Observer-based controller for VTOL-UAVs tracking using direct Vision-Aided Inertial Navigation measurements.

Area of Science:

• Robotics
• Control Systems
• Navigation

Background:

• Vertical Take-Off and Landing (VTOL) Unmanned Aerial Vehicles (UAVs) require precise navigation, especially in Global Positioning System (GPS)-denied environments.
• Existing navigation systems often rely on GPS, limiting UAV operation in challenging conditions.
• Vision-Aided Inertial Navigation Systems (VA-INS) offer a promising alternative for robust state estimation.

Purpose of the Study:

• To propose a novel observer-based controller for VTOL-UAVs that directly utilizes VA-INS measurements.
• To enable computationally efficient and accurate estimation of attitude, position, and velocity without explicit reconstruction.
• To ensure stable control of the UAV's six degrees of freedom (6 DoF) trajectory.

Main Methods:

• Development of an observer-based controller integrating data from a vision unit and a 6-axis Inertial Measurement Unit (IMU).
• Direct use of IMU and feature measurements for estimating attitude, gyroscope bias, position, and linear velocity.
• Formulation of the problem on a Lie Group using unit-quaternions for global and unique 6 DoF representation.

Main Results:

• The observer accurately estimates key motion components (attitude, position, velocity) and gyroscope bias.
• The observer-based controller effectively manages VTOL-UAV attitude, angular velocity, position, and linear velocity.
• Exponential stability of closed-loop estimation and control errors is mathematically proven from almost any initial condition.

Conclusions:

• The proposed observer-based controller provides a computationally efficient and robust solution for VTOL-UAV navigation in GPS-inaccessible areas.
• The Lie Group and unit-quaternion formulation ensures a globally unique representation and stable control.
• Both continuous and discrete versions of the approach are validated.