Non-destructive assessment of mango firmness and ripeness using a robotic gripper
Mangoes are climateric fruits, and their ripening process occurs rapidly during post-harvest time. The determination of fruit ripening stage is a basic issue during the packaging process in order to provide a supply of good quality fruit. Firmness, in particular, is considered a reliable indicator of mango maturity, and can be measured manually or with a penetrometer. However, the first method provides poor repeatability and the second is destructive. A recent study by a group of Spanish researchers (Blanes et al., 2015), evaluated the use of a robotic gripper for the automatic and non-destructive assessment of mango firmness. The tests were performed on 350 mangoes of the Osteen cultivar, divided in 7 sets (A, B, C, D, E, F e G) of 50 mangoes each, at different ripening stages. The gripper has three fingers, each of them containing an accelerometer that is connected to a data acquisition module. During the tests, the fruits were analysed to establish the major physical and chemical properties related with their ripening index: mechanical firmness, total soluble solids, pH, and acidity, as well as the flesh luminosity. Then, Partial Least Square regression models (PLS) were developed to explain these properties according to the variables extracted from the accelerometer signals. The authors proved that robotic grippers can handle 100% of the samples of the sets from A to F without damaging them. Only in the case of riper fruits (G), the gripper damaged 10% of the samples. With the PLS analysis, models with good correlation coefficients were obtained for the prediction of following parameters: mechanical firmness (r = 0.925), soluble solids (r = 0.892) and flesh luminosity (r = 0.893). In short, the study showed that with a robot gripper the non-destructive automatic assessment of mango firmness and ripeness is possible prior to its packaging.
Muscle separation by means of a multi-arm robotic system
The level of interest in robotic systems continues to increase in the meat industry based on their capacity of reducing meat processing costs. In this context, a recent study conducted by a team of French researchers (Nabil et al., 2015), presented the development of a robotic arm for beef muscles separation. This operation consists in cutting the aponeurosis, i.e. the tissue separating two muscles, over a length of about 50 cm using small incisions along the surface. In particular, process control of this multi-arms robot is based on both physical modeling of soft material and vision perception. The first allows estimating material behaviour, while the active perception system provides feedback on the material current surface shape during handling. The control system was designed using magnetic resonance imaging (MRI) techniques. Thanks to this technology, two different models could be developed, and the simulation results are presented in this study. In order to take into account material anisotropy, one of the MRI models was modified by introducing several non-linear parameters that contribute to enhance the realism of simulations. The cutting task was simulated using different knife positions, pull-off strengths and experimental cutting forces. Furthermore, the study introduced a new algorithm based on vision perception and curvature estimation of 3D surfaces. This algorithm enables the cutting tool path generation and updating starting from a starting point (Po) set by the robot operator. Finally, the authors point out the fact that the developed algorithm can be easily modified and implemented on robotic cutting systems for other types of meat.
Blanes et al., Food and Bioprocess Technology, 8, 2015, 1914-1924
Nabil et al., Robotics and Computer-Integrated Manufacturing, 32, 2015, 37-53