EP3164344B1 - Device and method for retrieving objects, in particular waste, using sound - Google Patents

Description

Technical field of the invention

The invention relates to the field of the treatment of objects and more particularly waste.

The invention more particularly relates to an object recovery device for characterizing an object inserted into a container of said recovery device.

State of the art

Our mode of consumption leads to the generation of more and more waste. It has become crucial to organize collection systems to recover this waste for recovery, and preferably to encourage recycling.

Recovery, or recycling, requires the identification of waste.

There are solid waste sorting techniques that use optics to recognize an object examined by image processing.

In this sense, the document FR2905880 discloses an object sorting device comprising a visual recognition station of an object by video means, an identification station of the composition of the object using a spectrometer, a metal differentiation station constituting the object using a inductive sensor, and a position measuring the weight of the object. The use of many sensors necessary for the various positions induces a prohibitive cost of the sorting device.

• la nécessité d'utiliser des systèmes d'éclairage idoines,
• la complexité des algorithmes de reconnaissance de forme,
• le coût,
• et l'entretien dû au nettoyage fréquent en vue de désencrasser l'optique et les systèmes d'éclairage. En outre, le document CN201808857 divulgue un dispositif et un procédé selon les préambules des revendications 1 et 7. Ce document propose en variante de reconnaître un objet inséré dans un contenant par un système acoustique enregistrant le signal sonore de l'objet inséré. L'efficacité d'un tel système reste insuffisante.

In addition, the use of optics has certain disadvantages:

• the need to use appropriate lighting systems,
• the complexity of shape recognition algorithms,
• the cost,
• and maintenance due to frequent cleaning to unclog optics and lighting systems. In addition, the document CN201808857 discloses a device and a method according to the preambles of claims 1 and 7. This document proposes alternatively to recognize an object inserted into a container by an acoustic system recording the sound signal of the inserted object. The effectiveness of such a system remains insufficient.

In this sense, there is a need to develop new techniques for characterizing an object, and not having all or some of the disadvantages listed above.

Object of the invention

The object of the present invention is to propose a solution that overcomes all or part of the disadvantages listed above.

This object is aimed in particular by means of a device for recovering objects, in particular waste, comprising a container in which said objects are intended to be inserted and a characterization module of an object inserted in the container, said device comprising an acoustic system configured to record a sound signal representative of the interaction of the object inserted with the recovery device said acoustic system being connected to said characterization module so that the sound signal participates in the characterization of said inserted object. In addition, the device comprises a lid mounted on the container and whose opening allows the introduction of the object to be inserted into the container, said acoustic system being configured to record a first component of the sound signal representative of the less an interaction of the object to be inserted with the lid. The characterization module comprises an element for determining the material of the inserted object configured to exploit the sound signal, said determining element comprising a setting element, at a predetermined cutoff frequency value, a signal ratio of the low frequency energy of said first component on a total energy of said first component, and a comparison element of the established ratio with a predetermined ratio threshold value. In addition, the acoustic system may be configured to record a second component of the sound signal representative of at least one return of the lid in a closed position after insertion of the corresponding object into the container and / or a third component of the sound signal representative of at least one shock of the object inserted in the container at the end of its fall into said container.

Preferably, the characterization module is configured to study the first and / or second and / or third components of the sound signal to characterize said inserted object.

According to an exemplary embodiment, the acoustic system comprises a microphone, in particular of Electret or digital MEMS type.

In addition, the characterization module may include an element for estimating the size and / or the weight of the inserted object using said sound signal.

According to an improvement, the device comprises a payment module, in particular financial module, of a user having inserted the object into the container, said payment module being connected to the characterization module so as to use at least one piece of information relating to the characterization of said inserted object to evaluate compensation.

According to an improvement, the device may comprise a module for determining the filling of the container, in particular configured to use at least one data from the acoustic system and / or the characterization module.

• ∘ une détermination, à partir de la première composante du signal sonore et à une fréquence de coupure prédéterminée, d'un rapport de l'énergie basse fréquence de ladite première composante sur une énergie totale de ladite première composante, et
• ∘ une comparaison du rapport déterminé avec une valeur seuil de rapport prédéterminée,

The invention also relates to a method for recovering objects, in particular waste, in a recovery device provided with a container, said method comprising a step of inserting an object into the container and a characterization step the inserted object having a step of studying a sound signal resulting from the interaction of the inserted object with said recovery device. According to the present invention, the study step comprises the use of a first representative component of at least one interaction of the object to be inserted with a lid of the recovery device mounted on the container, and comprises a step of determining the material of the inserted object comprising:

• A determination, from the first component of the sound signal and at a predetermined cut-off frequency, of a ratio of the low-frequency energy of said first component to a total energy of said first component, and
• A comparison of the determined ratio with a predetermined ratio threshold value,

• une deuxième composante représentative d'au moins un retour de l'opercule dans une position de fermeture après une insertion de l'objet correspondant dans le contenant, et/ou
• une troisième composante représentative d'au moins un choc de l'objet inséré dans le contenant à la fin d'une chute dudit objet inséré dans le contenant.

The study step may comprise the use of at least one of the following components of the sound signal chosen from:

• a second component representative of at least one return of the lid in a closed position after insertion of the corresponding object into the container, and / or
• a third component representative of at least one shock of the object inserted in the container at the end of a fall of said object inserted in the container.

• une estimation de la taille de l'objet inséré comprenant :
  • ∘ une détermination à partir des première et deuxième composantes d'une durée entre un instant de premier contact de l'objet inséré avec l'opercule et un instant de fermeture de l'opercule après insertion de l'objet dans le contenant, et
  • ∘ une comparaison de la durée déterminée avec au moins un seuil de durée,
• une caractérisation du matériau de l'objet inséré comprenant, à partir de la troisième composante du signal sonore, une détermination de fréquences dominantes et/ou d'une énergie multi-bandes et/ou d'un rapport d'énergie,
• une détermination du poids de l'objet inséré à partir de l'énergie de la troisième composante du signal sonore.

In addition, the study step may further comprise at least one of the following steps:

• an estimate of the size of the inserted object including:
  • ∘ a determination from the first and second components of a duration between a moment of first contact of the inserted object with the lid and a closing instant of the lid after insertion of the object in the container, and
  • ∘ a comparison of the determined duration with at least one duration threshold,
• a characterization of the material of the inserted object comprising, from the third component of the sound signal, a determination of dominant frequencies and / or of a multi-band energy and / or of an energy ratio,
• a determination of the weight of the object inserted from the energy of the third component of the sound signal.

Preferably, the step of characterizing the inserted object output: the material constituting the inserted object and / or the size of the inserted object and / or the weight of the inserted object.

According to an improvement, the method may comprise a step of estimating the filling of the container using at least one acoustic data.

• une détermination, à partir de ladite au moins une donnée acoustique, d'un nombre de rebonds moyens réalisés dans le contenant par les objets insérés,
• une détermination, à partir de ladite au moins une donnée acoustique, d'une moyenne de durée de chute des objets insérés dans le contenant,
• une détermination, à partir de ladite au moins une donnée acoustique, du nombre d'objets insérés dans le contenant.

Preferably, the step of estimating the filling comprises at least one of the following steps:

• a determination, based on said at least one acoustic data, of a number of average rebounds made in the container by the inserted objects,
• a determination, based on said at least one acoustic data, of an average duration of fall of the objects inserted into the container,
• a determination, from said at least one acoustic data item, of the number of objects inserted into the container.

According to another improvement, the method may comprise a step of valuing the inserted object in which at least one data item from the characterization of the inserted object is used to credit an account, in particular a user account belonging to the person who has inserted the object within the container.

Brief description of the drawings

• la figure 1 représente un dispositif de récupération selon un mode d'exécution particulier,
• la figure 2 est un diagramme représentant différentes étapes d'un procédé de récupération d'objets,
• la figure 3 représente l'amplitude en fonction du temps d'un signal sonore issu du système acoustique,
• la figure 4 illustre un rapport de l'énergie de la première composante en fonction de valeurs de fréquences de coupure,
• la figure 5 illustre l'amplitude de trois troisièmes composantes respectivement associée à un tas de journaux, un tas de magazines et un annuaire en fonction du temps,
• la figure 6 illustre l'énergie de trois troisièmes composantes respectivement associée à un tas de journaux, un tas de magazines et un annuaire en fonction de numéros de trame,
• la figure 7 illustre la durée moyenne d'une chute d'une petite bouteille dans le contenant en fonction du pourcentage de remplissage du contenant.

Other advantages and features will emerge more clearly from the following description of particular embodiments of the invention given as non-restrictive examples and shown in the accompanying drawings, in which:

• the figure 1 represents a recovery device according to a particular embodiment,
• the figure 2 is a diagram representing different steps of an object recovery process,
• the figure 3 represents the amplitude as a function of time of a sound signal coming from the acoustic system,
• the figure 4 illustrates a ratio of the energy of the first component as a function of cutoff frequency values,
• the figure 5 illustrates the amplitude of three third components respectively associated with a pile of newspapers, a pile of magazines and a directory according to time,
• the figure 6 illustrates the energy of three third components respectively associated with a pile of newspapers, a pile of magazines and a directory according to frame numbers,
• the figure 7 illustrates the average duration of a fall of a small bottle in the container according to the filling percentage of the container.

Description of preferred modes of the invention

The device and method described hereinafter differ from the prior art in that it is proposed to use acoustics to characterize a recovered object. Such use makes it possible in particular to reduce the costs by using a single or several microphones to perform all or part of the operations referred to in the positions of the prior art. This also results in a compactness of an associated characterization module of the object.

In this sense, an object recovery device 1, in particular waste, such as that illustrated in FIG. figure 1 comprises a container 2 in which said objects are intended to be inserted and a characterization module 3 of an object inserted in the container 2. The device 1 comprises an acoustic system 4 configured to record a sound signal representative of the interaction of the object inserted with the recovery device 1 (in particular the container 2 and / or a lid 5 mounted on the container 2). This acoustic system 4 is connected to said characterization module 3 so that said sound signal participates in the characterization of said inserted object. Thus, it is understood that the characterization module 3 comprises the software and / or hardware means for analyzing the sound signal from the acoustic system 4 in order to characterize the inserted object. For example, the characterization module 3 may comprise a microcontroller or a DSP (Digital Signal Processor) processor, and so on. The use of an acoustic system 4 greatly simplifies the device compared to what exists in the state of the art. The acoustic system 4 comprises a microphone, in particular Electret type or MEMS (Microsystem electromechanical) digital. Preferably, the acoustic system 4 comprises a single microphone, this to reduce integration costs. In general, the acoustic system 4 may include an analog to digital converter (also known by the acronym ADC in the field) connected to the microphone so that the sound signal is a digital signal, a digital signal amplifier, and a high pass filter applied to the sound signal (in its digital version) after amplification. Therefore, the sound signal studied by the characterization module 3 to characterize the inserted object is that at the output of the high pass filter.

The digitization of the sound signal coming from an Electret microphone is done, for example, using an ADC Alternately, it is also possible that the microphone is a digital MEMS and in this case the sound signal is directly exploitable.

More generally, in the context of a process for recovering objects, in particular waste, in a recovery device 1 (in particular as described) equipped with a container 2, the latter comprises ( figure 2 ) an insertion step E1 of an object in the container 2 and a characterization step E2 of the inserted object, implemented by the characterization module 3. The characterization step E2 of the inserted object comprises a E2-1 study step of a sound signal resulting from the interaction of the inserted object with the recovery device (including the container 2 and / or the cap 5 mounted on the container 2). It is then understood that, implicitly, the method comprises, prior to the study step E2-1, a step of acquisition of the sound signal to be studied by the acoustic system 4.

By "characterizing the inserted object" is meant, for example, the determination of the material constituting it (glass, paper, plastic, metal, etc.) and / or its size and / or volume and / or its weight. In the context of the sorting of objects, "characterizing the inserted object" can, for example, make it possible to determine whether the object is in a particular material, or not (in the case of sorting the glass in the example of the figure 1 ) The device 1 comprises a lid 5 mounted on the container 2 and whose opening allows the introduction of the object to be inserted into the container 2. The acoustic system 4 is then configured to record a first component of the sound signal representative of at least one interaction of the object to be inserted with the cover 5 and preferably a second component of the sound signal representative of at least one return of the cover 5 in a closed position after the insertion of the corresponding object (called "inserted object") into the container 2 ( ie after the interaction of the object inserted with said lid 5) and / or a third component of the sound signal representative of at least one shock of the object inserted in the container 2 at the end of its fall into said container 2 Said characterization module 3 is advantageously configured so as to study the first and preferably also the second and / or third components of the sound signal to characterize said inserted object.

The cap 5 is a member configured to vary between a closed position and an open position. In the context of the present invention, the cap 5 is preferably configured so as to always be biased towards its closed position, in particular by a return member, the passage from the closed position to the open position being carried out when an object is intentionally inserted into the container 2 by a user who solicits the object to be inserted against the lid 5 so as to counteract the forces maintaining said lid 5 in its closed position. Thus, as soon as the inserted object has completely passed the cap 5, the latter naturally returns to its closed position. In its closed position, the seal 5 closes a hole formed in the container 2 and allowing the insertion of objects into said container 2.

• une deuxième composante représentative d'au moins un retour de l'opercule 5 dans une position de fermeture après une insertion de l'objet correspondant dans le contenant 2, et/ou
• une troisième composante représentative d'au moins un choc de l'objet inséré dans le contenant 2 à la fin de la chute dudit objet inséré dans le contenant 2.

It is understood from what has been said above that, in the context of the method, the study step E2-1 comprises the use of at least a first component representative of at least one interaction of the object. to be inserted with the cover 5 of the recovery device mounted on the container, and preferably also one of the following components of the sound signal chosen from:

• a second component representative of at least one return of the cover 5 in a closed position after insertion of the corresponding object into the container 2, and / or
• a third component representative of at least one shock of the object inserted in the container 2 at the end of the fall of said object inserted in the container 2.

In general, the first component may include a sound associated with the point contact of the object with the cover 5 before opening of the cover 5 and / or a sound associated with the friction of the object with the cover 5 during the insertion of the object in the container 2.

The second component may incorporate a sound associated with the return of the cap 5 in its closed position.

The third component may preferentially incorporate a sound relative to the impact of the inserted object when it reaches the bottom of the container 2 formed by a part of the container 2 itself or by objects already present in the container 2.

The figure 3 illustrates a sound signal from the acoustic system 4 during an insertion phase of an object in the container 2. This signal comprises the first component C1, the second component C2 and the third component C3. These first, second and third components are here successive. Thanks to the study of the sound signal, it is possible to deduce from the first component C1 a first time instant T _C representative of the first contact of the object with the cover 5, of the second component C2 a second time instant T _R representative of the return of the cap 5 in a closed position. Moreover, it is also possible to deduce from the third component C3: T _Ch representing the time instant of the end of the fall including the impact at the end of the fall of the inserted object. To the figure 3 , ΔT _Ch represents a time duration of falling of the object inserted in the container 2 taking into account the duration between T _C and T _Ch .

More particularly, it is considered that the sound signal S (t) may be of type: $$\mathrm{S}\left(\mathrm{t}\right)=\mathrm{VS}1\left(\mathrm{t}+{\mathrm{T}}_{\mathrm{VS}}\right)+\mathrm{VS}2\left(\mathrm{t}+{\mathrm{T}}_{\mathrm{R}}\right)+\mathrm{VS}3\left(\mathrm{t}+{\mathrm{T}}_{\mathrm{ch}}\right)+\mathrm{B}\left(\mathrm{t}\right)$$

With C1 an impulsive type contact signal starting at time T _C , C2 an impulsive type of return signal starting at time T _R , and C3 a signal after the fall and broadband type starting at T _Ch and B represents the additive environmental noise. This noise is of low amplitude inside the container which is well soundproofed.

For example, for a glass bottle, the duration of the first and second components C1, C2 varies between 5ms and 10ms. For the third component C3, in these conditions it can last between 25ms and 120ms, this spread in time is due to possible rebounds of the inserted object.

At first, we are interested in the first component C1. This first component C1 is advantageously used to check whether the material of the inserted object (here it is preferable to determine the majority material in the object under consideration) is compatible with the container 2 (that is to say in glass if the container 2 is intended for the recovery of the glass, or in paper if the container 2 is intended for the recovery of paper). Therefore, the characterization module 3 comprises a determination element of a material of the inserted object configured to exploit the sound signal, according to the invention at least the first component C1 of the sound signal.

Preferably, the container 2 is either a glass container or a paper container. In this sense, the material determining element can, from the sound signal, determine whether the material of the inserted object is, if any, glass or paper, or if it is a unauthorized material in the container 2.

More particularly, the cap 5 is present, and the characterization module 3 makes it possible to determine the material of the inserted object (via the material determination module) by using the first component, which then includes a sound associated with the contact of the object. opercule 5 with the inserted object and / or a sound associated with the friction of the object with the cover 5 before the fall of said object in the container 2.

It is understood here that, preferably, the skilled person will choose the cap 5 according to its ability to discriminate more easily objects of material that it is desired to be stored in the container 2. In this sense, the seal 5 referred to above may contain plexiglass or metal. The lid 5 will preferably be in plexiglass for a container 2 for storing plastic objects, and metal for a container 2 for storing metal objects (cans, tin can, etc.). More generally, the container 2 is intended to receive objects of a particular material, and the choice of the material constituting the cap 5 will be made according to the particular material.

For example, in the case of glass the lid will preferably be metal, and in the case of paper the seal will preferably also metal.

le rapport d'énergie (entre 0 et 1) avec E_CB l'énergie basse fréquence de ladite première composante et E_CT l'énergie totale de ladite première composante.In order to verify / determine whether the material is the correct one, it is possible to extract from the first component C1 data that makes it possible to calculate the following ratio: $${\mathrm{R}}_{\mathrm{CBT}}={\mathrm{E}}_{\mathrm{CB}}/{\mathrm{E}}_{\mathrm{CT}}$$

the energy ratio (between 0 and 1) with E _CB the low frequency energy of said first component and E _CT the total energy of said first component.

The figure 4 illustrates the behavior of the first component C1 in the context where the cover 5 is metal for objects in different materials (glass, stone, aluminum, wood, plastic). The figure 4 comprises a curve of the ratio R _CBT as a function of the contact cutoff frequency F _CC of the first component for each of the listed materials. In particular, curve 1 is associated with glass objects (small and large bottles, bottle of wine, bottle of cider, bottle of champagne, jars), curve 2 is associated with stone objects (small and large pebble, pottery , pebbles, bricks, etc.), curve 3 is associated with wooden objects (dry wood, wet wood, etc.), curve 4 is associated with aluminum objects (bobbin, cans, etc.), and the curve 5 is associated with plastic objects.

It results from the analysis of figure 4 , that by taking a cutoff frequency F _DC between 1500Hz and 2000Hz and E _CB corresponding to the energy of the filtered signal passes low with F _CC , it is possible to establish a threshold of R _CBT equal for example to 0.65. Below this threshold, it can be said that the material of the object inserted in the container is glass, and above that the material of the inserted object is not compatible with the container 2. The figure 4 is obtained using a single low pass filter with less complexity.

In other words, the material determination element comprises an establishment element, at a predetermined cutoff frequency value, of a ratio of the low frequency energy of said first component C1 to a total energy of said first component C1. , and a ratio comparison element established with a predetermined ratio threshold value (in the glass example, if the established value of the ratio is below the predetermined ratio threshold value then the material of the inserted object is considered as being glass). In particular, it is understood that the predetermined value of cutoff frequency and the predetermined ratio threshold value are a function of a calibration of the device, in particular depending on the type of cap and the positioning of the acoustic system, the skilled person will therefore be able to choose these thresholds using the teachings of the present description. It is understood here that the low frequency energy of the ratio is that obtained by filtering at the predetermined cutoff frequency.

The same principle works to discriminate paper. In particular, in the context of paper, the energy is concentrated at 80% in the frequency band below 1 kHZ. In this sense, it is possible to know that the inserted object is made of paper if the ratio R _CBT established is greater than 0.8 for a frequency F _DC between 500Hz and 1000Hz. In this particular example the lid is metal. Here it is used a low pass filter.

Thus, in the context of the method, the study step E2-1 comprises a step of a determination of the material of the inserted object comprising: a determination, starting from the first component of the sound signal and at a cutoff frequency predetermined, a ratio of the low frequency energy of said first component to a total energy of said first component, and a comparison of the determined ratio with a predetermined ratio threshold value
Let's move on to the second C2 component. The latter can be used to determine the size of the inserted object.

To determine the size of the inserted object, the characterization module 3 can calculate a time duration between the beginning of the first component C1 (instant T _C ) and the beginning of the second component C2 (instant T _R ).

Discrimination between a small inserted object and a large inserted object can be done by comparing, for each inserted object, the difference between T _C and T _R at a predetermined threshold of discrimination of the size of the object (below this threshold it is a small object and above this threshold it is a large object).

Determining the size of the inserted object may also be useful in the context of a monitoring of the filling state of the container 2, or in the context of a payment of the user of the container 2 according to the objects inserted.

Thus, the characterization module 3 may include an element for estimating the size of the inserted object using said sound signal. Preferably, this size estimation element uses the first component C1 of the sound signal and the second component C2 of the sound signal by determining from these the interaction time of the object inserted with the cover 5 during in particular, this element for estimating the size of the object inserted, and therefore by extension a step of estimating the size of the inserted object of the process (belonging in particular to the study step E2-1), can process said sound signal so as to determine, from the first and second components, the duration between T _C (a moment of first contact of the object inserted with the cover 5) and T _R (a closing instant of the operculum 5 after insertion of the object in the container 2). It is based on the fact that the longer the duration between T _C and T _R will be shorter the object will be small (and therefore light depending on the identified material). In other words, a step of comparing the determined duration with a duration threshold can then suffice to estimate the size of the inserted object. The analysis of different drops of glass bottles has shown that the average duration between T _C and T _R for a large bottle is 40% higher than for a small bottle.

According to an improvement, the determined size can be used to estimate the weight of an inserted object when we know the material constituting it.

By calibration, it is then possible to generate a model on the basis of the time parameters between T _C and T _R. This model will be used to estimate the actual size and / or effective weight of an inserted object. This established, the model can easily be realized by the skilled person without it being necessary to describe here a detailed implementation.

Regarding the third component, the latter can be used in different ways. The sound associated with the shock of the object inserted in the container 2 at the end of its fall in the latter allows, for example, to validate that the object has finally been inserted into the container 2, thus validating the characterization. According to another example, the third component C3 makes it possible to determine a filling state of the container 2. According to yet another example, the third component C3 can be used to determine the weight of the inserted object, especially when said inserted object is paper, and therefore the container 2 is a paper storage container 2.

According to an example of determining the weight of the inserted object, the figure 5 gives the amplitude of the third component C3 as a function of time for successively a pile of newspapers, a pile of magazines, and a directory presenting different weights but the same volume / same size. It is then understood that it is possible to estimate a weight of the inserted object by analyzing the amplitude of the third component C3 and / or the duration of the third component C3. The size or volume of the object inserted can be estimated for example from the study of the first and second components C1, C2 as described above.

Preferably, for the glass, the size will first be estimated in order to estimate a weight, and for the paper the energy of the signal will be estimated with a calibration adapted to estimate a weight.

The analysis of the energy of the third component C3 shows that 90% of this energy for a pile of discarded paper is concentrated in the frequency band below 2kHz. The energy of the third component C3 can therefore be used to estimate the weight of a pile of discarded paper as illustrated by figure 6 .

In addition to the information described above and extracted from the third component C3, the latter may, alternatively or in combination, be divided into weighted frames by a weighting function. These slippery fields overlap to better smooth estimates. This decomposition in frames is adapted to the real-time processing, and makes it possible to follow the variability and non-stationarity of the sound signal over time. In this sense, it is possible to establish a curve representing the evolution of the energy of the third component C3 as a function of the number of the associated frame as illustrated in FIG. figure 6 . On this figure 6 third C3 components are shown for successive insertion of a pile of newspapers, a pile of magazines and a directory of the same volume but of different weights. Thus, by using a smoothing in frames of the third component C3, one realizes that the study of the associated energy also makes it possible to estimate a weight.

• F_di les fréquences dominantes de la troisième composante,
• E_Chi l'énergie du signal dans la bande i de la troisième composante (i=1 à 8), normalisé (entre 0 et 1) par rapport à l'énergie totale E_ChT.
• R_ChBT = E_ChB/E_ChT le rapport d'énergie avec E_ChB l'énergie basse fréquence de la troisième composante,
• N_re le nombre de rebonds de l'objet à la fin de sa chute lors de son interaction avec le fond du contenant 2, ou avec des objets déjà contenus dans le contenant 2. Lors du rebond, l'amplitude du signal est faible, et un seuillage sur l'énergie permet d'estimer N_re

It is also possible to extract the following data from the third component C3:

• F _di the dominant frequencies of the third component,
• E _Chi the signal energy in the band i of the third component (i = 1 to 8), normalized (between 0 and 1) with respect to the total energy E _ChT .
• R _ChBT = E _ChB / E _ChT the energy ratio with E _ChB the low frequency energy of the third component,
• N _re the number of rebounds of the object at the end of its fall during its interaction with the bottom of the container 2, or with objects already contained in the container 2. During the rebound, the amplitude of the signal is low, and a threshold on energy makes it possible to estimate N _re

In this sense, a step of characterizing a material of the inserted object can comprise, starting from the third component C3 of the sound signal, a determination of dominant frequencies and / or of a multiband energy and / or a energy ratio.

These extracted data being linked to the third component C3, they can also be used to determine the filling of the container 2, to check if the insertion has been made, or to determine the weight of the inserted object.

Moreover, from the sound signal of a glass object, it will be possible to classify the inserted object in a low level (the object breaks often without rebound in the container, the recorded signal is compact), medium level (the object often bounces several times in the container and breaks), high level (the object often bounces several times without breaking it in the container). The skilled person is fully able to analyze the sound signal to classify the fall of the object in the container 2 in one of these three levels.

It is understood from what has been said above that, preferably, the characterization of an inserted object will not be achieved by simple comparison of the sound signal studied with pre-recorded standard sound signals, but rather by extraction, from the sound signal. , empirical data (extracted data and / or levels described above) for characterizing said object for example by comparing the empirical data extracted from the sound signal to data from a calibration.

It has been mentioned above that the characterization of an inserted object could consist of determining the material constituting the inserted object and / or its size and / or its weight. In other words, the step of characterizing the inserted object and / or the characterization module 3 output: the material constituting the inserted object and / or the size of the inserted object and / or the weight of the inserted object. inserted object.

Examples of plastic and metal objects have been given above. Of course, the invention is not limited to this type of objects. The acoustics may very well also be used to estimate the volume of paper thrown into the container and / or to evaluate the filling rate of the container.

Preferentially, to promote the citizen approach of recycling objects, the user will be encouraged to perform a sort of objects by preferably placing in the same container 2 that objects of the same material. In this case, the characterization module 3, and by extension the characterization step, is configured to check whether the object inserted is in the right material, if the verification is positive then the user is paid. Preferably, the size and / or weight of the inserted object is used to determine the retribution of the user. If the check is negative then a fraud is detected and the user is not paid (in this case the user is warned that the inserted object is not intended for the container).

In other words, the recovery device 1 preferably comprises a payment module 6 ( figure 1 ), in particular financial, of a user having inserted the object into the container 2, said compensation module 6 being connected to the characterization module 3 so as to use at least one data relating to the characterization of said inserted object to evaluate the retribution. To summarize, the evaluation of the compensation implemented by the compensation module 6 preferably takes into account the material of the inserted object, the size and / or the weight of the material of the inserted object as well as a rate of representative of the value of the material constituting the object. When one places oneself within the framework of the process, the latter preferentially comprises a step of valuation E3 of the inserted object in which at least one datum resulting from the characterization of the inserted object is used to credit an account, in particular a user account belonging to the person who inserted the object within the container 2.

According to a particular embodiment, the method comprises, before the step of inserting the object into the container 2, a step of identifying the the user (by card or by smart phone). A contactless contact card can be made by RFID. For the smartphone, the use of a dedicated application and a wireless communication, for example by WIFI or Bluetooth, can be used. One or more steps of inserting an object can then be performed. When the user has finished, he implements a step of validation of the inserted objects from where it results preferentially a step of evaluation of a financial reward according to the inserted objects, and a stage of contribution of credit on a account, for example present on the card or on the smartphone. During the insertion steps, if an object not compatible with the container 2 is identified, a signal light, or a buzzer, can be used to signify to the user that it is fraudulent and that no payment will be made for the object concerned.

The payment can be credited to a personal card of the user that the latter has taken care to present to the payment module 6 before the insertion of the object in the container 2. Alternatively, the credit can be made to an account the user that the latter can, for example, consult from his smart mobile phone, in this case the compensation module 6 can use a connection, preferably wireless, to go credit the appropriate user account on a server.

In a particular example, the user will be rewarded for an insertion phase of a plurality of objects in the form of glass bottles in a container 2 forming a collection point for glass bottles. The compensation according to the weight for a given user can be done in the following manner by characterizing each of the objects inserted: $${\mathrm{P}}_{\mathrm{User}}={\mathrm{P}}_{\mathrm{BOY\; WUT}}*{\mathrm{NOT}}_{\mathrm{BOY\; WUT}}+{\mathrm{P}}_{\mathrm{P}}*{\mathrm{NOT}}_{\mathrm{P}}$$

Where _User represents a total weight thrown by a user when he came to the collection point,
P _G represents a fixed datum representing an average weight of a bottle considered as a large bottle,
P _P represents a fixed data representing an average weight of a bottle considered as a small bottle,
N _G and N _P represent respectively the number of large bottles and the number of small bottles actually discarded by the user. The discrimination between a small bottle and a large bottle is made by comparing, for each bottle inserted, the difference between T _C and T _R at a predetermined threshold of discrimination of the size of the bottle (below this threshold it is a small bottle and above this threshold it is a large bottle).

In another particular example, the user will be rewarded for a phase of insertion of a plurality of objects, being in the form of paper pile, in a container 2 forming a paper collection point. The compensation according to the weight for a given user can be done in the following way by studying each of the inserted objects:
For each jet i, a pile weight P _{Tas_i is determined} and the estimate of the weight of paper waste sent by the user (P _Usager ), finally P _Usager is equal to the sum of P _{Tas_i} for i ranging from 1 to N _Pile the number of heaps inserted.

It can happen, when a user brings objects into a recovery center (also called voluntary point of supply), the container 2 he wants to use is full. This is due to problems of management of the filling of the container 2. In this sense, it is preferable to equip the device 1 of recovery of a determination module 7 of the filling of the container 2, in particular configured to use at least data from the acoustic system 4 and / or the characterization module 3. When the determination element 7 of the filling of the container 2 detects that the filling of the container 2 exceeds a predetermined threshold, it can transmit an alert to the device manager. recovery 1 so that the latter quickly plans the emptying of said associated container 2.

For example, the filling level of the container 2 is estimated from the instant of the contact T _C of the object with the lid 5 and the instant T _Ch . The duration between these two instants will depend on: the discarded object, the size of the discarded object (small, large), the speed of the jet of the object given by the user, and the level of filling before the jet (cumulative volume).

To calculate a fill level of the container 2, it is then possible to estimate on the objects discarded by the user the duration between the instants T _C and T _Ch averaged over the number of small objects inserted in the container 2 and on the number of large objects inserted in the container 2, as well as a minimum duration between these two instants.

The figure 7 gives an example of the evolution of the average falling time of a glass bottle considered by the characterization step as a small bottle depending on the level of filling of the container 2. It is therefore clear that the duration of a fall of 'an object, or the duration average drop of objects of the same size to limit errors, used to estimate the filling level of the container 2.

According to a generalization of the method, the latter advantageously comprises a step of estimating the filling of the container 2 using at least one acoustic data, in particular from the acoustic system 4. In particular, the step of estimating the filling comprises at least one one of the following steps: a determination, from said at least one acoustic data, of a number of average rebounds made in the container by the inserted objects; a determination, based on said at least one acoustic data, of an average duration of falling objects, including the same size, inserted in the container 2; a determination, from said at least one acoustic data item, of the number of objects inserted in the container 2.

The container 2 described above may be overhead or buried.

According to a particular embodiment, to avoid any fraud, the outer envelope may preferably comprise a cavity at the bottom of which the cover 5 is placed. This cavity has a depth that prevents the user from interacting with his hand directly with the lid (for example to make believe that the volume of the object inserted in the container 2 is much greater than the actual volume of the object inserted in the container. container 2 to increase compensation).

According to one embodiment, the recovery device 1 can be electrically powered, for example by battery or with a solar panel connected to a management system and battery storage to allow in particular to the characterization module 3 to implement its tasks. In order to save energy, the recovery device 1 may include a sleep state and an active state. The active state consumes more power than the standby state. The transition from the waking state to the active state can be triggered by contact of an object with the cover 5, and the transition from the active state to the waking state can be implemented for example if the cap 5 has remained in its closed position for a duration greater than a threshold duration of standby.

The recovery device 1 may also include a human machine interface allowing feedback to the user by informing him of the material and / or size and / or volume of the object inserted into the container 2.

Claims (12)

1. Device (1) for collecting objects, in particular rubbish, comprising a container (2) into which said objects are intended to be inserted, a module (3) for characterizing an object inserted into the container (2), and an acoustic system (4) that is configured so as to record an audio signal representative of the interaction of the inserted object with the collecting device (1), said acoustic system (4) being connected to said characterizing module (3) so that the audio signal may participate in the characterization of said inserted object, said device including a lid (5) that is mounted on the container (2) and that is opened to allow the object to be inserted to be introduced into the container (2), characterized in that said acoustic system (4) is configured so as to record a first component (C1) of the audio signal, which component is representative of at least one interaction of the object to be inserted with the lid (5), and in that the characterizing module (3) includes an element for determining a material of the inserted object, configured so as to exploit the audio signal, said material-determining element including an element for establishing, at a preset cut-off frequency value, a ratio of the low-frequency energy of said first component (C1) to a total energy of said first component (C1), and an element for comparing the established ratio with a preset threshold ratio value.
2. Device according to Claim 1, characterized in that said acoustic system (4) is configured so as to record a second component (C2) of the audio signal, which component is representative of at least one return of the lid to a closed position after the corresponding object has been inserted into the container (2) and/or a third component (C3) of the audio signal, which component is representative of at least one impact of the inserted object in the container (2) at the end of its fall in said container (2).
3. Device according to Claim 1 or 2, characterized in that said characterizing module (3) is configured so as to study the first component (C1) and the second and/or third component (C2, C3) of the audio signal in order to characterize said inserted object.
4. Device according to one of Claims 1 to 3, characterized in that the acoustic system (4) includes a microphone, in particular a digital MEMS or Electret microphone.
5. Device according to any one of the preceding claims, characterized in that the characterizing module (3) includes an element for estimating the size and/or weight of the inserted object using said audio signal.
6. Device according to any one of the preceding claims, characterized in that it includes a module (7) for determining the fullness of the container (2), in particular configured so as to use at least one datum generated by the acoustic system (4) and/or the characterizing module (3).
7. Method for collecting objects, in particular rubbish, in a collecting device equipped with a container (2), said method comprising a step (E1) of inserting an object into the container (2) and a step (E2) of characterizing the inserted object including a step (E2-1) of studying an audio signal resulting from the interaction of the inserted object with said collecting device, characterized in that the studying step (E2-1) includes using a first component (C1) that is representative of at least one interaction of the object to be inserted with a lid (5) of the collecting device, which lid is mounted on the container (2), and in that the studying step (E2-1) includes a step of determining the material of the inserted object, comprising:

   ∘ determining, from the first component (C1) of the audio signal and at a preset cut-off frequency, a ratio of the low-frequency energy of said first component (C1) to a total energy of said first component; and

   ∘ comparing the determined ratio with a preset threshold ratio value.
8. Method according to the preceding claim, characterized in that the studying step (E2-1) includes using at least one of the following components of the audio signal chosen from:

   - a second component (C2) that is representative of at least one return of the lid (5) to a closed position (2) after the corresponding object has been inserted into the container (2); and/or

   - a third component (C3) that is representative of at least one impact of the inserted object in the container (2) at the end of a fall of said inserted object in the container (2).
9. Method according to the preceding claim, characterized in that the studying step (E2-1) includes at least one of the following steps:

   - estimating the size of the inserted object, this step comprising:

   ∘ determining, from the first and second components, a duration between a time of first contact of the inserted object with the lid (5) and a time of closure of the lid (5) after the object has been inserted into the container (2); and

   ∘ comparing the determined duration with at least one duration threshold;

   - characterizing the material of the inserted object, this step comprising, on the basis of the third component (C3) of the audio signal, determining dominant frequencies and/or a multiband energy and/or an energy ratio; and

   - determining the weight of the inserted object from the energy of the third component (C3) of the audio signal.
10. Method according to one of Claims 7 to 9, characterized in that the step (E2) of characterizing the inserted object delivers as output: the material from which the inserted object is made and/or the size of the inserted object and/or the weight of the inserted object.
11. Method according to one of Claims 7 to 10, characterized in that it includes a step of estimating the fullness of the container (2) using at least one acoustic datum.
12. Method according to the preceding claim, characterized in that the step of estimating fullness includes at least one of the following steps:

    - determining, from said at least one acoustic datum, an average number of times the inserted objects bounce in the container (2);

    - determining, from said at least one acoustic datum, an average fall time of the inserted objects in the container (2); and

    - determining, from said at least one acoustic datum, the number of objects inserted into the container (2).

Images