Interconnected maximum likelihood estimator and extended Kalman filter for inertial measurement unit calibration fusing three-dimensional camera information

Abstract

This article presents a method for calibration of inertial sensors (gyroscopes and accelerometers). In the proposed self-calibration method, interacted extended Kalman filter and maximum likelihood estimator are applied for estimation of motion state parameters and calibration parameters (biases and scaling factors), respectively. The extended Kalman filter is employed to estimate nonlinear Gaussian attitude kinematics fusing three-dimensional camera information given the estimated calibration parameters. These parameters are estimated in maximum likelihood–based system identification module and transferred to the state estimation module to estimate motion parameters. The calibration square error is then formulated and the effect of model mismatch is evaluated. The approach takes advantage of modular design, sensor fusion, and self-calibration. Self-calibration makes the method economically efficient because it does not rely on using a mechanical platform rotating the inertial measurement unit into different precisely controlled orientations. Experimental results based on data from a three-dimensional microelectromechanical system–based inertial measurement unit and a three-dimensional camera system are used to demonstrate the efficiency of the method. Comparisons are also provided to compare the method with state-of-the-art approaches.

Loosely coupled fusion of camera and inertial sensors for distributed error compensation in strapdown inertial navigation system

“…Readers are referred to Sadeghzadeh-Nokhodberiz and Poshtan (2015) and Törnqvist (2006) for more information about quaternions-based attitude representation.…”

“…The tightly coupled integration is optimal but time consuming due to high number of state parameters. In order to overcome this problem, decentralization of this integration scenario has been proposed in Sadeghzadeh-Nokhodberiz and Poshtan (2015) and Sadeghzadeh-Nokhodberiz et al (2014a, 2014b). Although the computational burden is less in loosely coupled integration, it suffers from fault propagation in the navigation algorithms.…”

Contactless multi-touch interactive technology based on adaptive filtering algorithm

“…Human-computer interaction is no longer confined to keyboard input and handling operation, but the emergence of a more novel way: tiny finger movement, vibration of sound in air, rolling of one's eyes and tongue. [3][4][5][6][7][8]10,[12][13][14][15][16][17][18][19][20] All of them can realize the transmission of information to complete the interaction between man and machine ''dialogue.'' However, the existing domestic and international human-computer interaction technology has a high requirement on the hardware, which is a large touchinteractive system with complex structure and low sensitivity.…”