Sensing and control for sheep shearing robots

Abstract

Sheep shearing is a difficult task for a robot and requires a robot with a delicate touch yet fast reactions, a large workspace, substantial load-carrying capabilities, and adaptive control. Tactile and proximity sensors supply the information needed to keep the cutter close to the skin. Knowledge of the shearing techniques being used at any time combined with force sensing and monitoring of unusual conditions in the adaptation mechanism of the robot provides the inputs to a real-time expert system which replans the shearing strategy when the lower level path and trajectory adaptation is not sufficient. A machine vision system is used to generate geometric models of the sheep's surface which are used to plan the arm trajectories and to provide feed-forward information into the cutter attitude control system. The sheep shearing robot, therefore, provides a working example of sensor-based control, trajectory adaptation, and online strategy planning. It is shown that, given a realistic understanding of the task, these techniques can be made to work in a real-time system, but considerable experimentation is required to have confidence that the resulting system will behave well under a wide variety of operating conditions.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>