A Nonlinear Input Shaping Technique for Motion Control of a Sensing Antenna

Abstract

Flexible links combined with force and torque sensors can be used to detect obstacles in mobile robotics, as well as for surface and object recognition. These devices, called sensing antennae, perform an active sensing strategy in which a servomotor system moves the link back and forth until it hits and object. At this instant, information of the motor angles combined with force and torque measurements allow calculating the positions of the hitting points, which are valuable information about the object surface. In order to move the antenna fast and accurately, this article proposes a new open loop control for driving this flexible link based sensor. The control strategy is based on an (IS) Input Shaping technique, in order to reduce link vibrations. The antenna performs free azimuthal and vertical movements. However, the vertical movement is clearly non-linear due to the gravity effect, which prevents the use of standard linear IS techniques. Then a new nonlinear IS has been developed in this article which includes a linearization term of the gravity. Experiments have shown the improvements attained with this technique in the accurate and vibration free motion of our antenna.