Nowadays, industries are looking for an efficient, fast and reliable control method at a reasonable cost to control any electromechanical machine, i.e; robots. Therefore, a question is raised, "Can machine learning algorithms or genetic algorithms replace traditional control theory?"
This might happen only when an AI can design a stable (traditional) control system that can be verified mathematically. Even in this case, it will take years for industry to adapt that method. Think about it. Will someone use a complex AI for a task for which a simple PID (or some other) controller is sufficient? It will be waste of time, effort, (maybe even money) considering the time you spend to develop the AI or people you hire for that or the tools you need to use. Never say never, but as Aleš Kralj has said above, it is unlikely for now.
They are tools that can help in the control design process. I do not believe so.
I don’t think so, but we can use AI methods to tune the gains for any traditional controller such as state feedback control.
- Badges
- Science topic