No abstract

No abstract

The work discusses the performance of a Stochastic Cloning Unscented Kalman filter (SC-UKF) which is used to fuse the incremental position and orientation information from the Visual Odometry (VO) using a stereo camera setup and the absolute attitude obtained from a low-cost inertial measurement unit. The system is designed for pedestrian tracking within an uncontrolled environment and employs a quaternion-based attitude representation within the filter state. The attitude is cloned and kept between lower rate VO samples, while inertial data is processed in real-time with a higher sampling rate. Corresponding to the same time span, the relative orientation from the VO is used to correct the IMU-based rotation difference between the cloned and the current attitude. The information of magnetic compass is included in order to improve the heading estimation along with the mechanism for magnetic disturbance compensation. The filter scheme is extended by implementing the INS mechanization equations for position estimation, where the VO data is used as a velocity observation to reduce the growth of the rate of the position error. The performance of the designed SC-UKF is compared to the one of SC-based Extended Kalman filter on a number of representative walking paths. The augmented system shows a better performance especially for the indoor segments such as corridors with insufficient illumination and stairs with monotone walls.

No abstract

The work presents an extension of the conventional Kalman filtering concept for systems of fractional order (FOS). Modifications are introduced using the Grünwald-Letnikov (GL) definition of the fractional derivative (FD) and corresponding truncation of the history length. Two versions of the fractional Kalman filter (FKF) are shown, where the FD is calculated directly or by augmenting the state vector with the estimate of the FD. The filters are compared to conventional integer order (IO) Position (P-KF) and Position-Velocity (PV-KF) Kalman filters as well as to an adaptive Interacting Multiple-Model Kalman Filter (IMM-KF).The performance of the filters is assessed based on a hand and a head motion data set. The feasibility of the given approach is shown.