Gyroscope-Aided Odometry Navigation Using a Highly-Precise Automotive MEMS IMU Complemented by a Low-Cost Sensor Array

This paper examines vehicle navigation employing a redundant array of two different types of MEMS inertial sensors in combination with wheel speed sensors. We compare the position precision of a purely inertial strap-down algorithm with motion-constraints (SMC, [1]) to gyroscope-aided odometry navigation (GAO). An RTK-corrected GNSS was used in parallel to generate the reference trajectory. Initial heading of both methods was determined by a dual-antenna GNSS moving baseline setup. Across five repeated experiments with a traveled distance of 1600m over 275 s each, GAO delivered a mean RMSE of 4.36 m, whereas SMC resulted in errors of 634 m. For GAO, we show that initial heading error was dominant compared to the influences of gyroscope noise. Determining accurate initial heading is therefore deemed crucial for applications with prolonged GNSS outages.