An Autonomous Land Vehicle Navigation System Based on a Wheel-Mounted IMU

Area of Science:

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

Background:

• Low-cost inertial systems suffer from significant drift errors, challenging land vehicle navigation in Global Navigation Satellite System (GNSS)-denied environments.
• Traditional methods like odometer/non-holonomic constraints/inertial navigation systems (INS) have limitations in accuracy and error suppression.

Purpose of the Study:

• To propose an autonomous navigation strategy using a wheel-mounted microelectromechanical system (MEMS) inertial measurement unit (IMU), termed wheeled INS, to mitigate navigation errors.
• To conduct a theoretical analysis of wheeled INS error characteristics and system observability.
• To develop a hybrid extended particle filter (EPF) for accurate state estimation.

Main Methods:

• Utilizing inertial mechanization algorithms to predict vehicle position, velocity, and attitude (PVA).
• Employing gyro outputs to derive forward velocity, treated as an observation with non-holonomic constraints (NHCs) for error state estimation.
• Performing system observability analysis to understand error characteristics.
• Implementing a hybrid extended particle filter (EPF) combining extended Kalman filter (EKF) and particle filter (PF) for state updates.

Main Results:

• Wheel rotation enhances observability of gyro errors, effectively suppressing azimuth, horizontal velocity, and position errors.
• The wheeled INS demonstrates superior navigation performance compared to traditional odometer/NHC/INS.
• Kinematic field tests over 26 km showed a maximum position drift rate of 0.47% and a root mean square (RMS) heading error of 1.13°.
• The EPF effectively handles system non-linearity and non-Gaussian noises, achieving high accuracy with tolerable computational complexity.

Conclusions:

• The proposed wheeled INS provides an accurate navigation solution for land vehicles in GNSS-denied environments.
• The integration of wheel rotation and advanced filtering techniques significantly improves navigation system performance and error suppression.
• This strategy offers a viable solution for autonomous navigation where GNSS is unreliable or unavailable.