differential flatness Automatice, vol. 39, pp. 1941-1946.
- V Hagenmeyer, E Delaleau (2003) Exact feedforward linearization based on differential flatness, International Journal of Control, vol. 76, pp. 537-556.
Article Robustness Analysis With Respect to Exogenous Perturbations ...
I don't know the detail of your model. But so far as my knowledge is concerned, robust controller based on signal compensation is the most practical way to control an unknown plant. The idea of implementation is rather simple by regarding all model uncertainty and external disturbances as an overall signal, and suppressing it by output error feedback. See publications by Prof. Zhong, Y.-S. from Google Scholar.
Maybe you can have a look and even try Simple Adaptive Control (SAC) in my papers and their references, with specific applications to robots. Classical Control designers have observed that adding simple adaptive algorithms add performance and robustness with uncertainty.