Wheel-AINS: A Vehicle Autonomous Positioning System Based on a Wheel-Mounted MIMU Array

Area of Science:

• Robotics and Autonomous Systems
• Navigation and Positioning
• Sensor Fusion

Background:

• High-precision vehicle autonomous positioning is challenging in satellite-denied environments (urban canyons, tunnels).
• Low-cost microelectromechanical systems inertial measurement units (MIMUs) are alternatives to expensive IMUs but suffer from noise and bias instability, leading to navigation errors.
• Existing solutions struggle with sustained autonomous navigation accuracy due to error divergence in single MIMU systems.

Purpose of the Study:

• To propose and validate a novel autonomous positioning system, the Wheel-AINS, utilizing a wheel-mounted MIMU array for enhanced vehicle navigation.
• To address the limitations of single MIMUs by employing differential fusion and kinematic constraints for improved accuracy and reduced cost.
• To provide a viable technical pathway for long-endurance vehicle navigation in satellite-denied environments.

Main Methods:

• Developed a Wheel-AINS employing a differential layout with multiple low-cost MIMU chips on rear wheels for redundant sensor arrays.
• Implemented differential fusion of symmetrically mounted chips to cancel common-mode noise and zero bias, using wheel rotation for speed estimation.
• Integrated estimated wheel speed and vehicle kinematic constraints within a Kalman filter framework to suppress inertial navigation system error divergence.

Main Results:

• The Wheel-AINS achieved an average position drift rate of 0.50% and an average heading Root Mean Square Error (RMSE) of 12.2° in urban road tests.
• Demonstrated an 80% reduction in closure error for a 2.49 km trajectory (10.43 m) compared to a single MIMU system.
• Ablation experiments showed MIMU array fusion reduced position RMSE from 155.0 m to 10.1 m; dual-wheel constraint further improved it to 8.2 m.

Conclusions:

• The proposed Wheel-AINS significantly enhances autonomous positioning accuracy in satellite-denied environments using low-cost MIMUs.
• Differential fusion and kinematic constraints effectively mitigate MIMU limitations, offering a cost-effective solution for vehicle navigation.
• The system presents a practical and robust approach for long-term autonomous navigation, overcoming challenges in GPS-unavailable conditions.