Learning-Based Control Compensation for Multi-Axis Gimbal Systems Using Inverse and Forward Dynamics

TL;DR

This paper introduces a neural network-based control method for multi-axis gimbal systems that models both forward and inverse dynamics to effectively compensate for disturbances and uncertainties, improving stability and performance.

Contribution

It presents a novel approach combining forward and inverse dynamics modeling with neural network disturbance compensation for enhanced gimbal control.

Findings

Neural network controller outperforms traditional controllers in simulations and experiments.

The method effectively handles complex disturbances and parameter uncertainties.

No tuning required for the neural network controller across operational ranges.

Abstract

Unmanned aerospace vehicles usually carry sensors (i.e., electro-optical and/or infrared imaging cameras) as their primary payload. These sensors are used for image processing, target tracking, surveillance, mapping, and providing high-resolution imagery for environmental surveys. It is crucial to obtain a steady image in all these applications. This is typically accomplished by using multi-axis gimbal systems. This paper concentrates on the modeling and control of a multi-axis gimbal system. A novel and fully outlined procedure is proposed to derive the nonlinear and highly coupled Equations of Motion of the two-axis gimbal system. Different from the existing literature, Forward Dynamics of the two-axis gimbal system is modeled using multi-body dynamics modeling techniques. In addition to the Forward Dynamics model, the Inverse Dynamics model is developed to estimate the complementary torques associated with the state and mechanism-dependent, complex disturbances acting on the system. A disturbance compensator based on multilayer perceptron (MLP) structure is implemented to cope with external and internal disturbances and parameter uncertainties through the torque input channel. Our initial simulations and experimental work show that the new NN (neural network)-based controller is performs better in the full operational range without requiring any tuning or adjustment when compared with well-known controllers such as cascaded PID, ADRC (Active Disturbance Rejection Control), Inverse Dynamics based controllers.