3D indoor area recognition for personnel security using integrated UWB and barometer approach

Area of Science:

• Robotics and Automation
• Sensor Fusion
• Machine Learning

Background:

• Real-time location tracking is critical for industrial worker safety, particularly in environments like power substations.
• Ultra Wideband (UWB) offers high accuracy and penetration for indoor positioning but faces challenges with complex topologies, multipath, and non-line-of-sight (NLOS) conditions.
• Traditional UWB trilateration demands extensive base station deployment and is susceptible to performance degradation in challenging indoor environments.

Purpose of the Study:

• To develop a robust 3D area recognition solution for industrial safety applications.
• To overcome the limitations of UWB positioning in complex indoor environments by integrating multiple sensor data.
• To enhance the reliability and efficiency of real-time location tracking for personnel security.

Main Methods:

• Integration of Ultra Wideband (UWB) time-of-flight (TOF) ranging with barometer measurements.
• Implementation of a multi-tier distributed joint probabilistic inference model.
• Utilization of machine learning algorithms, including clustering and prediction, for 3D area recognition.

Main Results:

• Achieved over 99.2% accuracy in 3D area recognition.
• Improved computing efficiency by 93% compared to traditional methods.
• Demonstrated less than 1m error in barometric height estimation, leading to 100% floor identification success with 3-4m floor separation.

Conclusions:

• The proposed integrated UWB and barometer system provides reliable real-time 3D area information, surpassing the need for mere coordinate-based positioning.
• This solution is highly suitable for personnel security in industrial scenes, offering enhanced accuracy and efficiency.
• The machine learning-based probabilistic inference model effectively addresses NLOS and multipath issues, improving overall system performance.