Quadcopter drones are widely valued for their ability to hover, maneuver in confined spaces, and integrate low-cost autopilot systems. However, these capabilities depend heavily on the quality of their attitude control system. One of the most critical movements in this system is pitch—the upward or downward rotation of the drone’s nose—which directly affects forward and backward motion.
In many UAV platforms, pitch control is managed using a PID (Proportional–Integral–Derivative) controller. PID remains one of the most practical solutions because it is computationally lightweight, easy to implement, and adaptable in real-world flight tuning. Yet, its success depends entirely on choosing the right gain values. Poorly tuned gains can create sluggish responses, oscillations, or weak tracking performance.
The study, titled Flight-Log-Based Performance Analysis of a PID Controller for Pitch Response in a Quadcopter UAV, examined the effectiveness of an initial PID controller by analyzing real-world flight log data. Instead of relying on simulations, the researchers used an empirical dataset exported from UAV Log Viewer, focusing on how closely the drone’s actual pitch followed the commanded pitch during flight.
The team analyzed a valid flight segment lasting 89.96 seconds, covering 2,649 recorded samples. Three main variables were observed:
- Desired Pitch – the target angle commanded to the drone
- Current Pitch – the measured angle during flight
- North-East Position – horizontal movement used to assess spatial stability
The initial PID gains applied in the pitch loop were:
- Kp = 0.8
- Ki = 0.1
- Kd = 0.05
These values were treated as a starting configuration rather than an optimized setup.
The results revealed a major gap between command and response. The desired pitch ranged from -1.34° to 4.14°, while the measured pitch remained confined between -0.73° and -0.50°.
In practical terms, the drone barely responded to changes in commanded pitch.
The quantitative indicators reinforced this conclusion:
- MAE (Mean Absolute Error): 1.65°
- RMSE (Root Mean Square Error): 2.04°
- IAE (Integrated Absolute Error): 148.46°
- Pitch Correlation: 0.03
- Maximum Absolute Error: 4.81°
These values confirm that the controller produced weak command-following performance, even though the drone stayed airborne and operational.
Interestingly, the horizontal trajectory remained bounded throughout the test. The drone covered approximately 35.32 meters northward and 16.65 meters eastward, with a total path length of 109.75 meters. This suggests that while the control system did not achieve precise pitch tracking, it also did not lead to unstable or divergent motion.
That distinction is crucial.
A drone may appear calm and stable in flight, but still fail to execute commands accurately. In mission-critical applications, such hidden performance gaps can reduce effectiveness and safety.
“The initial gain configuration was safe for diagnostic data collection, but inadequate for responsive pitch tracking,” the research team noted in their publication.
The study also demonstrates the value of flight-log-based assessment. By reviewing historical flight data, researchers and engineers can diagnose control issues without requiring immediate complex mathematical modeling. This approach is practical for academic laboratories, prototype testing, and industrial UAV development.
However, the researchers acknowledged several limitations. The analysis was based on only one flight dataset, without repeated trials under identical environmental conditions. It also focused solely on the pitch channel, excluding actuator outputs, motor data, and disturbance factors such as wind.
For that reason, the study should be viewed as an initial diagnostic evaluation rather than final controller validation.
The authors recommend retuning the PID gains to improve control authority and responsiveness. Future work should compare multiple gain configurations, include actuator and motor-output data, and repeat tests under consistent conditions to measure improvements more reliably.
For the drone industry, the study serves as a reminder that controller tuning is not just a technical adjustment—it is a foundation of flight safety and operational accuracy.
0 Komentar