Homogeneous Sensor Fusion Optimization for Low-Cost Inertial Sensors

Area of Science:

• Sensor Fusion
• Inertial Sensor Arrays
• Real-time Calibration

Background:

• Homogeneous inertial sensor arrays require robust calibration for accurate data.
• Degraded sensors can significantly compromise the precision of fused sensor data.
• Existing methods may lack real-time adaptability and automatic fault detection.

Purpose of the Study:

• To develop a real-time sensor fusion and calibration method for inertial sensor arrays.
• To enable automatic suppression of degraded sensors within the array.
• To maintain or improve the overall precision of the sensor estimation.

Main Methods:

• Adaptive weighting of sensor data based on Root Mean Square Error (RMSE) against a weighted average.
• Real-time estimation of sensor calibration constants, including gain and bias.
• Comparison of estimated angular velocity with a tactical-grade fiber-optic gyroscope as ground truth.

Main Results:

• The proposed method successfully estimates calibration constants in real-time.
• Degraded sensors were automatically suppressed, preserving estimation precision.
• Experimental validation with low-cost MEMS gyroscopes demonstrated the method's effectiveness.

Conclusions:

• The developed technique offers robust real-time calibration and sensor fusion for inertial sensor arrays.
• Adaptive weighting and degraded sensor suppression enhance data reliability.
• The method is applicable to various sensor arrays, including MEMS gyroscopes.