quadrotor_mpc

Trajectory tracking with a quadrotor.

The non-linear dynamics are linearized about a linearization trajectory $\bar{x}, \bar{u}$ while the quadrotor aims to track a different reference trajectory $x_{ref}, u_{ref}$.

Quadrotor and reference trajectory

The MPC controller solves the following optimization problem :

$$\begin{aligned} \min_{x_{1:N},u_{1:N-1}} \quad & \sum_{i=1}^{N-1} \bigg[ (x_i - x_{i,ref})^TQ(x_i - x_{i,ref}) + (u_i - u_{i,ref})^TR(u_i - u_{i,ref}) \bigg] + \frac{1}{2} (x_N - x_{N,ref})^TQ_f(x_N - x_{N,ref}) \\\ \textrm{s.t.} \quad & x_1 = x_{\text{IC}} \\\ & x_{k+1} = f(\bar{x}_k, \bar{u}_k) + \bigg[\frac{\partial f}{\partial x} \bigg|_{\bar{x}_k, \bar{u}_k} \bigg](x_k - \bar{x}_k) + \bigg[\frac{\partial f}{\partial u} \bigg|_{\bar{x}_k, \bar{u}_k} \bigg](u_k - \bar{u}_k) \quad \text{for } i = 1,2,\ldots,N-1 \\\ & u_{min} \leq u_i \leq u_{max} \quad ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ \text{for } i = 1,2,\ldots,N-1 \\\ \end{aligned}$$

Trajectory tracking with MPC controller

Technologies

The project was created with:

• Python 3.9.0
• numpy 1.24.2
• cvxpy 1.3.1
• matplotlib 3.6.2
• seaborn 0.12.2
• pandas 1.5.3
• jax 0.3.25

Installation

To use this project, install it locally via:

git clone https://github.com/elena-ecn/quadrotor_mpc.git

The dependencies can be installed by running:

pip install -r requirements.txt

To execute the code, run:

python3 main.py

License

The contents of this repository are covered under the MIT License.