# Multi-Objective Grey Wolf Optimizer-Tuned LQR Attitude Control of a Three-DOF Hover System

## TL;DR

This paper proposes a systematic method using a multi-objective grey wolf optimizer to tune LQR controller parameters for attitude control of a three-DOF hover system.

## Contribution

A novel framework combining multi-objective optimization and TOPSIS for LQR weight selection in hover system attitude control is introduced.

## Key findings

- The multi-objective grey wolf optimizer successfully tunes LQR parameters for improved attitude control performance.
- Simulation results demonstrate feasible tracking performance using the optimized LQR parameters.
- The framework provides a replicable method for LQR weight selection in hover-type systems.

## Abstract

Attitude control of unmanned aerial vehicles is a problem that needs to be solved in a reliable manner. The research presented in this paper examines a systematic approach to the design of an LQR state feedback controller for the three-DOF hover system. The state space model is used to derive the feedback gain K, with the diagonal elements of the weighting matrices Q and R used as design variables. A multi-objective grey wolf optimizer is used to obtain Q–R matrices based on closed-loop simulations under representative roll, pitch and yaw reference commands. There are four separate multi-objective optimization runs, each using one of four standard error indices which are the integral of absolute error (IAE), the integral of time-weighted absolute error (ITAE), the integral of squared error (ISE) and the integral of time-weighted squared error (ITSE). Each index is used to track roll, pitch and yaw errors at the same time and the resulting non-dominated solution sets are post-processed using TOPSIS to select a compromise knee-point design. The simulation results show that the adjusted LQR parameters lead to feasible tracking performance. The proposed framework provides a systematic and replicable method for LQR weight selection in hover-type attitude control problems under the considered simulation settings.