Curbar AIControl fundamentals, MPC, and deep RL for robotics.
An in-development robotics course built around one practical contrast: classical control versus learned control.
The planned syllabus moves from mechanics and control fundamentals into model-predictive control, deep reinforcement learning, and side-by-side comparisons on environments such as inverted pendulums, Reacher, and Hopper.
The aim is to build a clean bridge from traditional engineering intuition to modern robotics control methods, rather than treating them as separate disciplines.
The current robotics syllabus in Notion is structured as a progression from control and mechanics fundamentals into model-predictive control, deep reinforcement learning, and side-by-side comparisons on standard benchmark environments.
Establish the course framing and the central contrast between classical and learned control.
Build the modelling tools used later in both MPC and learned-control sections.
Introduce MPC as a strong classical baseline, then work through concrete robotics examples.
Move from classical controllers into policy-learning methods for harder control settings.
Compare classical and learned methods on matched environments rather than treating them as separate worlds.