Computing the worst case accuracy of a PKM over a workspace or a trajectory

Accuracy is clearly an important feature of PKM. Still it is difficult to evaluate what will be the influence of the sensor errors on the positioning errors except at one particular pose. We present a method that allow us, provided an analytical form of the inverse jacobian matrix of the robot, to estimate safely if the positioning errors may exceed some given thresholds over a given workspace or trajectory. This method is generic, i.e. it may be used for any PKM or any type of workspace or n-dimensional parametric trajectory (i.e. a surface, a volume).