|Narayanan, Priya - UMBC, GRADUATE STUDENT|
|Tasch, Uri - UMBC, PROFESSOR|
|Rostamian, Rouben -|
Submitted to: Transactions of the ASABE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: October 29, 2008
Publication Date: December 1, 2008
Citation: Narayanan, P., Lefcourt, A.M., Tasch, U., Rostamian, R., Kim, M.S. 2008. Orientation of apples using their inertial properties. Transactions of the ASABE. 51(6):2073-2081. Interpretive Summary: The use of automated systems to inspect fruit during processing offers the potential to increase the safety and quality of food consumed by the public. Such systems use cameras to take pictures of the fruit and computer software to determine if the fruit is contaminated, or if it is damaged or diseased. For apples, attempts to use automated inspection have been hindered by the difficulty of differentiating the stem and calyx regions of the apple from problem areas. One way of improving the effectiveness of automated inspection is to orient fruit prior to taking pictures so that the location of the stem and calyx is known. We discovered that rolling apples down a track consisting of two parallel rails can orient apples so that the location of the stem or calyx is always in the same location. Tests were conducted by rolling 100 Red and 100 Golden Delicious apples down two different test tracks with three different initial positions of the stem/calyx axis: perpendicular, at 45°, and parallel to the rails. Apples became oriented before traveling 50 cm for all cases except the parallel loading condition and theoretical dynamics suggest that parallel loading will not occur if a random loading mechanism is used. These results suggest that using tracks to convey and orient apples has the potential to significantly improve the ability to inspect apples for defects and contamination and to reduce commercial processing costs, and will be of interest to scientists, regulators, producers, and equipment manufacturers.
Technical Abstract: Machine vision has the potential to improve quality and safety of fruit available for consumption. However, lack of an effective and inexpensive method for appropriately orienting fruit for imaging has hindered development of optical technologies for sorting fruit. For example, it is difficult to differentiate stem and calyx regions from problem areas in images. One solution would be to orient fruit so that the location of these regions is known. We discovered that apples can be oriented using their inertial properties. Apples were rolled down two test tracks consisting of two parallel wooden rails using 100 Red and 100 Golden Delicious apples with three initial orientations of the stem/calyx axis: perpendicular, parallel, or at 45° to the rails. As the apples progressed down the tracks, they generally moved to an "oriented" position where the stem/calyx axis was perpendicular to the direction of travel and parallel to the plane of the rails. A camera was used to create movies of the orientation process and software was written to automatically analyze the movies. Apples started in a perpendicular position were de facto oriented. Apples started at 45° almost instantly became oriented. For the parallel condition, orientation was delayed and was successful in only about half of trials. However, theoretical dynamics suggest that this parallel loading result is an artifact and will not occur in a commercial setting where apples are loaded onto the track using an automated mechanism. Alternatively, potential methods for circumventing this problem case are discussed.