Error Compensation Method for Pedestrian Navigation System Based on Low-Cost Inertial Sensor Array

Area of Science:

• Navigation Systems
• Sensor Fusion
• Inertial Measurement Units

Background:

• Pedestrian navigation systems often use inertial measurement unit (IMU) arrays to reduce errors.
• Current fusion methods neglect individual IMU error compensation and zero-velocity interval corrections, limiting accuracy.
• Existing techniques fail to effectively minimize errors and enhance navigation precision.

Purpose of the Study:

• To propose an enhanced error compensation method for pedestrian navigation systems utilizing low-cost IMU arrays.
• To improve the calibration of multiple IMUs and correct position information during zero-velocity intervals.
• To significantly reduce measurement errors and boost navigation performance.

Main Methods:

• An improved calibration method using a sliding variance detector to segment data into stationary and motion intervals for independent IMU calibration.
• Application of compensation to velocity residuals within the zero-velocity interval after zero-velocity update (ZUPT).
• Fusion of measurements from multiple low-cost IMU arrays.

Main Results:

• A 3.01-fold increase in static noise performance for the calibrated IMU array.
• A 27.52% reduction in average horizontal position error for a single calibrated IMU compared to an uncalibrated one.
• A 2.98-fold reduction in average horizontal position error for the calibrated IMU array compared to a single calibrated IMU.
• Further reduction in average horizontal position error by 0.73 m for a single IMU and 64.52% for the IMU array after residual velocity compensation.

Conclusions:

• The proposed error compensation method significantly improves the noise performance and accuracy of IMU arrays in pedestrian navigation.
• Independent IMU calibration and ZUPT corrections are crucial for enhancing navigation system performance.
• The method offers a substantial improvement over existing techniques, particularly in reducing horizontal position errors.