Enhanced Indoor Mobile Robot Localization via Lie-Group IMU-UWB Fusion and Dual-Stage Kalman Filtering.

Area of Science:

• Robotics
• Sensor Fusion
• State Estimation

Background:

• Single-modality localization in indoor mobile robots leads to poor accuracy and robustness.
• Conventional pose computation methods are complex and prone to errors.
• Existing sensor fusion techniques may not effectively filter measurement noise.

Purpose of the Study:

• To propose a novel indoor robot localization method, IMU_UWB_ESKF, for improved accuracy and robustness.
• To leverage Lie-group theory for numerically stable pose propagation and measurement updates.
• To implement a dual-stage Kalman filtering strategy for enhanced sensor fusion and noise reduction.

Main Methods:

• Tightly fuses inertial measurement unit (IMU) and ultra-wideband (UWB) measurements using a Lie-group state representation.
• Formulates IMU and UWB data on the Lie algebra for stable pose propagation and updates.
• Employs a dual-stage Kalman filtering approach: EKF for measurement correction and an error-state Kalman filter for refined fusion.

Main Results:

• The proposed IMU_UWB_ESKF method demonstrates stable real-time localization.
• Achieved improved robustness and accuracy, especially during dynamic motion.
• Successfully implemented on a wheeled robot platform within the ROS framework.

Conclusions:

• The IMU_UWB_ESKF method offers a robust and accurate solution for indoor mobile robot localization.
• The Lie-group state representation and dual-stage Kalman filtering effectively mitigate sensor noise and drift.
• The approach is well-suited for real-time indoor navigation tasks requiring high localization performance.