Using quadrotors, students learn robot intelligence in various ways and also gain multidisciplinary knowledge in physics, mechanics, control, and information technologies (IT). This course aims to introduce classical model-driven and recent data-driven approaches for achieving control objectives. More specific and hands on experiments with a quadrotor system only will be provided to clearly understand abstract mathematical expressions.

- Basic physics
- Control Engineering
- Linear algebra
- Python coding.

Projects (90%) and attendance (10%)

- Lecture material

Week 1. Mathematical modelling
Week 2. Experiments and algorithms for parameter estimation
Week 3. Navigation
Week 4. Algorithms for inertial navigation systems
Week 5. Control systems
Week 6. Model based control algorithms
Week 7. Fusion with other sensor systems
Week 8. Hardware composition
Week 9. Deep neural networks
Week 10. Disturbance estimation with deep neural networks
Week 11. Deep reinforcement learning focussing on soft actor critic algorithms I
Week 12. Deep reinforcement learning focussing on soft actor critic algorithms II
Week 13. Experiments of hovering quadrotors with deep reinforcement learning
Week 14. Experiments of rolling quadrotors with deep reinforcement learning
Week 15. Simulation of flying inverted pendulums with deep reinforcement learning
Week 16. Demo presentation

- Lecture
- Team projects