UAV Adaptive PID Gain Scheduling for Changing Payloads
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing UAV flight controllers face challenges in deriving suitable tuning constants for PID controllers, especially when carrying and dropping payloads, as detailed models are often unavailable, and gain scheduling is difficult due to changing physical characteristics.
Innovation Solution
Implementing a trained machine learning model to predict tuning constants for PID controllers based on real-time data, allowing gain scheduling to adapt to payload changes without prior knowledge of payload weight or detailed models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a detailed model of the UAV is used to tune PID controller, then suitable tuning constants can be derived, but a detailed model is not always available and the system becomes more complex
Solution Approach 1:
The system performs self-identification of its own dynamic characteristics by analyzing flight data collected during normal operation. The microcontroller automatically generates the UAV model and derives PID tuning constants without external intervention or detailed manual modeling, allowing the system to tune itself based on actual flight behavior.
Solution Approach 2:
The patent replaces the traditional mechanical/modeling-based tuning approach with an automated computational system. The microcontroller uses algorithms to process flight data, identify system dynamics, and calculate optimal PID parameters automatically, substituting manual modeling work with automated electronic computation.
2Adaptability or versatility
If gain scheduling is used to adapt to different operating conditions, then controller performance improves, but gain scheduling becomes difficult when payloads change
Solution Approach 1:
The system dynamically adapts PID tuning constants based on real-time flight conditions and detected payload characteristics. Rather than using fixed gain schedules, the microcontroller continuously identifies current operating conditions and adjusts parameters on-the-fly, making the system adaptable to unexpected payload changes without pre-programmed gain schedules.
Solution Approach 2:
The system uses feedback from flight data sensors to detect changes in operating conditions and payload characteristics. This feedback loop enables the microcontroller to automatically adjust PID parameters in response to actual system behavior, making gain scheduling responsive to real-world conditions rather than relying on pre-defined schedules.
3Ease of manufacture
If PID controller parameters are tuned offline, then suitable values can be determined, but the system cannot adapt to changing conditions during operation
Solution Approach 1:
The system performs self-tuning during flight operations by automatically identifying its dynamic characteristics and calculating optimal PID parameters. The microcontroller continuously monitors flight data and adjusts controller parameters without requiring external intervention or pre-offline tuning, enabling both initial setup and ongoing adaptation.
Solution Approach 2:
The system collects and analyzes flight data during normal operation to prepare and update the UAV model and PID parameters in advance of future operations. By continuously learning from accumulated flight data, the system proactively adapts to changing conditions and prepares optimal control parameters before they are needed.
Data Source
AI summary
Techniques for an unmanned ariel vehicle (UAV) having a closed-loop controller that is gain scheduled based on a trained machine learning model. The closed-loop controller could be a PID controller. Real-time data pertaining to the motor being controlled is input to the trained machine learning model. Examples of the real-time data includes, but is not limited to, motor current, motor voltage, and an estimate of motor thrust. The trained machine learning model may also input an error term of the closed-loop controller. Tuning constants for the closed-loop controller are derived based on the prediction from the machine learning model. Gain scheduling for the closed-loop controller may thus be performed “online” while the UAV continues on its mission. Controller gain scheduling may be performed to account for changes in a payload carried by the UAV.