This repository presents a simulation-based control framework for trajectory tracking of a quadrotor under aerodynamic influences. Initially focused on Proportional-Derivative (PD) control, the project implements gradient-based optimization for fine-tuning controller gains and sets the foundation for future research in robust control using Model Predictive Control (MPC).
🎯 Objectives Model the full nonlinear dynamics of a quadrotor.
Design and implement a PD control strategy for trajectory tracking.
Optimize PD control gains using gradient-based optimization to minimize trajectory tracking error.
Analyze controller performance on multiple reference trajectories: Circular, Helical, and Lissajous.
Lay the groundwork for future implementation of MPC-based control.
⚙️ Methodology
-
Dynamic Modeling A nonlinear 6-DOF model of a quadrotor is developed, incorporating translational and rotational dynamics under external disturbances.
-
PD Control Architecture A traditional PD controller is applied to regulate position and orientation. Controller performance depends on the tuning of Kp and Kd gains.
-
Optimization of Control Gains An objective function based on the Integral of Squared Error (ISE) is minimized using gradient-based optimization to identify optimal control parameters for each trajectory.
📈 Simulation Results Tracking performance was validated on three complex trajectories:
Circular Trajectory – Demonstrated consistent convergence to reference path.
Helical Trajectory – Showed stability and smooth elevation tracking.
Lissajous Trajectory – Captured dynamic path changes with reduced overshoot.
✅ Conclusion The optimized PD controller successfully enhanced the quadrotor’s tracking accuracy across diverse trajectory types.
Gain values significantly varied with the trajectory, highlighting the importance of task-specific tuning.
The current PD-based solution provides a strong baseline for robust trajectory tracking.
🔮 Next Steps: Model Predictive Control (MPC) To overcome the limitations of fixed-gain PD controllers, future work will implement Model Predictive Control:
MPC can handle multi-variable systems and constraints in real time.
It allows predictive optimization over a moving horizon, enabling the system to adapt to changing dynamics and external disturbances.
This approach is expected to offer robustness, adaptability, and optimal performance in uncertain and complex environments.
👤 Author Neelkumar Subhashbhai Ahir Supervisor: Prof. Garima Bhandari