Performance analysis of supply chain in olive pitting machines by artificial vision and neural networks

The pitting machines of olives are characterized because his optimum working is linked to a good adjust. selection of a right feed disc to the variety of olives and his caliber, the geometric features of supply chain, etc The first of this elements pins up the optimum input of olives in the supply chain keeping off buckets with gap or more than one olive there. The second element pins up the right position of the olives to be pitted, keeping off this olive be pitted for no main axe. The proposed paper analyses the right position of the olives in the buckets of the supply chain, for it we have used: 1.- An artificial vision system with external triggering be able to take photos for each buckets in front of the camera. 2.- A neuronal network classified, in a right teaching way, it can classify the bucket in four possible cases: empty, normal, incorrect olive position "barco", and anomalous case (two olives in the same bucket, broken olive and incorrect olive position however no barco olive). The analysis has carried out using tool-box of Matlab neuronal network, this previous analysis verified the viability of a neural network for this kind of classification. The paper values as a last input the use of physics chips with neuronal network to classify: a) Intel Curie b) NeuroMem CM1K It has been achieved to train a neuronal network implemented in physical chips to classify pictures of olives which are circulating on the feed lines of pitting machines of olives for the first time and in a successful way. The use of physical chips Intel Curie and mainly Neuromem CM1K due his greater capacity and scalability, it has been achieved so it is possible a great potential for classification.