Fractional-order Model of the Dynamics of a Flexible Sensing Antenna for Haptic Underwater Applications

Abstract

This paper develops a mechanism to detect the instant at which an underwater haptic flexible antenna detects an object. This sensor is implemented as an aid to navigation in mobile underwater robots. It is used to localize obstacles by performing an active sensing strategy in which the servomotor system moves the beam back and forth until it hits an object. This impact instant is estimated by our mechanism which subsequently triggers an estimator of the contact point of the antenna with the obstacle and switches the rover control system from free to constrained navigation modes. Our detection mechanism is based on detecting if the difference between a measured force/torque signal and the value predicted by a dynamic model surpasses a threshold. This threshold and, then, the delay in the estimation of the impact instant, depend on the accuracy of this model. Several models have been fitted to experimentally recorded data showing that a fractional-order model of complexity equivalent to that of the standard integer model outperforms by more than five times the accuracy of this last one. This results in a much more efficient impact detection mechanism that has a much lower delay time.