The course covered theoretically far more, but the following methods/algorithms were implemented in practice:
- Localization: EKF / Particle Filter (Odometry Motion Model, Landmark Sensor Model) on a car in PyBullet.
- Motion Planning: A* / LPA* / D* / RRT / RRT* on 2-DoF/3-DoF robotic arms with scene obstacles, RRT* on a non-holonomic car in PyBullet.
- Reinforcement Learning: Behavior Cloning / DAgger on Reacher, Policy Gradient on Inverted Pendulum, all of which are parts of MuJoCo environment.
This repo content discusses some observations and analyses of the methods' performance based on different parameters, reinforced with visual proofs (plots, screenshots, GIFs)
The code for the methods is NOT provided due to the course policy