Authors:
Jielei Zhang
;
Jie Feng
and
Bingfeng Zhou
Affiliation:
Peking University, China
Keyword(s):
Inertial Measurement Unit, Sensor Fusion, Inertial Navigation, Trajectory Reconstruction.
Related
Ontology
Subjects/Areas/Topics:
Computer Vision, Visualization and Computer Graphics
;
Hardware Technologies for Augmented, Mixed and Virtual Environments
;
Interactive Environments
;
Mobile Interfaces
Abstract:
In this paper, we present a novel sensor-fusion method that reconstructs trajectory of mobile devices from
MEMS inertial measurement unit (IMU). In trajectory reconstruction, the position estimation suffers seriously
from the errors in the raw MEMS data, e.g. accelerometer signal, especially after its second-order integration
over time. To eliminate the influence of the errors, a new error model is proposed for MEMS devices. The
error model consists of two components, i.e. noise and bias, corresponding to different types of errors. For
the noise component, a low-pass filter with down sampling is applied to reduce the inherent noise in the data.
For the bias component, an algorithm is designed to detect the events of movement in a manner of sensor
fusion. Then, the denoised data is further calibrated, according to different types of events to remove the bias.
We apply our trajectory reconstruction method on a quadrotor drone with low-cost MEMS IMU devices, and
experiments show the eff
ectiveness of the method.
(More)