Visual conveyor tracking for "pick-on-the-fly" robot motion control

Abstract

The paper presents a method and related algorithm for robot motion control in order to pick objects moving on conveyor bells; the instantaneous location of the moving objects is evaluated by a vision system consisting of a stationary, down looking matrix video camera, a controller-integrated frame grabber and image processor and the AVI version of V/V+ programming environment. The algorithm for visual tracking of the conveyor belt for "on-the-fly" object grasping is partitioned in two stages: (i) visual planning of the instantaneous destination of the robot; and (ii) dynamic replanning of the robot's destination while tracking the object moving on the conveyor belt. The control method uses in addition the concept of the conveyor belt window (CBW). The paper discusses in detail the motion control algorithm for visually tracking the conveyor belt on which objects are detected, recognised and located by the vision part of the system. Experiments are finally reported in what concerns the statistics of object locating errors and motion planning errors function of the size of the objects and of the belt speed. These tests have been carried out on a SCARA-type 4 d.o.f robot with artificial vision and the AVI AdeptVision software extension of the V/V+ highlevel, structured programming environment.