Novel pairwise coupled kinematic solution for algebraic angular acceleration estimation of serial link manipulators

We consider low-noise angular acceleration estimation for multi-axis robotic manipulators. The proposed model uses pairwise coupled inertial measurements across a section of the kinematic chain, which is reduced to a single rigid body. Experimental validation is built upon compact low-power micro-electro-mechanical (MEMS) components, installed on a full-scale heavy-duty mobile manipulator. While the model itself has a built-in mechanism for common-mode disturbance rejection, an adaptive transversal filter is devised for a further improvement. The results indicate a 40-80 fold suppression of high-frequency perturbations with respect to a baseline motion derivative, the discrete difference. As inertial sensors require no mechanical contact to rotating axes and the number of parameters is kept low, the model is easily applicable to motion control feedback of typical heavy-duty manipulators.