This project aims to control a double inverted pendulum using model predictive control (MPC). The starting position is set to the origin of coordinates with both pendulums hanging down. The MPC approach is be able to, within physical limitations, bring the cart and pendulums in any arbitrary position while avoiding given obstacles. For the problem at hand a direct MPC approach utilizing orthogonal collocation is used. Additionaly a cost function is proposed that includes a quadratic set-point tracking objective. The algorithm developed is documented in a Jupyter Notebook according to the task requirements.
- Download and install Conda
-
Install conda enviroment:
-
Ubuntu:
conda env create -f environment.ymlThen:
source activate mpc_env -
Windows:
conda env create -f environment.yml
Then:
conda activate mpc_env -
-