Session
Session 1
Board Number
1
Pose Estimation of a Cable-Driven Parallel Robot Using Kalman Filtering and Forward Kinematics Error Covariance Bounds
This paper introduces a novel Unscented Kalman filter (UKF) approach for the pose estimation of a cable driven parallel robot (CDPR). The filter fuses accelerometer, rate gyroscope, and winch encoder data and is compared to a dynamically-updated covariance on the forward kinematics pose estimation error. The filter is tested on experimental data collected by a six degree-of-freedom CDPR test bed as well as simulated truth data. The results show that the UKF is capable of providing similar pose estimation accuracy compared to forward kinematics alone.