Use of Trunk Roll Constraint to Improve Heading Estimation in Pedestrian Dead Reckoning Navigation Systems

T. Jakel, D. Gebre-Egziabher

Abstract:	In this presentation a method for increasing the accuracy of the heading estimate in IMU-based personal navigation system is discussed. Heading estimation without use or emittance of any external electromagnetic signals for reference is a critical but challenging component of pedestrian dead reckoning navigation systems. This is because in the environments where these systems are normally used, electromagnetic signals may be unavailable. Even when available, these signals may be intentionally or inadvertently modified for an extended period of time. In such environments, the navigation system must determine initial heading through gyrocompassing or from an alignment using information from other sensors such as magnetometers or radio frequency multilateration systems. Subsequently, these external measurements may be used periodically to arrest the heading drift when signal conditions and trajectory allow it. Due to the size, weight, power, and cost constraints imposed on a pedestrian navigation systems as well as current IMU performance limitations, the gyroscopes used to determine heading exhibit significant drift. To deal with this drift problem without having to rely on information external to the navigator, the use of human motion models as constraints has been proposed. For example, one such motion model used as a constraint predicts the onset of turning motion of the pedestrian by thresholding the outputs of the yaw gyroscope. That is, unless a significant yaw rate is detected, it is assumed that the user is moving in a straight line. Since this is also the gyroscope integrated to determine the change in angle over the period of turning, such constraints delay the open loop gyro integration time onset which increases the accumulated drift, especially for sudden turns which can occur during indoor navigation. In this presentation the trunk roll angle and angular rate are presented as additional signals used to predict the onset of pedestrian turning motion. The signature of the trunk roll motion relative to foot and body reorientation is described. Integration of the trunk roll motion based turning prediction method with existing yaw threshold methods is described and analyzed. Experimental data from ten subjects captured in a gait laboratory using a Vicon motion tracking unit is used for validation. In these experiments the subjects were instrumented with five low cost IMU units; one on the right foot, one on the right ankle, one on the right thigh, one on the lower back and one on the chest. The outputs of the foot-mounted IMUs are used with Zero Velocity Updates (ZUPTs) to determine the subjects gross body motion. The torso mounted IMUs are used to sense upper body motion and this information is used in activating the trunk-roll constraint. The analysis of experimental data shows that trunk rolling motion precedes heading change and can be used together with yaw rate threshold methods to predict the onset of turning motion. Thus, it provides a smaller latency in the turn detection. This, in turn, means a more accurate open-loop gyro integration time leading to lower accumulated gyroscope drift.
Published in:	Proceedings of the 26th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2013) September 16 - 20, 2013 Nashville Convention Center, Nashville, Tennessee Nashville, TN
Pages:	448 - 460
Cite this article:	Jakel, T., Gebre-Egziabher, D., "Use of Trunk Roll Constraint to Improve Heading Estimation in Pedestrian Dead Reckoning Navigation Systems," Proceedings of the 26th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2013), Nashville, TN, September 2013, pp. 448-460.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In