Abstract
This paper presents a position estimation method using information from lateral line-array (LA) sensors for active singulation of overlapping objects in real time. We derive constraints imposed on the placement of LA sensors capable of compensating for object slippage on conveyor while constructing its 2D boundary in dark rooms. The models (based on ellipsoids that geometrically represent a wide variety of natural objects) characterize the clustering of objects by quantifying the degree of overlap. As illustrative applications, we present experimental results obtained with live broilers illustrating the effectiveness of this method in detecting transitions between overlapping objects (with unknown sizes) for high-speed singulation.