Location-Aware Range-Error Correction for Improved UWB Localization

Area of Science:

• Robotics
• Sensor Fusion
• Wireless Communication

Background:

• Existing ultra wideband (UWB) ranging correction methods rely on channel information.
• These methods often lack robustness and can overfit.
• There is a need for improved ranging accuracy without additional hardware.

Purpose of the Study:

• To present a novel localization scheme, location-aware ranging correction (LARC), for UWB signals.
• To enhance ranging accuracy by incorporating preliminary location estimates into the correction model.
• To improve localization performance without requiring extra hardware like inertial measurement units (IMUs).

Main Methods:

• Developed the LARC scheme integrating location-based features into a range-error prediction model.
• Utilized an open-access UWB measurement dataset for distances up to 20 meters.
• Employed a simple regression model executable in real-time on a CPU.

Main Results:

• Reduced the 90th percentile (P90) ranging error by 58% to 15 cm for an unseen trajectory.
• Improved the 2D localization P90 error by 21% to 18 cm.
• Demonstrated robustness in a changed environment with moved metallic objects, achieving 56% better P90 ranging (16 cm) and a 17% improvement in 2D localization (18 cm).

Conclusions:

• The LARC method significantly enhances UWB ranging and localization accuracy.
• The approach is robust to environmental changes and does not require additional hardware.
• LARC offers a practical and efficient solution for real-time UWB positioning.