Global Positioning System (GPS): A Low-Cost Velocity Sensor for Correcting Inertial Sensor Errors on Ground Vehicles

Abstract

This paper demonstrates the ability of a standard low-cost Global Positioning System (GPS) receiver to reduce errors inherent in low-cost accelerometers and rate gyroscopes used on ground vehicles. Specifically GPS velocity is used to obtain vehicle course, velocity, and road grade, as well as to correct inertial sensors errors, providing accurate longitudinal and lateral acceleration, and pitch, roll, and yaw angular velocities. Additionally, it is shown that transient changes in sideslip (or lateral velocity), roll, and pitch angles can be measured. The method utilizes GPS velocity measurements to determine the inertial sensor errors using a kinematic Kalman Filter estimator. Simple models of the inertial sensors, which take into account the sensor noise and bias drift properties, are developed and used to design the estimator. Based on the characteristics of low-cost GPS receivers and IMU sensors, this paper presents the achievable performance of the combined system using the covariance analysis from the Kalman filter. Subsequent simulations and experiments validate both the error analysis and the methodology for utilizing GPS as a velocity sensor for correcting low-cost inertial sensor errors and providing critical vehicle state measurements.

Filling the gap between low frequency measurements with their estimates

“…In order to demonstrate the applicability of the proposed approach to a multi-rate sensor fusion problem, we select a pair of sensory data and relate them under a Kalman filter with the kinematic model introduced in [10]. This is not an unique choice, as long as we were able to solve one measurement from another, we are able to apply the proposed approach.…”

“…We adopt the kinematic model introduced in [10] to model the underlying kinematic relation between the high frequency measurement v and its bias as:…”

“…In Fig. 1 the position estimate (red line) is obtained by applying the Kalman filter as described in [10]. This Kalman filter runs state-update at the sampling frequency of v (1000 Hz) and receive measurements at the sampling frequency of u (32Hz).…”

Study on Tracked Combine Harvester Dynamic Model for Automated Navigation Purposes

“…The sideslip angle is calculated from the difference between the direction of the tracked combine harvester given by the RTK-GPS and the heading given by the IMU [9].…”

Estimation of road inclination and bank angle in automotive vehicles

“…These include [4], which is based on transfer functions from the steering angle and road bank angle to the yaw rate and lateral acceleration, and [5], [6], which use an extended Kalman filter (EKF) to estimate the sideslip angle, which is in turn used in a linear unknown-input observer to estimate the road bank angle. Several GPS-based designs, such as [7], [8], estimate the road inclination or bank angle. In this paper, we do not assume availability of GPS measurements.…”