Robotic Processing of Variable Products

This paper summarises research and commercial developments in robotics for handling and processing tasks involving a high degree of variability. The variability is often derived from the characteristics of individual items to be processed e.g. natural product items such as meat portions or fish fillets which vary uniquely on an individual basis. Research into flexible knowledge-biled programming, sensor measurement, and sensor feedback to accommodate task variability is described. Example developed systems include scaled wall framing construction for domestic buildings using expert knowledge on building techniques, palletising of variable-size oil canisters, automated de-boning of fish fillets, and robotic cutting of lamb carcasses. The projects described were all carried out by members of the Advanced Manufacturing Group at Industrial Research in New Zealand.