Real-Time Multi-Target Localization from Unmanned Aerial Vehicles

Area of Science:

• Aerospace Engineering
• Computer Vision
• Geomatics Engineering

Background:

• Unmanned aerial vehicle (UAV) electro-optical imaging systems require improved reconnaissance efficiency.
• Accurate real-time multi-target localization is crucial for UAV operations.

Purpose of the Study:

• To propose a real-time multi-target localization scheme for UAV electro-optical stabilized imaging systems.
• To enhance the accuracy and efficiency of UAV-based target geo-location.

Main Methods:

• Developed a target location model and calculated geodetic coordinates using homogeneous coordinate transformation.
• Implemented real-time zoom lens distortion correction and a recursive least squares (RLS) filtering method based on UAV dead reckoning.
• Established a multi-target localization error model using Monte Carlo theory.

Main Results:

• Lens distortion correction reduced circular error probability (CEP) by 7%, enabling simultaneous localization of 50 targets.
• The RLS algorithm reduced CEP by 25% compared to single-image localization by adaptively estimating location data from multiple images.
• Localization results were within allowable error ranges during actual flight tests at 1140 m altitude.

Conclusions:

• The proposed real-time multi-target localization scheme is effective and can be implemented on small circuit boards.
• The system significantly benefits small UAVs requiring multi-target geo-location capabilities.
• This research improves the accuracy and efficiency of UAV reconnaissance and surveillance.