A modified newton-euler method for dynamic computations in robot fault detection and control

We present a modified recursive Newton-Euler method for computing some dynamic expressions that arise in two problems of fault detection and control of serial robot manipulators, and which cannot be evaluated numerically using the standard method. The two motivating problems are: i) the computation of the residual vector that allows accurate detection of actuator faults or unexpected collisions using only robot proprioceptive measurements, and ii) the evaluation of a passivity-based trajectory tracking control law. The modified Newton-Euler algorithm generates factorization matrices of the Coriolis and centrifugal terms that satisfy the skew-symmetric property. The computational advantages with respect to numerical evaluation of symbolically obtained dynamic expressions is illustrated on a 7R DLR lightweight manipulator.