A non linear cartpole balancing system was provided. The objective was to balance the cartpole at a single point. Hence, the system was linearized about that single point. This linearized system was used to solve for the infinite horizon controller using LQR.
A similar non linear helicopter hovering system was provided. The objective was to stabilize the hovering system. The infinite horizon controller was determined for the linearized system.
The objective was to stabilize and observe the hopper under different levels of perturbations for a set of points.
Given a non-linear system, the objective was to trabsform the system into a LTV setting and run LQR. A reference trajectory was provided, that is approximately feasible.
Refer to requirements.txt to install all the necessary dependencies. To set up Anaconda environment, refer https://www.anaconda.com/ For Mujoco-py installation, refer https://github.com/openai/mujoco-py