Inertial Line-Of-Sight Stabilization Using a 3-DOF Spherical Parallel Manipulator with Coaxial Input Shafts
Alexandre Le, Guillaume Rance, Fabrice Rouillier, Damien Chablat
TL;DR
The paper investigates inertial line-of-sight (LOS) stabilization using a coaxial-input 3-DOF spherical parallel manipulator (CoSPM) with three 3-RRR branches. It delivers a detailed kinematic analysis, deriving the geometric model, first-order kinematics, and a polynomial formulation to study singularities, and proves geometric invariance with respect to the coaxial bearing axis $\chi_3$, ensuring the prescribed workspace avoids both Type-1 and Type-2 singularities. A speed-control law is proposed, yielding decoupled, identical loops that map LOS-frame angular rates to actuator commands via $\mathbf{J}$ and $\mathbf{T}$, with a carefully tuned $K_0(s)$ achieving robust disturbance rejection and satisfactory stability margins ($60^{\circ}$ phase, $14.2$ dB gain). The approach is validated through MATLAB/Simulink simulations and a SimMechanics-based digital model, showing sub-milliradian residuals and modest actuator torques (transient $\leq 4$ Nm, steady $\leq 2.5$ Nm) under sea-state disturbances. The work highlights practical viability of CoSPMs for inertial LOS stabilization while acknowledging limitations in control robustness and suggesting future work on joint-stops management and a full dynamic model for certification under uncertainty.
Abstract
This article dives into the use of a 3-RRR Spherical Parallel Manipulator (SPM) for the purpose of inertial Line Of Sight (LOS) stabilization. Such a parallel robot provides three Degrees of Freedom (DOF) in orientation and is studied from the kinematic point of view. In particular, one guarantees that the singular loci (with the resulting numerical instabilities and inappropriate behavior of the mechanism) are far away from the prescribed workspace. Once the kinematics of the device is certified, a control strategy needs to be implemented in order to stabilize the LOS through the upper platform of the mechanism. Such a work is done with MATLAB Simulink using a SimMechanics model of our robot.
