Research on Object Detection and Tracking Methods for aLow-Speed Mobile Platform

Area of Science:

• Robotics
• Computer Vision
• Sensor Fusion

Background:

• Autonomous following systems face challenges in dynamic indoor environments due to occlusion and interference.
• Existing positioning systems struggle with accuracy and stability in complex settings.

Purpose of the Study:

• To enhance the positioning stability and accuracy of autonomous following systems.
• To develop a robust system for leader trajectory estimation and tracking in indoor environments.

Main Methods:

• Implemented an Ultra-Wideband (UWB) positioning network using an optimized Double-Sided Two-Way Ranging (DS-TWR) protocol.
• Integrated computer vision-based gait analysis (GaitPart framework) to address Non-Line-of-Sight (NLOS) UWB limitations.
• Utilized an Extended Kalman Filter (EKF) to fuse UWB and gait data for trajectory estimation, tracked by a Model Predictive Controller (MPC).

Main Results:

• The fused UWB and gait analysis system achieved precise and robust leader trajectory estimation.
• Tracking deviation remained within 50 mm for most trajectory points.
• Maximum deviation was 115 mm during turns, demonstrating strong robustness.

Conclusions:

• The proposed UWB and gait analysis fusion method significantly improves autonomous following system performance.
• The system exhibits strong robustness and practical utility for intelligent vehicle applications.
• This approach offers a viable solution for accurate positioning in challenging indoor environments.