An UWB/Vision Fusion Scheme for Determining Pedestrians' Indoor Location

Area of Science:

• Robotics and Autonomous Systems
• Computer Vision
• Sensor Fusion

Background:

• Accurate pedestrian indoor localization is crucial for various applications.
• Existing methods like ultra-wideband (UWB) and vision-based Simultaneous Localization and Mapping (SLAM) have limitations.
• UWB offers robust ranging but can suffer from multipath interference, while vision-based SLAM faces challenges with scale ambiguity and textureless environments.

Purpose of the Study:

• To develop a robust indoor pedestrian localization system by fusing UWB and vision data.
• To address the limitations of individual localization techniques, specifically scale ambiguity in monocular SLAM and performance in challenging visual conditions.
• To achieve high-precision and reliable positioning for pedestrians in diverse indoor settings.

Main Methods:

• Proposed an ultra-wideband (UWB) localization algorithm utilizing an extended Kalman filter (EKF) for precise ranging.
• Developed a method to resolve scale ambiguity and enable real-time repositioning for monocular ORB-SLAM (Oriented FAST and Rotated BRIEF-Simultaneous Localization and Mapping) using EKF.
• Implemented a sensor fusion approach combining UWB and vision data to leverage the strengths of both technologies.

Main Results:

• The UWB localization algorithm achieved an indoor positioning accuracy of 0.3 meters.
• The integrated system demonstrated reliable positioning accuracy on the order of 0.2 meters in experimental tests.
• The fusion approach successfully solved scale ambiguity for monocular ORB-SLAM and enabled repositioning upon vision track failure, enhancing overall system robustness.

Conclusions:

• The proposed UWB and vision fusion algorithm provides a reliable and accurate solution for pedestrian indoor localization.
• The system effectively handles environments with sparse textures and frequent changes in illumination.
• The fusion strategy improves the performance of individual UWB and vision localization methods, making it suitable for practical indoor navigation.