Abstract
This paper presents a method to determine the 2-D profile and motion of a live product (such as chicken for poultry meat processing) on amoving conveyor from a lateral optical sensor that consists of an orthogonal pair of line array (LA) scanners. Unlike most line array (LA) scanners designed to provide a 2-D image of a static object, the lateral optical sensor presented here offers a practical means to detect object slippage on the conveyor in real time. Three examples are given to illustrate the effectiveness of this sensing method. The first simulates the 2-D boundary of a geometrically well-defined object on an accelerating conveyor, which offers intuitive insights on the effects of conveyor dynamics and object slippage on the accuracy of the 2-D boundary measurement. The second experimentally demonstrates the extendibility of LA sensors to detect both engineering and natural objects. The final example illustrates the application of the lateral optical sensor as a real time feedback sensor for active singulation of natural objects.
Note to Practitioners-This paper was motivated by the problem of automated handling of live chickens on moving conveyors for poultry meat processing. The method presented here can also apply to other object handling applications where the absolute location of the randomly shaped moving object is used as a triggering signal for subsequent operation. Existing approaches generally rely on the use of the conveyor speed with a beam switch (or a line scan array) and are unable to account for object slippage or voluntary motion. This paper suggests a new approach using a pair of orthogonal optical line sensors along with the known conveyor speed to determine the absolute location and lateral profile of the object on the moving conveyor; both engineering and natural objects are considered. In this paper, we formulate a locating method that exploits the fast scan rate of photoelectric sensors to detect discrepancies between the position/velocity of the object and the conveyor. We then illustrate how this information can be used to compensate for object slippage due to conveyor acceleration and perform object-based location calibration. Preliminary physical experiments conducted on a small number of live chickens using (300 mm-long) line sensors (consisting of 128 emitter/receiver pairs spaced at 2.5 mm) sampled at a rate of 40-50 ms suggest that this method has a significant potential in applications such as active singulation where object injuries due to unpredictable voluntary motion must be minimized.