Food Orientor

Abstract

A method of automatically orienting symmetric and asymmetric food items, such as apples for example, is provided. Individual items of food are manipulated by a programmable manipulator within the view of one or more depth imaging cameras. Digital three dimensional characterizations of the surface of the food items are generated by the depth imaging camera or cameras and are utilized by a computer connected to the depth imaging camera or cameras to locate the stem and blossom of each food item. Asymmetric food items, such as apples with dropped shoulders as well as symmetric food items can be properly oriented and processed automatically.

Claims

1. A method of orienting food items for processing, comprising: imaging a food item, creating a digital three dimensional characterization of the surface of said food item, utilizing said digital three dimensional characterization of the surface of said food item to locate a stem and a blossom in three-dimensional space, and orienting said food item to a proper orientation by: rotating said food item about a first axis until said stem and said blossom lie in a pre-determined first plane; and rotating said food item about a second axis until said stem and said blossom lie along a pre-determined line.

2. The method of claim 1, wherein said food item is an apple having a stem indent in which said stem is located and a blossom indent in which said blossom is located and wherein a principal axis of curvature algorithm is utilized to locate said stem indent and said blossom indent.

3. The method of claim 1, wherein said digital three dimensional characterization is a three dimensional model of the surface of said food item, wherein a plurality of three dimensional models of properly oriented food items is stored in a digital library, and wherein an iterative closest point algorithm is utilized to compare said generated three dimensional model with said stored three dimensional models of properly oriented food items and to cause a programmable manipulator to correct the orientation of the food based on the comparison.

4. The method of claim 1, wherein a manipulator grips said food item using the fin ray effect.

5. The method of claim 1, wherein a manipulator utilizes an orienting cup with two off-center drive wheels beneath said orienting cup to cause said food item to rotate to present the entire surface of said food item to said depth imaging camera.

6. The method of claim 1, wherein said imaging is performed by an RGB-D camera.

7. The method of claim 1, wherein two or more depth imaging cameras are utilized to image said food item.

8. The method of claim 1, wherein four RGB-D cameras are positioned around said food item, and each camera generates a three dimensional model of the food item.

9. The method of claim 1, where a manipulator follows a preset motion pattern to present the entire surface of the food item to the camera.

10. The method of claim 1, wherein a position of a manipulator is tracked using encoders.

11. The method of claim 1, wherein said food item is an apple.

12. The method of claim 1, wherein one or more RGB-D cameras are utilized to detect the color of any bruises to said food items that require that food item to be either separated for special processing or to be discarded.

13. A method of automatically orienting food items, wherein each food item has a stem and a blossom comprising the steps: imaging a food item as said food item is manipulated to create a three dimensional characterization of said food item, comparing, using an iterative closest point algorithm, the created digital three dimensional characterization of the surface of said food with a digital library of a plurality of three dimensional models of properly oriented and symmetric food items to locate a closest match between the created digital three dimensional characterization of the surface of said food and one of three dimensional models in the digital library, and correcting the orientation of said food item to a proper orientation based on the located closest match.

14. The method of claim 13, wherein said correcting further comprises: actuating a manipulator to correct the orientation of said food item to said proper orientation.

15. The method of claim 13, wherein said correcting uses a manipulator to grip said food item using a fin ray effect.

16. The method of claim 13, wherein said correcting uses a manipulator that comprises an orienting cup with two off-center drive wheels beneath said orienting cup to cause said food item to rotate to present the entire surface of said food item to said depth imaging camera.

17. A system of automatically orienting food items, wherein each food item has a stem and a blossom comprising: an imager configured to image each food item, a data processor configured to create a digital three dimensional characterization of the surface of a food item being manipulated and utilize said digital three dimensional characterization of the surface of said food item being manipulated to locate said stem and said blossom in three-dimensional space, and wherein said data processor is configured to correct the orientation of said food item to a proper orientation by: rotating said food item about a first pre-determined axis until said stem and said blossom lie in a first pre-determined plane; and rotating said food item about a second pre-determined axis until said stem and said blossom lie along a pre-determined line.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0023] FIG. 1 is a schematic cross-sectional illustration of a symmetrical apple;

[0024] FIG. 2 is a schematic cross-sectional illustration of an unsymmetrical apple having a “dropped shoulder;”

[0025] FIG. 3 is a schematic cross-sectional illustration of an asymmetrical apple having a “dropped shoulder” as properly oriented by the present invention;

[0026] FIG. 4A illustrates an asymmetric apple having a “dropped shoulder” as presented to a manipulator;

[0027] FIG. 4B illustrates the apple of FIG. 4A as properly oriented in accordance with the invention;

[0028] FIG. 5 is a more detailed, perspective view of the manipulator shown in FIGS. 4A and 4B with a modification in the swivel joint supports;

[0029] FIG. 6 is a block diagram of the control system of a first embodiment of the invention;

[0030] FIGS. 7A and 7B are perspective views of an alternate manipulator;

[0031] FIG. 8 is a perspective view of a further alternate manipulator;

[0032] FIG. 9 is a schematic representation of a second embodiment of the invention; and

[0033] FIG. 10 illustrates the preferred use of 4 depth imaging cameras.

DETAILED DESCRIPTION OF THE DRAWINGS

[0034] FIG. 1 is a schematic cross-sectional representation of a symmetrical apple 10 having a stem 14 at upper end 11 with stem indent 13 and a blossom end 12 with blossom indent 15 and blossom 16. The upper indent 13 contains stem 14. The blossom indent 15 contains blossom 16. Axis 18 extends vertically from stem 14 to blossom 16 and is the correct axis along which coring tubes move to remove the undesirable seed cell 19. Apple 10 is symmetrical relative to axis 18.

[0035] FIG. 2 is a schematic cross-sectional representation of asymmetric apple 30, since apple 30 is not symmetric relative to axis 38, which extends through stem 34 and blossom 36. The stem indent 33 is positioned in “dropped shoulder” 31. When apple 30 enters a prior art orientor, the relatively flat blossom end 32 with indent 35 is positioned as shown in FIG. 2. The prior art orientors locate indent 35 and align apple 30 as shown in FIG. 2, with vertical axis 37 not extending through stem 34 and seed cell 39. A coring machine removes a cylindrical core along axis 37, but misses the seed cell 39 entirely. This problem is overcome by the present invention.

[0036] FIG. 3 is an illustration of the asymmetric apple 30 of FIG. 2 properly oriented by the invention and positioned in a vertical coring device 40.

[0037] FIGS. 4A-4B are schematic diagrams of the concept of the invention. Current feed and singulation equipment (not shown) is used to automatically transport individual produce items such as apple 105 into programmable manipulator 100 as shown in FIG. 4A. The manipulator 100 will grab hold of the apple 105 regardless of orientation and present it to a depth imaging camera, preferably an RGB-D camera 200. The RGB-D camera 200 images each produce item such as apple 105 repeatedly as the produce item is manipulated and collects color information (Red, Green, and Blue) for every point on the surface of apple 105 as well as distance of each point from the camera. The manipulator will automatically follow a preset and programmable pattern to manipulate apple 105 to present the entire surface of the apple 105 to the camera 200. The manipulator's position is tracked using known encoders (not shown).

[0038] As shown in FIG. 4A, apple 105 has a dropped shoulder 107 and is asymmetric about an axis between its stem 106 and blossom 108. The apple core 139 is shown between stem 106 and blossom 108. Axis 138, extending from stem 106 to blossom 108, is the proper axis to be oriented vertically for transfer to coring and other processing equipment, as noted above. Axis 137 represents an axis incorrectly located by typical prior art orientors described above. Stem indent 133 includes stem 106. Blossom 108 is located in blossom indent 135. The bottom 132 of apple 105 is relatively flat.

[0039] Manipulator 100 has a generally Y-shaped base 90 which includes a support sleeve 91 which rotates about axis x-x as support shaft 110 is rotated. Support shaft is actuated by pneumatic drive means not shown. Sleeve 91 is formed integrally with and carries shoulders 92 and 93, which in turn pivotally carry arms 101 and 102, at pins 103 and 104. Support shaft 110 also is connected to, carries and actuates a four bar linkage system including pivotable arms 101 and 102 and linkage arms 94 and 95. Linkage arms 94 and 95 are pivotally connected to arms 101 and 102 by pins 96 and 97 and carried by linkage support base 96, which in turn is carried by the upper end 110a of shaft 110 (shown in phantom). As shaft 110 is advanced upwardly in FIG. 4A by pneumatic or other means, the four base linkage system causes arms 101 and 102 to separate, spreading cups 121 and 122 apart. As shaft 110 is retracted downwardly, preferably by pneumatic action, the four bar linkage causes arms 101 and 102 to move closer together to cause support cups 121, 122 to contact the apple 105 without crushing it. As shaft 110 rotates around axis x-x, the entire manipulator 100 and apple 105 are rotated in unison. The rotation of shaft 110 about axis x-x and the rotation of shafts 111 and 112 about axis y-y enable manipulator 100 to present the entire surface of apple 105 to camera 200. Swivel joints 131 and 132 are carried by shafts 111 and 112. Swivel joints carry support cups 121 and 122. Shaft supports 141 and 142 are carried at the ends of arms 101 and 102, and support shafts 121 and 122. One small servo motor 143 is carried by shaft support 142. Servo motor 143 drives both shafts 111 and 112, since shafts 111 and 112 are only rotated together when an apple is carried by manipulator 100.

[0040] Once the stem 106 and blossom 108 have been located as described above, the manipulator 100 is actuated to orient the stem 106 vertically with respect to blossom 108.

[0041] FIG. 4B illustrates proper orientation of stem 106 vertically aligned with blossom 108. It is to be understood that the apple 105 may be properly oriented with the stem 106 either vertically above blossom 108 as shown in FIG. 4B, or stem 106 may be vertically below blossom 108.

[0042] To orient apple 105 from the position shown in FIG. 4A to its proper orientation shown in FIG. 4B, manipulator 100 is rotated about the x-x axis until stem 106 and blossom 108 lie in a plane including the x axis and the z axis (shown best in FIG. 5), wherein the z axis is perpendicular to the plane including the x and y axis. Once the stem 106 and blossom 108 lie in the plane including the x and z axes, cups 121 and 122 are rotated around the y-y axis to reach the proper orientation shown in FIG. 4B. Once the proper orientation is achieved, cups 121 and 122 are separated by support shaft 110 moving upwardly in FIG. 4B by a pneumatic drive (not shown) to release the oriented apple to known transfer apparatus for further processing.

[0043] It is to be understood that in FIGS. 4A and 4B, camera 200 and manipulator 100 are connected to a computer not shown for clarity. The drive mechanism for manipulator 100 is also not shown for clarity.

[0044] In a first embodiment of the invention, images from camera 200 are used to create a 3D model of the surface of apple 105. This 3D model is a “Digital three dimensional characterization” of the surface of the produce item, apple 105, being manipulated. An Iterative Closest Point (ICP) algorithm is used to compare the generated model with a previously created pre-existing digital library of a plurality of 3D models of properly oriented symmetric and asymmetrical apples (or other produce items) to locate the closest match between the current generated model and the digitally stored models. The motions needed to correct the orientation of the apple from its current state to a preferred state are calculated. Those motions are then carried out by the manipulator 100 to produce a properly oriented apple as shown in FIG. 4B. The Iterative Closest Point (ICP) algorithm is known in the art and readily available.

[0045] FIG. 5 is a perspective and more detailed view of preferred manipulator 100 shown in FIGS. 4A and 4B, in which shafts 111, 112 shown in FIGS. 4A and 4B have been shortened, so that swivel joints 131 and 132 are carried by shaft supports 141 and 142. As noted above one small servo motor 143 (FIG. 4A) is carried by shaft support 142 and is not visible in FIG. 5. As noted above, manipulator 100 has arms 101 and 102 carried by shaft 110 which rotates about axis x-x. Support cups 121 and 122 are carried by swivel joints 131, 132 carried by shafts 111 and 112. Support cups 121, 122 are utilized to contact and manipulate apple 105 to achieve correct orientation.

[0046] FIG. 6 is a block diagram of the control system for said first embodiment is shown generally as 50. A known feed system 51 is actuated to transport an apple to the manipulator 100. The manipulator 100 is actuated to rotate the apple in a preprogrammed sequence stored at 52. One or more depth imaging cameras 200 is actuated to scan the apple repeatedly as it is rotated by manipulator 100.

[0047] The three dimensional model 250 of the current apple being scanned is fed into computer 300. An Iterative Closest Point (ICP) algorithm 260 is used to compare the model 250 with a digital library 320 of three dimensional digital models of properly oriented symmetrical and asymmetrical apples to determine the closest match and the proper pose for the current apple. The manipulator motions to properly orient the current apple are calculated at 330 and fed to the manipulator at 340. The manipulator is actuated at 350 to properly orient the current apple, and the apple is then transferred to coring or peeling at 360.

[0048] FIGS. 7A and 7B show an alternative manipulator 400 that grips the apple using the fin ray effect. Fins 410 and 420 are carried by arms 401 and 402, which in turn are carried by a support shaft 405 which rotates about axis x-x, in similar fashion to manipulator 100 shown in FIGS. 4A, 4B and 5. The fins 410 and 420 grip the produce as shown in FIG. 7B, as known in the manipulator art.

[0049] FIG. 8 shows a further manipulator 500 based on an orienting cup 520 with 2 wheels 530, 540 off center rather than one rotating wheel in the center. Orienting cup 520 has an opening (not visible in FIG. 8) formed in its bottom, as known in the art. Apple 505 is rotated by off-center wheels 530 and 540 so that the entire surface of apple 505 can be repeatedly imaged by one or more depth imaging cameras (not shown in FIG. 8). When the entire surface of apple 505 has been imaged, either or both wheels 530, 540 are actuated to properly orient apple 505.

[0050] FIG. 9 is a schematic diagram representing a second embodiment of the invention shown generally as 700. Four depth imaging cameras 711-714 are spaced around manipulator 740 and are repeatedly imaging an asymmetric apple 720 having a stem 734 in stem indent 733 and a blossom 736 in blossom indent 735 as apple 720 is being manipulated. As the apple 720 is manipulated by manipulator 740 (such as manipulator 100 shown and described above the indents) 733 and 736 are presented to cameras 711-714 along with the entire surface of apple 720. As the apple continues to be rotated, a three dimensional image of the surface of apple 720 is fed by cameras 711-714 to computer 760. Computer 760 is connected to the drive 770 for manipulator 740. Computer 760 utilizes the known Principal Axis of Curvature (PAC) algorithm to create a three dimensional map of the slope for each point on the surface of apple 720. The stem indent 733 and blossom indent 735 have the highest clusters of high slopes, which clusters are utilized to locate the indents. When the indents are located, the location of stem 734 and blossom 736 are known and the manipulator 740 is actuated by computer 760 to properly orient the apple 720 for processing as described above. Normal orientation for apples aligns stem 734 vertically with blossom 736, with either the stem being above or below the blossom. In using the PAC algorithm, no digital library is required. Although 4 depth imaging cameras are shown in FIG. 10, acceptable results can be achieved with a single camera. A block diagram of the control system for this embodiment is the same as the diagram of FIG. 6, except that the 3D image at 250 is of the slopes for each point on the surface of the apple, the algorithm at 310 is the PAC rather than ICP, and the digital library shown at 320 is not used with the PAC algorithm.

[0051] The Principal Axis of Curvature (PAC) algorithm is known and is not described in detail here. The basic two steps of the PAC algorithm are:

[0052] 1. Represent the segmented apple point cloud in terms of curvature by multiplying surface normal by curvature magnitude. The result is a point cloud that represents the magnitude of surface changes vs. direction. This translates the higher curvature around the stem and blossom indents into a point cloud that is elongated along the stem-blossom axis.

[0053] 2. Estimate the principal axis of the curvature representation of the apple using a robust version of PCA. The principal axis is the axis of maximum variability. The principal axis, or first principal component, represents the estimated axis of the apple. Perform a few iterations of principal axis estimation with outlier removal.

[0054] As shown in the diagram of FIG. 10, the most preferred embodiment of the invention utilizes four RGB-D cameras 810, 820, 830 and 840 placed around manipulator 100, shown in FIGS. 4A, 4B and 5. Computer 860 is connected to manipulator 100 and all four cameras 810, 820, 830 and 840.

[0055] One or more depth imaging cameras may be utilized. The RGB-D cameras are preferred, since they also provide color information. The color information is utilized to detect dark or discolored regions on the surface of bruised apples which are not appropriate for automatic coring or peeling. Such bruised apples are separated and either discarded or processed by alternate means.

[0056] As used herein and in the claims, the phrase “digital three dimensional characterization of the surface” refers to any useful digital depiction, model or representation of the shape of the surface or of any characteristic of the surface such as slope.

[0057] As used herein and in the claims, the phrase “depth imaging camera” refers to any camera capable of generating three dimensional images or characterizations of the surface of an object within the view of said camera.

[0058] It is to be understood that locating the stem and blossom is done in most instances in the case of apples by locating the stem indent and/or blossom indent using the PCA algorithm and assuming that the stem and blossom are located at the center of each respective indent. Accordingly, as used herein and in the claims, the phrase “locating the stem and blossom” is used broadly to include locating the stem indent and blossom indent. In addition, for many varieties of peaches, pears and apricots where the stem and/or blossom indents may be too small to use the PCA algorithm, the first embodiment using a digital library and the ICP algorithm would be the appropriate method.

[0059] In the case of peaches and apricots, the proper orientation is required for removing the stone or pit. The present invention may be utilized to locate not only the stems and blossoms of peaches and apricots, but also the “suture line” of these items. Proper orientation of the suture line is significant in removing the stone or pit, as is known in the art.

[0060] The foregoing description of the invention has been presented for purposes of illustration and description and is not intended to be exhaustive or to limit the invention to the precise form disclosed. Modifications and variations are possible in light of the above teaching. The embodiments were chosen and described to best explain the principles of the invention and its practical application to thereby enable others skilled in the art to best use the invention in various embodiments suited to the particular use contemplated.