A Sensor Fusion Approach to Observe Quadrotor Velocity.

Area of Science:

• Robotics
• Control Systems
• Computer Vision

Background:

• The increasing deployment of Unmanned Aerial Vehicles (UAVs) necessitates advancements in their autonomous navigation, particularly for GPS-denied environments like indoor spaces.
• Visual odometry offers a promising alternative to traditional positioning systems (e.g., GPS) for enabling robust UAV navigation.
• Quadrotor platforms require sophisticated methods for accurate state estimation to achieve reliable autonomous flight.

Purpose of the Study:

• To develop and validate a translational velocity observer for quadrotor UAVs.
• To enhance autonomous navigation capabilities by fusing inertial and visual sensor measurements.
• To provide a GPS-independent navigation solution suitable for indoor environments.

Main Methods:

• A translational velocity observer was designed using the Immersion and Invariance observer design technique.
• The observer fuses data from inertial measurement units (IMUs) and visual sensors.
• Lyapunov stability theory was employed to formally prove the convergence of the observer's error to zero.

Main Results:

• The proposed observer successfully estimates the translational velocity of a quadrotor by integrating inertial and visual data.
• Numerical simulations demonstrated the observer's effectiveness in achieving accurate velocity estimation.
• The Immersion and Invariance technique provided a systematic approach to observer synthesis.

Conclusions:

• The developed observer enhances the autonomous navigation of UAVs, particularly in GPS-limited conditions.
• Fusion of inertial and visual measurements offers a robust solution for quadrotor velocity estimation.
• The Lyapunov-based stability analysis guarantees the observer's performance and reliability.