Kalman Filter Based Pose Estimation for CDPRs (Cable Driven Parallel Robots)

Abstract

This work extends previous work by taking the novel a novel extended Kalman filter (EKF) approach for the pose estimation of a cable driven parallel robot (CDPR) and comparing it with a novel Unscented Kalman Filter (UKF) based approach. Both filters fuse accelerometer, rate gyroscope, and winch encoder data together to create a more accurate covariance pose estimate. Both filters are tested on experimental data collected by a six degree of freedom CDPR test bed. The current results show minimal difference between the two filters but offer promising possibilities for future work.