Graduate Controls and Optimization project (MEAM 5170 at Penn): an LPV-MPC quadcopter controller for cluttered forest firefighting environments. The full pipeline combines global planning, nonlinear quadrotor dynamics, and model-predictive tracking to fly the drone from takeoff through randomly generated obstacles to a fire zone.
- RRT* for collision-free global planning with smoothing and clearance checks
- LPV-MPC trajectory tracking via PyDrake convex optimization, solving a QP at every timestep
- Constraints on path tracking, obstacle avoidance, and actuator limits (thrust/torque)
- Nonlinear quadrotor dynamics with velocity, attitude, and clearance bounds
- Spline-defined fire-zone boundary as the target region
- 3D visualization of the full forest flyover, validated at multiple integration steps
Repo: https://github.com/cedrichld/mpc-fire-quadrotor
quadrotor_trajectory_TPV_ref_traj.mp4
quadrotor_trajectory_TPV_forest3.mp4
First_drone_video.mp4
