Robot Learning Laboratories

Labs done for the Robot Learning class. I wrote a report for each lab discussing the main intuitions, issues and the results.

Brief description of the labs, and links to the sources:

1. Implementation of the Extended Kalman Filter in ROS. Read the Report.
2. Reinforcement Learning fundamentals: implementation of the Linear Quadratic Regulator (LQR), a standard control strategy, to control Gym's Cartpole environment and then comparing the results with a basic Reinforcement Learning algorithm, seen as a black-box, where I designed and tested different reward functions. Read the Report
3. Q-learning: implementation of a tabular Q-Learning method to control to control Gym's Cartpole environment. Read the Report.
4. Policy Gradient Algorithms: study of the REINFORCE algorithm, its variant with the baseline, and two Actor-Critic methods: SAC, PPO. Read the Report.