FR2775204A1 - Sorting of fruit and vegetables based on at least one of their characteristics - Google Patents

Abstract

The sorting machine delivers items one-by-one between an entry station (PE) and a packaging station (PP). External or internal properties are automatically determined while the fruit or vegetable is held at one of a number of measuring stations (PM1, PM2, PM3) placed between the entry and packaging stations. The measurements include size, color, density, weight, firmness and sugar levels.

Description

Method and installation for sorting fruits and vegetables, allowing the measurement of at least one of their characteristics.

Field of the invention
The present invention relates to a method and a fruit sorting facility, for measuring at least one of their external or internal characteristics such as, for example: size, firmness, density, ripeness, sugar content, coloration, appearance, etc.

 More specifically, the machine and the method according to the invention are intended for the automatic sorting of approximately spherical shaped fruits such as apples, peaches, oranges, mandarins, tomatoes, etc., presenting, for example, individually, a weight of between 100 g and 400 g.

State of the art
Currently, fruit and vegetable sorting and packing machines, often called "sizing machines", operate in continuous streams of fruit. The fruits are dumped from a container that is either a "pallox" (box of large dimensions and capacity: about 1,20 x 1,00 x 0,80 m, and 350 to 400 kg of fruit) transportable by means of a handling machine, ie a current box (capacity of 15 to 20 kg of fruit). These containers filled with fruit to sort come directly from the orchard or field, or cold rooms where they were stored after a step of pre-sorting said fruits.

 The spilled fruits are put on line, on one or more parallel lines materialized by ginning carriers and are individualized in buckets or in diabolos. They scroll rapidly, continuously, at typical rates of the order of 3 to 7 fruits / second.

 According to the techniques using electronics, fruits are sorted, scrolling, according to at least two criteria: the size estimated by weighing, and the color determined by optical sensors (photocells or cameras). The most sophisticated systems use CCD cameras to estimate the size by the diameter and to sort the appearance defects, by the implementation of an artificial vision system. These different criteria are designated as determining the external quality of the fruits.

 Depending on these categories, each fruit is ejected from the main sorting line and then routed to a specific output conveyor. There are typically 5 to 40 outputs per line, depending on the sorting facilities.

 The final crating is most often manual, and sometimes automatic.

This architecture is optimized for the sorting rate, but it has limitations, for example: - no measurement of internal quality - the sensors measure the fruit in
scrolling, which excludes sensors operating in direct contact with the fruit
in particular, these machines do not include sensors allowing
to assess the internal quality, in particular by measuring the firmness or
sugar levels.

- no total visualization: the fruits roll in the field of vision, there is no
therefore no guarantee to visualize all of their surface.

- lack of modularity: the addition of new sensors to measure
new sorting criteria is difficult, as well as an increase in
number of outputs.

- fruit damage: fruits suffer shocks at different levels
between them before going online, with the elements of the line at each
load break, this is especially true when ejecting the line
main.

- important dimensions. scroll rate constraints and
avoidance of shocks result in line lengths of several
dozens of meters.

- by automatic orientation no automatic means of orientation
Geometric fruit does not exist; The final aspect of crating can not
so be optimized only manually. In the sorting facilities
Manual, the presentation of fruits in the trays is often optimized in
function of the most ripe face or peduncle.

 In summary, current fruit sorting processes and facilities only allow the measurement of a small number of characterization elements of the latter and the measurements they allow are incomplete and / or insufficiently precise.

Objects of the invention
A first object of the invention is to overcome the aforementioned drawbacks by proposing a method and an installation for extending the number of measurements of characterization elements that can be taken into account for performing a selective sorting of fruits, depending many criteria, while allowing measurements in good conditions by which the results obtained are more complete and / or precise.

 According to a first characteristic arrangement of the method and the installation according to the invention, applicable to installations or machines of the kind in which the fruits or vegetables are animated by a scrolling movement, one by one, between an entry station and a plateauing station, a measure of at least one of the characteristics of the fruits included among the following characteristics. size, color, appearance, density, weight, firmness, sugar content, is carried out during a step of stopping said fruit at a measuring position arranged between the input station and the setting station of the plateau sorting facility.

 The machine according to the invention comprises an input station and a sorting station for sorted fruits or vegetables, between which at least one measuring station is disposed of at least one of their aforementioned characteristics, said machine comprising means ensuring the scrolling, one by one, of the latter, this machine being remarkable in that it is equipped with means for stopping momentarily the scrolling of each fruit or vegetable following his arrival at the measuring station or each measuring station for the duration of the measurement.

 Thanks to the invention, it is possible to develop an installation and a method making it possible to measure, if not all, at least almost all the fruit quality characterization criteria, and to carry out these measurements in conditions for high efficiency and improved accuracy.

According to an example of a very interesting application, the method and the installation according to the invention can be used for the classification of fruits according to their sugar level or sugar content;
For this purpose, according to another important feature of the invention, an optical interface providing illumination and / or light collection, such as a bundle of optical fibers Y or an integrating sphere, is brought into contact with each fruit, said optical interface being connected to a spectrometer itself connected to a computer.

 With this process and this facility, it is possible to make individual measurements of the internal quality of the fruits, including measurements of their sugar level.

 According to another example of a very interesting application, the method and the installation according to the invention can be used to measure the weight and the firmness of the fruits.

 For this, according to another important feature of the invention, the fruits are brought to a drop point from which they fall, one by one, of a low height, on an impactor comprising in particular a weighing platform resting on a strain gauge generating signals for determining, on the one hand, the firmness of each fruit, by means of a specific algorithm, and, on the other hand, the weight of said fruit.

 According to another example of a very interesting application, the method and the installation according to the invention can be used to carry out individual measurements of one or more external qualities of the fruits, in particular measures of their size, of their color, their diameter, or their appearance (presence or absence of spots or scab, presence or absence of peduncle, etc.).

 For this, according to another important characteristic feature of the invention, each fruit is presented to an inspection station in the field of vision of an artificial vision system, by means of two grippers each having a degree of freedom tilt and a degree of freedom of rotation in the axis of their gripping head, one of these grippers to capture each fruit and to orient, by a rotational movement, most of the surface of the fruit, in the direction of the artificial vision system, said fruit being then grasped by the second gripper which also directs, by a rotational movement, the remaining previously hidden surface, towards said artificial vision system.

 With this arrangement, it is possible to inspect the entire surface of the fruit.

 By taking into account the information collected by the weighing system and the inspection system, information on fruit density can also be obtained.

 It will be understood that the installation according to the invention may comprise one or more of the above-mentioned sets or measuring stations.

 In its most complete configuration, the fruit or vegetable sorting machine according to the invention allows an individual measurement of all the external and internal qualities of the latter, as well as a final orientation of the controlled products.

In such a configuration, the machine according to the invention aims to - automatically receive each fruit at a single point of entry; - inspect the fruit by presenting it, at a standstill and according to any
orientation, to different quality sensors; these sensors can
in particular to measure the size, the color, the diameter, the defects of appearance, the
weight, firmness, or sugar level; - classify the fruit according to various combinations of the results of the
different measures - orient the fruit for its final output at a fixed point - possibly, automatically condition the fruit respecting
the orientation given.

Since the treatment of each fruit is individual, its corollary is a relatively limited rate, which is typically of the order of 1000 to 2000 fruits per hour;
According to an example of very interesting application, the machine can be placed on one of the outputs of a conventional calibrator continuous stream. It then performs a high quality sorting of fruits whose caliber and / or color have good commercial potential. The machine selects, among these fruits, those with the best qualities of taste and maturity, and ensures their conditioning.

 According to another example of very interesting application, the machine can be used for the approval of a batch by sampling. It is then fed with a representative fraction of the incoming batch, and it allows the complete measurement of the characteristics of the sample (the packaging is not in this case an important advantage, the fruits can be evacuated on a single output conveyor) . Since the measurements carried out by the method and the machine according to the invention are non-destructive of the integrity of the fruits, there is no disadvantage in applying them to a relatively large fraction of the incoming batch: from 0.5% to 5%.

This results in a fast, accurate and objective method of characterization, which makes it possible to evaluate the value of the batch and to decide on its acceptance.

According to another example of a very interesting application, a subassembly or station of the machine can be used for the classification of fruits according to their sugar level only. Before the measurement, the fruit is stabilized at a standstill in a device for measuring the sugar level that indexes its position (sucker, diabolos games, etc ...). The measurement is always performed by an actuator automatically bringing, in contact with the fruit, the optical interface or interfaces, to carry out the measurement. On the other hand, the orientation capacity of the fruit can be omitted in a simplified version. The fruit can be released - either manually, the operator conditioning the fruit according to the class
announced by said device, - or automatically by output on specific conveyors
According to another example of a very interesting application, another subassembly or station of the machine can be used for the classification of the fruits according to their visual appearance only: the caliber, the color and the defects of appearance are then determined. The possibility of arbitrary orientation of the fruit allows an optimal classification according to these criteria, avoiding the hidden parts. Again, the input and output can be configured in both manual and automatic modes.

Brief description of the drawings
The above and other objects, features and advantages will become more apparent from the following description and accompanying drawings, in which:
Figure 1 is a top view of the implementation of a sorting machine shown in its most complete version for measuring all the fruit characterization criteria necessary for a high quality sorting.

 Figure 2 is a partial view on a larger scale of Figure 1.

 Figure 3 is a synopsis of the operation of robotic fruit sorting stations of the machine.

 Figure 4 is a front view, schematic, of the tray and outlet station.

 Figure 5 is a partial view on a larger scale of Figure 4.

 Figure 6 is a front view illustrating the transfer kinematics of the fruit in the inspection station, then in the tray setting.

 Figures 7A and 7B are side views showing two different positions of the same gripper.

 FIGS. 8A to 8F are views illustrating the 100% visualization of a sphere in four views according to the geometric procedure used in the method and the sorting machine of the invention.

 Figures 9 and 10 are views, respectively from the side and from above, of the impactor ensuring the measurement of firmness and weight.

 Figure 11 is a side view illustrating the approach of an optical fiber bundle constituting the optical interface providing illumination and / or light collection.

Description of an Example of Implementation of the Invention
Reference is made to the drawings to describe an interesting example, although in no way limiting, of implementing the method and of producing the machine according to the invention.

 An example of an installation or machine configured to allow the measurement of a maximum of criteria for characterizing fruit or vegetables is described below, it being observed, however, that certain stations or subassemblies of this installation could be used separately to measure one or more characterization elements only, depending on the applications or wishes of the users.

 The various stations or subassemblies of this installation are described below in an order determined by the routing of the fruits, from an entry station to an exit station.

According to this interesting example, the machine mainly comprises (FIG. 1): a PE entry station for automatically bringing the fruits one by
one and on demand; a PM robotic measurement station with four degrees of freedom, comprising
all the sensors, a PP set-up station ensuring the automated packaging of the
fruits after examination and grading; - two PS output stations full trays; two PC stations for loading the empty trays and the cells of
receipt of fruits.

Compared to this diagram and as indicated above, several variants are possible, because to limit costs, one can, for example independently: - replace the PE input station by a manual feed; - replace the PP tray and the PS output stations with a
manual conditioning, removing PC stations.

 On the other hand, the robotic measurement station PM may itself advantageously consist of several sets, namely (FIG. 2): a PM1 station for measuring the firmness and weight of the fruits; a PM2 station for inspecting their external appearance; a PM3 post for measuring the internal quality (sugar content) of said fruits.

The entry station
The PE input station (FIG. 1) can itself have two possible types of power supply: - an automatic crate bouncer that creates a flat layer, in the case
manual feeding by crate; a simple conveyor 1 coming from one of the sorting outlets of a known sizer
in itself.

 The box bouncer known per se is a rocker that returns the box, step by step, as the fruits are consumed downstream of the feed system. A flexible tarpaulin forms a lid and prevents the fruit from falling on the conveyor.

 Then, the fruits are brought on a pre-alignment conveyor or ginner 2 placed at 90 "of the main conveyor 1. It consists, in a manner known per se, of two mats forming a V and moving at different speeds, in order to stretch the fruits one behind the other Two lateral stops 3 fitted to the fixed longitudinal walls of the conveyor-ginner 2, improve the stretching of the fruits, these abutments 3 being offset along the axis of circulation, to avoid any flow blocking effect.

 Finally, the fruits go down slightly to a conveyor diabolos 4, known in itself, whose speed is greater than that of the conveyor belt preceding for better stretching. At this stage, the risk of having two fruits in front are eliminated, but it remains possible to have an empty slot on a diabolo.

 The three elements mentioned above are each controlled by an industrial programmable logic controller (PLC). The three PLCs are synchronized, and at the same time they start and stop the above mentioned three components of the input station.

 According to a characteristic arrangement of the method and of the installation of the invention, the three components of the input station are automatically controlled by a control system located downstream from said input station, this control system being advantageously constituted by computer at least one of the measuring stations. For example, the entrance station is set in motion, on demand, by the computer of the inspection station, as and when it needs new fruits, the power of the measuring station or the all the cells or measuring stations thus operating in a sequential and asynchronous method, each fruit or vegetable not entering said measuring station or the set of measuring stations or cells, only when the latter is ready for use. treat. This avoids any risk of jamming the measuring station. It has been validated that the online fruiting function is performed correctly in batch operation, whereas the known systems of this type are validated only in continuous operation.

Set of measures
The PM measurement set of fruit characteristics is schematically illustrated in Figure 3.

 This measuring assembly is itself constituted, in its most complete version, of a cell or station for measuring weight and firmness PM1, a cell or inspection station of the external appearance PM2 and of a cell or station measuring the PM3 sugar level.

 On the right side of Figure 3, there is also shown, following the cell or measuring station of the PM3 sugar level, a robotic station BR trays fruit examined.

1) Firmness and weight measuring station
The measuring cell PM1 of the weight and the firmness of the fruits F which, according to the advantageous example illustrated, is disposed below the exit of the diabolos conveyor 4, comprises (FIG. 9) a device for retaining the fruit F1 arriving. at the end of said conveyor, to prevent it from falling before the desirable moment.

This programmed separate fall of each fruit is, for example, obtained by means of a wheel of flexible material 5 controlled by a control automaton and disposed above the downstream end of the diabolo conveyor 4.

 The PM1 cell for measuring weight and firmness still essentially comprises (FIGS. 9 and 10) a weighing platform or support 6 on which the fruit falls from a low height (one to two cm), a strain gauge 7 placed under this plate and connected to a computer, a fixed barrier 8 bordering the plate 6 on two perpendicular sides, a movable ejection fork 9 mounted with a pivotability around a fixed vertical axis 9a, and a pneumatic cylinder 10 or similar actuator connected to the ejection fork and for ejecting the fruit to a first gripper V1, by pivoting said ejection fork. As shown in Figure 10, one of the branches of the fork 9 constitutes the ejector itself, while its second branch is a movable protective barrier.

 The slight drop of the fruit on the plate 6 generates a brief signal on the strain gauge: it is a damped sinusoid, which lasts between 20 and 40 ms, and is recorded and analyzed by the computer. A specific algorithm deduces the firmness of the fruit. After about a second, the bounces are over and there is only one continuous signal on the strain gauge: this signal represents the weight of the fruit.

 With this method and arrangement, the firmness and weight of the fruit are measured using the same device.

2) Station for measuring external characteristics
According to the method of measuring the external qualities of the fruits, each fruit is seized at its exit from the PM1 station by a first gripper which brings it to the PM2 inspection station and is then rotated around the PM2. axis of its gripping head so as to expose its different visible faces in the field of view of a camera or other artificial vision system, then said fruit is grasped in the position where it is at the end of this rotation, by a second gripper which is then also rotated about the axis of its gripping head, in order to place the face of the previously masked fruit, in the field of vision of said artificial vision system, after which this second gripper brings the inspected fruit into a position where it can be grasped by a tray setting device.

 The cell or inspection station for the external characteristics of the fruits comprises (FIG. 6) a lighting chamber 11, for example consisting of an integrating half-sphere, and in the lower part of which an annular fluorescent light is installed. 12. With an annular reflector 13 disposed in the space defined by the annular lighting tube, it avoids direct lighting of the fruits. The inner wall of the hemispherical chamber It is provided with a white diffusing coating, which guarantees a very uniform illumination of the fruit F3 during inspection.

 A hole 14 is formed in the top of the chamber 11, and an artificial vision system, for example constituted by a color camera 15 or a camera combining color and infrared channels, is fixedly disposed above this hole. In this way, the open central zone of the lower part of the chamber 11 is placed in the field of view of the camera 15, and constitutes the inspection place of the fruits which are placed under uniform indirect lighting, on all their visible surface, during their inspection, which avoids glosses of specular origin.

 According to an important characteristic arrangement of the invention, each fruit is presented at the inspection location, in the field of view of the camera 15, by means of two grippers V1, V2 having a degree of freedom of tilt and a degree of of freedom of rotation in the axis aa, a'-a ', respectively, of their gripping head, one of these grippers (for example gripper V1) for grasping each fruit and to orient, by a movement of rotation about the axis aa, most of the fruit surface, towards the artificial vision system, said fruit being then grasped by the second gripper (for example V2 gripper) which also directs by a rotational movement around from its axis a'-a ', the remaining surface of the previously hidden fruit, towards said artificial vision system.

 The grippers Vi, V2 are preferably constituted by suction cups, the suction energy being supplied, in a manner known per se, by an electric turbine (not shown) specific to each suction cup. The suction is controlled by a solenoid valve, also in a manner known per se.

 In order to be able to bring the fruits to inspect in the lighting chamber 11, from below, without touching the annular illumination, a specific kinematics has been adopted for the tilting of the support of each gripper V1, V2.

 The principle of this kinematics is illustrated in Figures 7A and 7B, which represent two different positions of the same gripper. The support of each gripper consists of two articulated moving parts 16 and 17.

Two toothed pulleys R1 and R2 and a synchronous belt 18 ensure the kinematics: the pulley R1 is secured to the fixed frame, while the pulley R2 is secured to the support 16. The support 17 is rotated by an actuator (not shown), for example an electric motor, the drive shaft 19 is concentric with the axis of the toothed pulley R1. We consider a starting position (Figure 7A) characterized by a gamma angle between the two supports.

When, from this position, the support 17 rotates at an angle alpha (FIG. 7B), the support 16 moves, and is driven in its own rotation by the pulley R2, and rotates in the opposite direction, a beta angle. The ratio between the angles, alpha / beta, is equal to the ratio of the diameters of the pulleys d (R2) / d (R1).

This ratio is preferably less than one and equal to 2/3 according to an interesting embodiment.

 The rotation of the grippers V1, V2 about their axis a-a, a'-a ', respectively, is provided by a motor 21 and a reduction gear 22 carried by the support 16 of said grippers (Figure 11).

An object of the machine and method of the invention is to allow to visualize the fruit in four successive views. For this, we proceed as follows;
The grippers V1, V2 are brought, via the actuators described above, in a position according to which their gripping heads V1 'and V2' are close to one another and their axes aa and a'-a ' form an optimum angle of the order of 1 20a.

 The fruit being held by the gripper V1, three views are taken successively by the camera 15, each of these views being separated from the next by a rotation of 1200 in the axis a-a of said gripper. These views referenced I1 to I3 are shown in FIGS. 8A to 8C, where the shaded areas represent the parts viewed in at least one image.

 The fruit is then transferred (FIG. 8D), without changing position, in the gripper V2, by switching the solenoid valves controlling the grippers V1 and V2 (the solenoid valve driving V1 cuts the suction, while the solenoid valve driving V2 is set in aspiration).

 The fruit is rotated 1800 by rotation of the gripper V2 around its axis a'-a '(Figure 8E), to view the previously masked underside of said fruit.

 We take a last view I4, perfectly centered on the area not yet visualized. Figure 8F illustrates this view in the axis of the camera, showing in hatched lines, the areas visible in at least one of the three previous views, for a perfectly spherical fruit. The size of the suction cup V1 is represented by an inner circle CI. We control well and that the visualization of the fruit is total.

Inspection by camera or other artificial vision system of the entire surface of the fruit provides the system with information to orient the fruit. The points of interest potentially recognized by the vision system are the colored faces, the calyxes and the peduncles, and the defects of aspect. This orientation makes it possible to present a preferential face of the fruit: - to the NIR (Near Infra-Red) sensor to optimize the measurement of sugar,
when the sorting facility includes a sugar level measurement station; - to the end customer, in the commercial container, to enhance the value of the
product.

3) Measuring the sugar level
According to the method of the invention, the measurement of the sugar level is made by bringing into contact with the fruit an optical interface ensuring

 After the visual inspection, the fruit, still held in the V2 gripper, is brought to the PM3 sugar station (Section C). The measurement is carried out by near-infrared spectrometry, a method known per se. The principle is to bring a strong infrared lighting under the skin of the fruit, then to analyze by spectrometer and by computer the composition of the light backscattered by the fruit. The spectrum of this light is characteristic of the sugar level of the area examined.

 An optical fiber bundle 23 having a Y structure known per se, brings light and collects the scattered light. for this purpose, it mixes illumination fibers from a light source, and measurement fibers which reduce the backscattered signal by a spectrometer.

 The fiber bundle 23 is mounted on a mobile support 24 whose forward movements towards the fruit or recoil are performed by means of a pneumatic cylinder 25 or a similar actuator whose movable member ensures good contact of the fiber with the skin of the fruit, this good contact is necessary to make a meaningful measurement. The compression of the cylinder ensures the permanent contact between fruit and fiber bundle.

For proper operation, the spectrometer requires two conditions fulfilled by the invention - to allow automatic reference measurement with a known object. In
this purpose is circulated in the machine according to the method of
the invention, for example at regular intervals, a reference object. We can
so use an opaque ball and that does not absorb light, which the
diameter is close to that of the fruits. To take the reference measurement,
it is circulated in the machine instead of a fruit, by arranging the
handling cycle. For example, we can make the first "fruit" of everything
lot examined.

- allow periodic cleaning of the end of the fiber bundle:
dust or greasy films on the surface of the fruit can reduce the
sharpness of the signal after a few dozen fruits. It is therefore essential to
to be able to periodically modify the sort cycle to introduce an object of
cleaning. It is also planned, for this purpose to use a ball, to
diameter close to that of fruits, and rough surface: the ball is seized
by the gripper V2, then rubbed dry against the end of the beam of
optical fibers, by a programmed reciprocating movement of said gripper.

For convenience, the cleaning object automatically describes a cycle
complete, with return to the starting point. This point is advantageously
constituted by a support accessible to the robotic crating arm described
below.

 The invention allows to arbitrarily choose, around the equator of the fruit, one or more measuring points, turning the gripper V2 of any angle.

 Finally, as indicated above, using the information from the vision, which in particular gives the position of the calyxes and peduncles, one can minimize the risk of making measurements of the sugar level on these points, by appropriate rotations of the two grippers controlled by the central unit which manages the operation of the actuators thereof.

Tray station
The PP trays of the machine (FIGS. 1, 2, 3 and 4) comprise - a spherical robot with three axes - a system for automatic handling of fruit trays.

1) Spherical robot
The spherical robot represented very schematically and designated by the reference 26 in FIGS. 3 and 4, is a spherical coordinate manipulator (two concurrent rotations and a translation towards the object to be grasped and / or towards the deposit location of the examined fruit. ). It is described in detail in the document
EP-A-0270469.

 This spherical coordinate manipulator is suspended, in a vertical position, at a gantry 27 disposed at a level higher than that of the grippers V1 and V2 of the inspection stations PM2 and PM3. According to a first exemplary embodiment, it has no vision system, so that, in this case, it always takes the fruits at a fixed exit point. Its geometry allows it to access a large volume of work illustrated by the twelve trays visible in the center of Figure 1. According to one embodiment, it arranges the fruit on a single layer in twelve trays dimpled, each with a capacity from 20 to 30 fruits. The spherical robot is controlled by a specific computer (see below).

It is enough to indicate to the computer the position of a corner of each plate, as well as some dimensions characterizing the system of cells, and the conditioning is completely automatic. The cell type can be reprogrammed instantly by the system operator.

 This example is in no way limiting. The same arm could also access, for example, three boxes of 15 to 20 kg of fruit, and store the fruits on three or four layers. According to another embodiment, it is possible to equip the robotic arm with a vision system, as also described in document EP-A-0,270,469. The vision system then makes it possible to automatically identify the positions and orientations of the boxes, provided that they remain in its work space, as well as the state of the filling at any time.

2) Tray handling
The system for handling and automatically evacuating the fruit trays is organized according to the invention as follows:
Two identical main structures placed face to face, support the trays, and the mechanisms ensuring their movements. Each structure serves 6 trays, with an inclination that allows easy access by the spherical robot arm, ie an inclination substantially perpendicular to its working axis (Figure 4).

 The empty trays P are fed from a subassembly (20A, 20B 20C, 20D) (FIG. 1), composed of a conveyor belt which can be raised and lowered, actuated for example pneumatically, towards the low or high positions. for respectively supplying the lower positions (28B, 20D) or the upper positions (28A, 28C).

 The empty trays are transmitted to the belt conveyors 29A to 29D, which transport them to forks 28A to 28D, in the arms of which roll guides are arranged, so that the trays descend by gravity to their PI filling position.

 These forks 28 can pivot, for example with pneumatic actuators, with respect to the axes 35A, 35B (FIG. 5) so that, once the trays are filled, they descend, from the working position P1 to the emptying position P2. Due to their fork-shaped design, their free inner part encounters, in its downward movement, the belt conveyors 30A to 30D, on which are deposited the trays that were carried by the roller guides of the forks.

 These conveyors 30 are actuated in the opposite direction of the conveyors 29, and take the trays to the evacuation positions P3, where roll conveyors 36A, 36B, etc., oriented perpendicularly to the preceding ones (FIGS. 1 and 5), which are arranged, are arranged. the full trays of the system. For this purpose, these roller conveyors, preferably controlled by pneumatic pistons, have a movement up and down the entire structure. On going up, they separate the full trays of belt conveyors 30, placing them on the rollers, which, mechanically actuated, carry the trays transversely (PS position of Figure 1).

 It is perfectly possible to evacuate a full tray while launching the simultaneous supply of an empty tray, ie without stopping the conditioning cycle, which makes it possible to avoid downtime in the cycle.

Control architecture
The machine according to the invention operates with any computer architecture that fulfills the logical and algorithmic functions indicated in the previous description, as well as the sorting cycle described below. In a preferred embodiment, central processing units (CPUs) of computers and
API, in a modular way:
a UC supervision for the overall supervision of the line and the interface
user; o a CU inspection for the inspection station, weighing, and measurement of
firmness;
a vision UC for vision
a sugar CU for the measurement of sugar; a storage unit for robotic storage of fruits in trays;
a handling unit for handling empty and full trays;
three APIs for the entry station.

The dialogue between the CUs is organized in a heterogeneous network with the following characteristics, which constitute industrial standards guaranteeing the reliability
physical support = Ethernet
dialog protocol = TcP JlP.

The functional hierarchy follows the following schema (where the arrow denotes a master => slave).

UC supervision => UC inspection => UC vision
=> UC sugar
=> UC storage
=> UC handling
=> entry point
The structure described above is not limiting; in particular, one can change the order of passage in front of the sensors or the number of reorientation gestures of the fruit, without changing the spirit of the invention.

 An interesting example of carrying out the inspection cycle is described below, which constitutes a good compromise between measurement accuracy and work rate.

 First of all, according to FIG. 3, the machine is designed for a series organization of the cycle, making it possible to simultaneously process 4 fruits. Each of sections A, B, C and D can process one fruit at a time.

In this diagram, the critical steps to optimize are the transfers between sections
transfer Entry =>A;
transfer A
transfer B
C transfer
The simplest is to describe each of the 4 cycles involving a
section, knowing that a given fruit passes successively in each section A,
B, C and D:
Cycle section A:
=> Starting the entry station
Waiting for fruit impact (Transfer Entry => A)
Firmness measure of the fruit
Waiting for fruit stabilization
Weighing the fruit
Ejection by jack to gripper V1 (transfer A => B)
Return of the ejector cylinder
+ in parallel: stop of the entry station after a pre-set time after
the impact.

 The purpose of the pre-set time is to bring the next fruit as close as possible to the fall on the impactor, but without success.

Cycle section B:
Reception of the fruit in V1 (transfer A => B)
Tilt from V1 to vision
=> Acquisition Image It
V1 rotation of 1200, and image processing I1
I2 image acquisition
V1 rotation of 1200, and I2 image processing
I3 image acquisition
I3 image processing
V1 rotation for optimized orientation using data from images I1 to
I3, in order to prepare the final orientation of the fruit in the tray
(optional)
Transfer of fruit V1 to V2 (transfer B => C)
Return of V1 to section A
Cycle section C:
Reception of the fruit in V2 (transfer B => C)
Rotation of the fruit of 1800 in the axis of V2 to visualize the hidden part
I4 image acquisition
Failover from V2 to NIR, and image processing I4
=> Optical fiber jack approach or other optical interface
Acquisition of the measurement of the sugar level
Treatment of the measurement of the sugar level
Retraction of the optical fiber jack or other optical interface
Tilting of V2 to robotic arm 26.

Transfer of fruit held by V2 to arm (transfer C => D) Return of V2 to vision
If a second sugar measurement is necessary, the following sequence must be inserted before switching V2 to arm
r V2 rotation of 90 "
Approaching the optical fiber jack or other optical interface.

Acquisition of the sugar measure Treatment of the sugar level
Retraction of the optical fiber jack or other optical interface.

Cycle D section:
=> Bring the arm to V2
Reception of the fruit in the arm (transfer C => D) Place the fruit in the appropriate tray.

 In the foregoing description, the starting points of each cycle are arbitrary. The indicated order is the one that best facilitates understanding.

However, this order is not appropriate in practice, because it is obvious that the transfers between two sections must be synchronized in the real cycles.

For example, the transfer A => B is the penultimate operation of the section
A. It also appears as the first operation of section B. In the order indicated, it would be necessary to wait until the end of cycle A to start cycle B, and one would lose the parallelism.

 To minimize waiting times, starting points should be placed differently in each section. In the embodiment described, the starting points actually implemented are marked by an arrow (=>).

 In the previous example, cycle B now has its transfer A => B two operations before the end. The cycles of sections A and B can be performed almost perfectly in parallel.

 The resulting cadence of the machine is best given by the slowest cadence of each section. It is of the order of 1000 to 2000 fruits per hour.

 According to one characteristic arrangement, the information from two different sensors is used to perform a multisensory merging in which the digital data provided by the two sensors are combined to obtain statistically more accurate measurements than the original data.

 The multisensory fusion applied in the invention consists in exploiting the correlations existing between the data of different sensors; for example, a color change and an increase in sugar can both be related to fruit ripening. By exploiting the color data of the fruit, we can then make a prediction of the sugar level more accurate than with spectrometry alone. The fusion is implemented by a neural network algorithm that works by learning: with a batch of fruits whose characteristics are perfectly known (thanks to destructive measurements); he finds the combination of the two data that gives the best measurement prediction. Then, this combination is applied to other fruits.

 Fusion is particularly useful in the case of unclear raw data, which is the case of non-destructive measurements.

 A very important particular case is where the raw data of one of the sensors is invalid: the fusion process then proposes a backup measure based on one or more other sensors.

 The invention poses a specific problem of multicriteria classification, because it brings into play for the first time a large number of sensors online.

 According to another characteristic arrangement, the information coming from at least two different sensors is combined in the form of a multi-criteria classification, the number of final classes being substantially less than the number of classes obtained by a simple combination of the value classes retained for each sensor.

The problem will be better understood by an example: a classic tomato grader uses 8 color classes and 5 template classes: it has 40 outputs, corresponding to all possible caliber / color pairs. If now we introduce 3 classes of appearance defects, 3 classes of sugar, and 2 classes of firmness, the same method leads to 360 final classes, which is unmanageable, both in production and in marketing.
The principle adopted for the classifier is therefore the following
a final class is defined by all the fruits that respect a fork
given values for each of the measurements made;
for each measure and each class, the bounds of the range are
provided by the user and can be modified at will in a table of
quality
moreover, a priority system exists: the program first checks whether the fruit
is acceptable in class N "1, then in class N" 2, etc ... and it stops as soon as it has
found a suitable class;
a special class (discards) is assigned to fruits that have not satisfied any
class.

 For greater flexibility, it is possible for lists of ranges of different values to return to the same class. For example: very large or very small fruits = class outside caliber.

 According to the exemplary embodiment illustrated for the invention, twelve classes are implemented, which corresponds to the number of trays served. The only constraint is to have at least as many output paths as classes, we can use less than twelve classes.

Claims (27)

 packing station (PP) sorted fruits or vegetables.  measuring station (PM1, PM2, PM3) disposed between said input station and said  automatically, during a step of stopping said fruits or vegetables at a  size, color, appearance, density, weight, firmness, sugar content, is operated  of their characteristics included among the following  sorted fruits or vegetables, characterized in that measuring at least one of  one by one, between an entry station (PE) and a packing station (PP)  1. Process for sorting fruits and vegetables with a scroll movement, 2. Sorting method according to claim 1, characterized in that one proceeds to  a measure of the weight and firmness of the fruits or vegetables, making  drop said fruits or vegetables, from a low height, one by one, on a  plate (6) resting on a strain gauge (7), the damped sinusoid  created by this fall on the strain gauge being recorded and analyzed  by an algorithm to deduce the firmness of the fruit or vegetable of which  the weight results from the continuous signal generated by said strain gauge  when the bounces of said fruit or vegetable are finished. 3. Sorting method according to one of claims 1 or 2, characterized in that  each fruit or vegetable is presented at an inspection post (PM2) in the  field of view of an artificial vision system (15), by means of two  grippers (V1, V2) having a degree of freedom of tilt and a  degree of freedom of rotation in the axis of their gripping head, one of these  grippers for grasping each fruit and guiding, by a  rotation, most of the fruit surface, in the direction of  artificial vision system, said fruit being then grasped by the second  gripper which also directs, by a rotational movement, the surface  remaining previously hidden, towards said vision system  artificial. 4. Sorting method according to any one of claims 1 to 3, characterized  in that each fruit or vegetable is presented at a rate measurement station  of sugar (PM3) where at least one contact sensor constituted by a  optical interface (23) providing illumination and / or light collection, is  brought into contact with each fruit or vegetable, said optical interface being  connected to an infrared spectrometer itself connected to a computer. 5. Sorting method according to one of claims 3 or 4, characterized in that  the information collected at the inspection station (PM2) concerning the  of each fruit or vegetable, are used to choose a preferred surface  of the latter, to present either at the optical interface (23) or in the  final packaging of said fruit or vegetable. 6. Sorting method according to any one of claims 3 to 5, the  vision system is constituted by a color camera (15), characterized in that  visualization of the entire surface of the fruits  or spherical vegetables, with four shots  successive. 7. Sorting method according to any one of claims 3 to 6, characterized  in that the fruits or vegetables are placed under indirect lighting  uniform throughout their visible surface during their inspection by the  of vision (15). 8. Sorting method according to claim 4, characterized in that one carries out  automatically periodic cleaning of the part of the optical interface  (23) which is in contact with fruits or vegetables. 9. Sorting method according to claim 5, characterized in that the choice of the  preferential face of the fruit or vegetable is operated by appropriate rotations  grippers (V1, V2) controlled by the central unit which manages the  operation of these. 10. Sorting method according to any one of claims 4 to 9, characterized  in that information from at least two different sensors,  are used to perform a multisensory merge in which the  digital data provided by the two sensors, are combined to  obtain statistically more accurate measurements than the original data. 11. Sorting method according to any one of claims 4 to 10,  characterized in that the information from at least two sensors  different, are combined in the form of a multi-criteria classification, the  number of final classes being significantly lower than the number of  classes obtained by a simple combination of the value classes retained  for each sensor. Sorting method according to one of Claims 1 to 11, characterized  in that the supply of the measuring station or the whole of the station of  measurement (PM) by the entry station (PE) is carried out according to a method  sequential and asynchronous, each fruit or vegetable not entering the said  measuring station or in said measuring station set, only  when he is ready to treat it. 13. A fruit or vegetable sorting machine comprising an entry station (PE) and a  packing station (PP) of the sorted fruit or vegetables, between which  at least one measuring station (PM) of at least one of the  following characteristics of the said fruits or vegetables, size, color,  appearance, density, weight, firmness, sugar content, said machine comprising  means ensuring the one-to-one scrolling of the latter, characterized in  what it is equipped with means to stop automatically and  momentarily the scrolling of each fruit or vegetable as a result of its  arrival at the measuring station (PM) or at each measuring station (PM1, PM2,  PM3), during the duration thereof. 14. Sorting machine according to claim 13, characterized in that it comprises  a measuring station (PM1) for the firmness and weight of the fruits or vegetables,  the measuring station (PM1) comprising a weighing platform or support (6)  based on a strain gauge (7) connected to a computer. 15. Sorting machine according to claim 14, characterized in that said station  measuring the firmness and weight of the fruit or vegetables, includes a  device allowing a programmed separate fall of each fruit or  vegetable, on the weighing platform or support (6). 16. Sorting machine according to claim 15, characterized in that said  device is constituted by a wheel of flexible material (5) disposed at  above the downstream end of a diabolo conveyor (4) of the entry station,  and slave to a controller. 17. Sorting machine according to any one of claims 13 to 15,  characterized in that an ejection fork (9) is secured to an actuator  (10) ensuring its pivoting and the ejection of the weighed fruit or vegetable. 18. Sorting machine according to any one of claims 13 to 17,  characterized in that it comprises a measuring station of  external characteristics of fruits (PM2) with a vision system  artificial, for example constituted by a color camera (15) connected to a  computer and in the field of view of which can be brought two  grippers (V1, V2) each having a degree of freedom of tilting  limited amplitude and a degree of freedom of rotation in the axis (a-a, a'-a ') of  its gripping head (V1'-V2 '), these grippers (V1, V2) being subject to  actuators (16, 17, 18, 19, R1, R2; 21-22) providing, respectively,  their tilting and rotation and can be brought into a position  according to which their gripping heads (V1'-V2 ') are close to one of  the other and their axes (a-a, a'-a ') form an optimum angle of the order of 1200 Sorting machine according to claim 18, characterized in that the  grippers (V1, V2) consist of suction cups connected to a  suction system controlled by solenoid valves. 20. Sorting machine according to one of claims 18 or 19, characterized in that  that the inspection station of the external characteristics of the fruits or vegetables  (PM2) comprises a lighting chamber preferably having a shape  hemispherical, and in the open lower part of which is arranged a  annular fluorescent lighting (12) on the inner side of which is placed a  annular reflector (13), this chamber comprising, on the one hand, on its wall  internally, a white diffusing coating and, at its apex, a hole (14) at  above which is located the artificial vision system (15) whose  field of vision corresponds to the central opening (1 la) of said chamber  (11). 21. Sorting machine according to any one of claims 18 to 20,  characterized in that each gripper (V1, V2) is carried by a support  consisting of two moving parts (16, 17) of which one (16) is integral with a  toothed pulley (R2), on which a synchronous belt (18) is wound  winding on a second toothed pulley (R1) secured to the fixed frame of the  machine, the other moving part (17) being rotated by an actuator (19). 22. Sorting machine according to any one of claims 13 to 21,  characterized in that it comprises a measuring station of a quality  fruit or vegetables, such as their sugar content, characterized in  it comprises at least one optical interface (23) ensuring lighting  and / or light collection, such as, for example, a bundle of fibers  Y-optics or an integrating sphere, said optical interface being connected  to a spectrometer itself connected to a computer. 23. Sorting machine according to claim 22, characterized in that the source of  light and the spectrometer are, respectively, a light source  infrared and an infrared spectrometer. 24. Sorting machine according to any one of claims 13 to 23,  including a station (PP) for automatic and individualized packaging  fruits or vegetables, in cases or trays (P), characterized in that  said station comprises a spherical coordinate manipulator (two  rotations and a translation towards the place intended for the reception of the fruit  or vegetable). 25. Sorting machine according to any one of claims 13 to 24,  characterized in that it comprises a handling assembly  automated trays or crates (P) for packaging fruit or  vegetables, to evacuate any solid tray and to supply the said  together with a new empty board, without stopping the cycle of  packaging, this set comprising, for each tray, a  supply conveyor (29A, 298,.), a tray holder  tilting (28A, 28B,.), an evacuation conveyor (30A, 30B,.) disposed  below and remote supply conveyor, and means  (31A, 31B, ...) allowing the tilting and the positioning of the support  tray, either in the loading position or in the position of  unloading. 26. Sorting machine according to claim 25, characterized in that the set of  automated handling of trays, includes roller conveyors  (36A, 36B,.), Disposed at the ends of the evacuation conveyors (30A,  30B, ...) and oriented perpendicular thereto. 27. Sorting machine according to any one of claims 13 to 18,  comprising an entry station (PE) having at least one conveyor  ginner (2) consisting of two mats forming a V and moving to  different speeds, characterized in that two lateral stops (3)  equip the fixed longitudinal walls of said conveyor-ginner, these  stops being offset along the axis of circulation of the latter.