Pose-estimation methods for underactuated cable-driven parallel robots

Abstract

Cable-Driven Parallel Robots (CDPRs) displace the end-effector (EE) by means of cables configured in a parallel fashion. When a CDPR employs fewer cables than the degrees of freedom (DoFs) of its EE, it is underactuated and underconstrained. As a result, only a limited subset of the EE DoFs can be controlled assigning specific cable lengths, and determining the EE pose through direct kinematics (DK) is unfeasible, since it also depends on mechanical equilibrium. To estimate the EE pose, then, it may be convenient to acquire and combine redundant measurements. This paper introduces a sensor fusion technique that aims at achieving the optimal estimation of the pose of a generic UACDPR. Sensor fusion is achieved by an iterative nonlinear weighted least-square algorithm, which is solved by a Gauss–Newton-like method. A novel criterion for terminating iterations, which is physically sound and straightforward to implement, is put forward. Different redundant measurements are experimentally compared, so as to show the performance of the method with different sensors involved.