This topic is about applying reinforcement learning to control systems including robotics and general process control. The idea is simple, but profound — having a digital twin of a given mechanical system or process, make a control system capable of learning the appropriate behavior model for either performing certain tasks while following a certain trajectory under specific constraints. See the attached video for several examples of this.
Requirements to take this topic:
- At least basic programming knowledge;
- MATLAB/Simulink or Python experience;
- Basic modeling and control knowledge;
- Experience with deep learning problems, e.g., having passed some TalTech or other universities’ courses;
- Interest to work with Game Engines such as Unreal Engine 4 and Unity3D.
In case of interest, the student will be given a small test assignment to see if he or she fulfills the requirements to tackle this topic.