Skip to main content
To KTH's start page To KTH's start page

Inertial Sensor Arrays

Sensor Fusion and Calibration

Time: Thu 2022-04-21 13.00

Location: F3, Lindstedtsvägen 26 & 28, Stockholm

Language: English

Subject area: Electrical Engineering

Doctoral student: Håkan Carlsson , Teknisk informationsvetenskap

Opponent: Professor Thomas Seel, Friedrich-Alexander Universität, Germany

Supervisor: Professor Joakim Jaldén, Teknisk informationsvetenskap

Export to calendar

QC 20220331


Motion estimation using inertial sensors is today used in a wide range of applications, from aircraft navigation to inflatable bicycle helmets. The accuracy with which the motion can be estimated using inertial sensors depends on how large the measurement errors are. One approach to reducing the inertial sensors' measurement errors is to use more sensors than what is necessary for motion estimation. By averaging the measurements from a redundant amount of sensors, the impact of independent errors can be reduced. But by placing multiple inertial sensors on a rigid body, more information about the motion is available than what can be obtained from simple averaging. For instance, point-wise accelerations of a rigid body contain information on the rotation of the rigid body. This thesis examines and proposes methods for how to fuse the measurements from an inertial sensor array and how systematic measurement errors present in the sensors can be estimated and calibrated.

The inertial sensor array contains multiple accelerometers and multiple gyroscopes. In motion estimation applications, it is common to estimate the angular velocity from the gyroscopes measurements and then integrate the angular velocity into an orientation. The angular velocity can also be estimated from multiple accelerometers. This thesis proposes different models for fusing the accelerometer and gyroscope measurements for more accurate orientation estimation. By increasing the accuracy with which the orientation can be estimated, the integrated error in the position and velocity estimates can be decreased.

The performance of the fusion algorithms for multiple inertial sensors depends on how large the systematic measurement errors are. The amount of rotational information from multiple accelerometers depends on how well the locations of the accelerometers are known. Other calibration parameters in the inertial sensor array are sensor biases. These calibration parameters are estimated in conventional calibration by exposing the inertial sensors to a known reference motion. However, creating such reference motion requires external equipment that may not be available to the user. Therefore, this thesis proposes methods to jointly estimate the motion and the sensor parameters, thereby omitting the need for external calibration equipment.