Neural Network-Based Robust Guaranteed Cost Control for Image-Based Visual Servoing of Quadrotor

Area of Science:

• Robotics and Control Systems
• Artificial Intelligence in Engineering
• Aerospace Engineering

Background:

• Image-based visual servoing (IBVS) is crucial for autonomous quadrotor navigation.
• Existing control methods face challenges with time-varying dynamics, angle constraints, and uncertain disturbances in quadrotor IBVS.
• Robust control strategies are needed to guarantee performance and stability under these conditions.

Purpose of the Study:

• To design a neural network (NN)-based robust guaranteed cost control for quadrotor IBVS.
• To address the control problem of the time-varying lateral subsystem with angle constraints and uncertain disturbances.
• To ensure stability and effectiveness of the proposed control strategy.

Main Methods:

• A two-loop control structure was implemented, combining linear quadratic regulator (LQR) for the outer loop and optimal robust guaranteed cost control for the inner loop.
• The lateral velocity system's optimal robust control problem was transformed into solving a modified Hamilton-Jacobi-Bellman equation.
• Adaptive dynamic programming and time-varying neural networks with a designed estimated weight update law were utilized for implementation.

Main Results:

• The proposed NN-based robust guaranteed cost control effectively manages the quadrotor's lateral dynamics under constraints and uncertainties.
• The two-loop control structure successfully decoupled and addressed the yaw, height, and lateral subsystems.
• Simulations and theoretical proofs validated the stability and effectiveness of the developed control system.

Conclusions:

• The NN-based robust guaranteed cost control offers a viable solution for enhancing quadrotor IBVS performance.
• The adaptive dynamic programming approach provides an effective way to solve complex optimal control problems in real-time.
• The proposed method demonstrates significant potential for improving the robustness and reliability of autonomous quadrotor systems.