A Comparison of Visual Servoing from Features Velocity and Acceleration Interaction Models

Visual Servoing has been widely investigated inthe last decades as it provides a powerful strategy for robotcontrol. Thanks to the direct feed-back from a set of sensors,it allows to reduce the impact of some modeling errors and toperform tasks even in uncertain environments. The commonlyexploited approach in this field is to use a model that expressesthe rate of change of a set of features as a function of sensortwist. These schemes are commonly used to obtain a velocitycommand, which needs to be tracked by a low-level controller.Another approach that can be exploited consists in goingone step further and to consider an acceleration model forthe features. This strategy allows also to obtain a naturaland direct link with the dynamic model of the controlledsystem. This study aims at comparing the use of velocity andacceleration-based models in feed-back linearization for VisualServoing. We consider the case of a redundant manipulator anddiscuss what this implies for both control techniques. By meansof simulations, we show that controllers based on featuresacceleration give better results than those based on velocityin presence of noisy feedback signals.