Cable-Driven Parallel Robot Pose Estimation Using Extended Kalman Filtering With Inertial Payload Measurements

This letter introduces two novel extended Kalman filtering (EKF) approaches to fuse payload accelerometer and rate gyroscope data with forward kinematics to estimate the payload pose of a cable-driven parallel robot (CDPR). An Euler-angle-based EKF and a rotation-vector-based multiplicative extended Kalman filter (MEKF) are proposed for this purpose. An unconstrained attitude parameterization identity is used to derive an analytic form of the Jacobian involved in the iterative forward kinematics calculations, which facilitates the use of different attitude parameterizations. Monte-Carlo simulations are performed with two levels of realistic sensor noise and bias, as well as calibration errors. The numerical results demonstrate more accurate pose estimates using the EKF and MEKF compared to forward kinematics computations alone.

Automated summary

This study introduces two novel extended Kalman filtering approaches to fuse payload sensor data and forward kinematics for pose estimation in cable-driven parallel robots. Monte-Carlo simulations demonstrate that using EKF and MEKF results in more accurate pose estimates compared to forward kinematics calculations alone.