UAV Controller Design

PID-Based UAV Pitch Autopilot with Sensor Fusion and Robustness Validation

🚀 Project Summary

Designed and validated a UAV pitch control system using PD/PID controllers and complementary sensor fusion.
Achieved <2% overshoot and ~3 s settling time under realistic sensor noise and bias conditions.
Demonstrated robust closed-loop performance across multiple feedback strategies and disturbance levels.

---

🧠 System Overview

This project models and controls UAV pitch dynamics using:

• Plant Model: Second-order pitch dynamics
• Controllers: P, PD, and PID control strategies
• Sensors:
  • Gyroscope (rate measurement with bias and noise)
  • Accelerometer (angle estimation with noise)
• Sensor Fusion: Complementary filter for reliable attitude estimation

---

🔁 System Architecture

Reference → PID Controller → UAV Model → Sensors → Complementary Filter → Feedback

---

📊 Key Results

Closed-Loop Performance (Best Case: Filtered Feedback)

• Overshoot: ~1.97%
• Settling Time: ~3.1 s
• Steady-State Error: ~0.0047 rad

![Closed Loop Comparison]

---

🔍 Sensor Behaviour Comparison

• Accelerometer-only: High noise leads to oscillatory response
• Gyro-only: Smooth short-term response but drifts due to bias
• Filtered (Best): Combines noise reduction and drift correction

![Sensor Estimation Under High Noise Conditions]

---

⚙️ Key Engineering Insights

• Increasing Kp improves response speed but increases overshoot
• Derivative action (Kd) improves damping and reduces oscillations
• Integral action (Ki) removes steady-state error but must be tuned carefully
• Gyro drift occurs due to bias accumulation during integration
• Complementary filtering balances noise rejection and drift correction

---

🧪 Robustness & Validation

System tested under increasing disturbance levels:

• Baseline noise
• Moderate disturbance
• Severe disturbance

Results:

• Raw accelerometer feedback degraded significantly with noise
• Gyro-only feedback showed increasing drift
• Filtered feedback remained stable and met performance requirements

![Disturbance Test]

---

✅ Requirements Verification

ID	Requirement	Limit	Result
REQ_01	Overshoot (baseline, filtered)	< 5%	PASS
REQ_02	Settling time (baseline, filtered)	< 5 s	PASS
REQ_03	Steady-state error	< 0.05 rad	PASS
REQ_04	Filtered SSE < Raw SSE	inequality	PASS
REQ_05	Filtered SSE < Gyro SSE	inequality	PASS
REQ_06	Overshoot (moderate disturbance)	< 10%	PASS
REQ_07	Overshoot (severe disturbance)	< 15%	PASS

---

⚠️ Limitations

• Simplified second-order model (not full aircraft dynamics)
• Complementary filter introduces slight estimation lag
• Performance degrades under extreme disturbance conditions

---

▶️ How to Run

pip install numpy matplotlib control
python uav_controller.py

• Generates plots automatically
• Outputs verification results
• Saves figures in images/

---

📁 Project Structure

UAV-Controller-Design/
├── README.md
├── uav_controller.py
├── images/
└── .gitignore

---

👤 Author

Saad Wajih