Sink test

Abstract

The claimed invention provides a method of assessing liquid absorption of a sample and an associated apparatus. The method comprises: identifying, using at least one of one or more measurement devices, a location of a surface of a liquid inside a container; introducing the sample to the liquid; identifying, using at least one of the one or more measurement devices, based on the location of surface of the liquid, a contacting of the surface of the liquid by the sample and a submersion of the sample in the liquid; recording a first time corresponding to the contacting and a second time corresponding to the submersion; and calculating a sinking time based on the first time and the second time.

Description

[0001] SINK TEST

[0002] FIELD OF THE INVENTION

[0003] The present invention relates to the assessment of liquid absorption by a sample. Typically, the sample is placed in a container of water and the sinking of the sample is monitored.

[0004] BACKGROUND

[0005] The manufacture of goods designed to absorb liquid is common across many sectors, from homewares to industrial growth substrates. It can be important to ensure consistency between different samples of the manufactured goods in order to produce goods of an acceptable quality. For instance, the quantity of liquid that can be absorbed in a particular time may be important to the end user.

[0006] As a measure of quality of an absorbent product, one may measure the absorbency. The absorbency of a material is its ability to take in and retain liquids within its structure. A high absorbency means that liquid is taken in by the material, and a large amount of liquid is retained inside.

[0007] It is known that the absorbency of a sample may be assessed using a variety of different methods. One such method is a “sink test”, in which the time taken for a sample to be wetted by a liquid is measured. The sample is placed on the surface of the liquid, floats, and begins to absorb the liquid. Once the sample is sufficiently wetted, i.e. it has taken on enough of the liquid such that its density is greater than that of the liquid, it will begin to sink into the liquid. This test requires a sample that has a lesser density than the liquid when not wetted, and a greater density than the liquid when wetted.

[0008] Such a test can be especially relevant for materials with an aperiodic or amorphous structure, such as mineral wool. If the structure of the material can vary, measurement of the absorbency can ensure consistency across various different samples. For more regular materials, this deviance may be lesser or less important. Therefore, commonly, in an effort to ensure a consistent quality of product, an operator will monitor the manufacturing process of products such as mineral wool and take samples from the output to perform sink tests. This has several shortcomings, however.

[0009] Firstly, it requires manual input from a human operator, who must select a sample and perform the test.

[0010] Secondly, performing the test requires close monitoring of the sample in the liquid. This includes starting and stopping a clock at the precise moments of the sample entering the liquid and full submersion, which is a difficult task. The accuracy of the test and ability to conduct consistent repeats is therefore reduced. To increase the accuracy and consistency of repeats of the test can require the involvement of a second operator, or more time on behalf of the first operator, which is also undesirable.

[0011] Thirdly, the time taken to perform the test limits how many samples can be tested, which can lead to production errors being detected only after a large amount of inferior product is already manufactured.

[0012] There is therefore a need for more accurate and consistent testing of absorbency whilst simultaneously reducing the amount of operator input involved.

[0013] SUMMARY OF INVENTION

[0014] According to a first aspect, there is provided a method of assessing liquid absorption of a sample, the method comprising: identifying using at least one of one or more measurement devices, a location of a surface of a liquid inside a container; introducing the sample to the liquid; identifying, using at least one of the one or more measurement devices, based on the location of the surface of the liquid, a contacting of the surface of the liquid and a submersion of the sample in the liquid; recording a first time corresponding to the contact and a second time corresponding to the submersion; and calculating a sinking time based on the first time and the second time. The method allows for more accurate testing of the absorbency of a sample over the state of the art, whilst reducing the amount of operator input required. The sample may comprise mineral wool, also known as rock wool, formed from fiberised rock.

[0015] The location of the surface of the liquid may be identified using a first measurement device of the one or more measurement devices. The location of the contacting of the surface of the liquid by the sample, and the submersion of the sample in the liquid, may be identified using a second measurement device of the one or more measurement devices. The second measurement device may be the same device as the first measurement device, and is referred to herein as the “second” measurement device for disambiguation only. There may be a single measurement device which locates both the surface of the liquid and the sample.

[0016] The one or more measurement devices may be a single measurement device. This improves the efficiency of the method and reduces the space required for the operation of the method.

[0017] The one or more measurement devices may also comprise a computer vision system. Such a system may automate the process of measuring the absorbency, freeing up the operator to perform other tasks. The specifics of the computer vision techniques used are discussed below. This arrangement also allows for greater consistency to be applied when testing multiple samples since there is greater precision possible. This is due to the computer vision system allowing identification of surface contact and submersion with greater consistency between tests than a human operator is capable of.

[0018] The computer vision system may use classical computer vision algorithms, such as edge detection algorithms. In other examples, binary-based methods or contour-based methods of corner detection could be used. In still further examples, a Moravec, Harris or SUSAN corner detection algorithm could be used.

[0019] However, in typical examples, a machine learning algorithm or neural network is used, trained on a corpus of annotated training data including images of samples. The images are annotated to denote the region of the image occupied by the sample. In this way, the machine learning algorithm learns to detect the location and bounds of the sample in the images. In some examples, just the corners of the sample are tracked and the bounds of the sample determined from this. In other examples, the full boundary of the sample is found.

[0020] Alternatively, the one or more measurement devices may comprise a rangefinding sensor, for instance, a laser and corresponding sensor. The second measurement device may also comprise a range-finding sensor. In one example, a LIDAR sensor may be used. LIDAR is a method for determining ranges by targeting an object or a surface with a laser and measuring the time for the reflected light to return to the receiver.

[0021] When a range-finding sensor is used as the first measurement device, it may be arranged such that it measures the location of the surface of the liquid. In some examples, such as when using a LIDAR sensor, this will involve the transmission of light at the surface of the liquid, which is subsequently reflected to a sensor. Alternatively, if a level indicator is used, as per the discussion below, a rangefinding sensor may also be used to determine the location of a level indicator in the same way.

[0022] Using a range-finding sensor may allow for increased computational efficiency when compared to a computer vision system. Further, a range-finding sensor may be used when the sample or the liquid is not clearly visible, for instance, in the case of an opaque container or opaque liquid.

[0023] When a range-finding sensor is used as the second measurement device, it may also be used to determine the location of the sample via the transmission of light at the sample, which is subsequently reflected to a sensor.

[0024] In some examples, the one or more measurement devices comprise a float level. A float level measures the level of a liquid using the principle of buoyancy, where a floating element moves inside the level. Such a level enables the continuous detection of the surface of the liquid, independent of physical and chemical changes of the media such as foaming, conductivity, dielectric, pressure, vacuum, temperature, vapours, condensation, bubble formation, boiling effects and density change. The location of the floating element in the level may be detected using magnetic means, for instance, by disposing magnets or magnetic material on the floating element, or by mechanical means, such as cables, tapes, pulleys and gears.

[0025] The use of a float level as a first measurement device may simplify the method and increase the accuracy by providing a direct measurement of the location of the surface of the liquid. Determining the location of the surface of the liquid may be challenging due to reflection, refraction and the motion of the surface itself, which do not complicate measurement using a float level.

[0026] In some examples, the one or more measurement devices comprise a hydrostatic device, such as: a displacer, a measurement device which functions via Archimedes’ principle; a bubbler, which senses fluid depth and determines the location of the surface by measuring hydrostatic pressure at a level underneath the liquid surface; or, a differential pressure sensor, operating by detecting the total pressure difference between the fluid at the bottom of the container and the ambient pressure.

[0027] In some examples, the one or more measurement devices comprise a load cell or strain gauge device. A load cells may be advantageous because of their noncontact nature, which avoids disturbing the liquid.

[0028] Both hydrostatic devices and load cells may be advantageous for at least the same reasons as a float level.

[0029] In some examples, the one or more measurement devices comprise an ultrasonic level transmitter. The second measurement device may also comprise an ultrasonic sensor. Such ultrasonic sensors measure the distance between a transducer and a target, such as the surface of the liquid, and subsequently calculate the location of the surface using the time required for an ultrasound pulse to travel from a transducer to the target and reflect to a sensor. Identifying the location of the surface of a liquid may comprise directly measuring the location of the surface, using at least one of the one or more measurement devices. Alternatively, or in addition, the location of the surface may be identified via measurement in advance, or by using a preset value for the location of the surface, or by determining the location of the surface from the location of a separate component, such as from the location of one of the one or more measurement devices. For instance, in some examples, a measurement device of the one or more measurement devices may be floating on the surface of the liquid or located at a known location relative to the surface of the liquid. In such an example, the location of the surface of the liquid is derived rather than directly measured.

[0030] Similarly, the identifying a contacting of the surface of the liquid may comprise detecting, using at least one of the one or more measurement devices, the sample touching the liquid. Alternatively, or in addition, the identifying a contacting of the surface of the liquid may be determined from a measurement of a different quantity. For instance, in some examples, the velocity of the sample as it is introduced may be measured by one or more measurement devices. From this, based on a location of the surface of the liquid, the contacting may be identified. In other examples, the identifying a contacting of the surface of the liquid may be performed by identifying a location of the sample at a known time, and deriving or approximating the time at which the sample is predicted to contact the surface.

[0031] The above concepts apply similarly to the identification of the submersion of the sample, which may be directly measured, or may be derived or inferred from measurement of different quantities. For instance, the sample may be detected at a depth some distance below the surface of the liquid, and, based upon the speed the sample descends through the liquid, the submersion may be identified as having occurred at an earlier time.

[0032] In some examples, the identifying a contacting of the surface of the liquid comprises identifying the sample being sufficiently close to the surface of the liquid. In such examples, the additional travel time for the sample to reach the surface of the liquid may be neglected or compensated when identifying the contacting.

[0033] Identifying the location of the surface of the liquid may further comprise identifying the location of a level indicator. The level indicator may be identified using a computer vision system, which allows for the accurate assessment of the amount of liquid in the container and the location of the surface. Alternatively, the location of the level indicator may be detected by any suitable second measurement device, such as those set out above. The container used is preferably (substantially) transparent, permitting a side-on view of the liquid and the level indicator. Alternatively or additionally, the container may comprise a window or transparent panel for viewing the liquid and / or level indicator through. Optionally, the level indicator is at least partially immersed in the liquid.

[0034] In order to detect the location of the level indicator, any one, or multiple, of the aforementioned measurement devices could be used. If a machine learning method is used to locate the level indicator, a corpus of corresponding training data may be used.

[0035] Multiple measurement devices may be used in combination to determine the location of the surface of the liquid, or to determine the location of the sample, or a contacting of the surface of the liquid by the sample and a submersion of the sample in the liquid. For instance, both a computer vision system and a rangefinding sensor may be used to determine the location of the surface of the liquid, or the location of the level indicator. This may improve the accuracy of the method.

[0036] When certain liquids, such as water, are used, it can be difficult to detect the location of the surface due to the complex appearance of the surface. For instance, some computer vision systems may fail to detect the surface. This is not only because the water is transparent, but also reflection and refraction effects play a large role. Similar factors would also apply if other liquids are used. The presence of the level indicator allows the computer vision system to accurately determine the location of the surface. In typical examples, the level indicator takes the role of a floating object on the surface of the liquid, with a shape and coloration known. In other examples, the level indicator may be a colourant or dye added to the liquid or may take the form of a liquid-sensitive marking placed in or on the container.

[0037] Once the sample is introduced to the liquid, the sample can be located and tracked by the second measurement device. This allows the claimed invention to detect if / when the sample has dropped below the surface of the liquid and is therefore fully submerged. The sample may take any shape, but preferably the sample is a cuboid, which comprises sharp corners, such as corners that have the largest angles between edges meeting at a corner of no more than 150 degrees (°), no more than 135° no more than 120°, no more than 100° or no more than 90° that bound the sample.

[0038] Detection of the second time (the time of submersion) is typically done by determining when all of the sample has passed below the location of the level indicator, and therefore passed below the surface of the liquid. In some examples, the second time will instead be a time when the sample passes a known amount below the surface of the liquid, or when the sample reaches the bottom of the container. It will be appreciated that any such choice leads to a sinking time that is simply proportional to the sinking time as defined above.

[0039] A processor may be connected to memory which stores the sinking time, which may also store previous sinking times or examples. Alternatively or in addition, the processor may transmit the sinking time via a wired or wireless connection to a computer, a server or a device for monitoring the measurement of the sinking time.

[0040] Optionally, the output of the first or second measurement devices, such as image(s) or video analysed by the computer vision system, may be shown to an operator. They may be shown to an operator on a computer display. The image(s) or video may be annotated to show the detected location of the level indicator and sample and calculated position of the liquid surface. They may be further annotated with additional information, such as the certainty of the computer vision system that objects identified are indeed the level indicator or sample.

[0041] In some cases, the surface of the liquid may not be stable. This could be due to a variety of regions, such as vibration from machinery, or due to disturbance from a previous sample measurement. An unstable surface might lead to ripples or waves in the liquid, which could cause the sample to become wetted faster or slower than expected, or otherwise cause the sample to be difficult to track. Therefore, in some examples, the method further comprises, such as, before the step of introducing the sample to the liquid, detecting that the location of the surface of the liquid changes less than a stabilising threshold over a stabilising time. This allows the claimed invention to measure if the surface of the liquid is stable.

[0042] In some examples, this may comprise detecting that the location of the level indicator changes less than a stabilising threshold over a stabilising time.

[0043] The stabilising threshold may be 10 cm (centimetres), 5cm, 2cm, or 1 cm. The stabilising time may be 1 minute, 30 seconds, 10 seconds, or 1 second.

[0044] As reducing the amount of manual input required makes the process more efficient and allows for a greater throughput of samples being tested, in some examples, the sample is introduced to the liquid automatically.

[0045] Optionally, before the sample is introduced, the sample is held by a gripper, wherein introducing the sample to the liquid comprises the gripper releasing the sample allowing the sample to fall into the container. The gripper may release the sample from any height, but the sample may start close to the surface of the liquid. This minimizes disturbance to the surface when the sample is introduced.

[0046] The gripper may comprise two (substantially) parallel surfaces, (mechanically) pressed against the sample in use. This provides an ability to hold the sample. These surfaces may be slanted, shaped or sloped at the bottom (i.e. at a base or lower end) to provide a platform for the sample to rest upon. Alternatively, the gripper may comprise a plurality of fingers used to hold the sample. The gripper may comprise a magnet, if the sample is magnetic, or may comprise hooks, slats, bands, nets or moving surfaces.

[0047] As mentioned previously, it may be preferable that a calm, undisturbed surface of the liquid is preserved. This may be preferable not only because it increases the consistency of the absorbency measurement, but also because determining the location of the surface via the location of the level indicator is more accurate when the surface is calm. To this end, it may be advantageous if the level indicator is located inside a first chamber of the container, the sample being introduced in a second chamber separated from the first chamber. This allows any disturbance to the surface of the liquid upon introduction of the sample to be kept away from the level indictor, increasing the reliability of the surface level measurement. The first chamber and the second chamber should be connected such that the liquid can flow between the two freely, allowing the level to equilibrate. In this way, the level indicator can still be used to locate the surface of the liquid in the second chamber. The first chamber and the second chamber may be two distinct containers connected together to allow the liquid to flow, or may be divided or separated sections of the same container.

[0048] In order to further reduce the need for manual input from an operator, it may be desirable for the method to further comprise the automatic removal of the sample from the container after a sinking time has been calculated. In some examples, this removal may comprise emptying the container, including the liquid, and refilling the container before a subsequent measurement is made. However, typically, only the sample is removed from the container.

[0049] This automatic removal may comprise the use of a movable platform, employed to lift the sample out of the liquid. The movable platform may comprise perforations, holes, or tines to allow the liquid to flow through and / or around it. In some examples, the movable platform may be moved by electrical or mechanical means, via pneumatic systems, hydraulic systems, servos, stepper motors, or linear actuators. The sink test functions only if the sample is wettable and, by wetting by the liquid, has a density greater than the density of the liquid. If the sample is not wettable, is wetted but has a density lesser than that of the liquid, or for some other reason doesn’t sink, the test cannot be performed. In this situation, it would be advantageous notice that the sample is not sinking and terminate the test as a failure, as opposed to waiting for a submersion. Accordingly, in some examples, the method further comprises detecting, using the first measurement device, the second measurement device, or a third measurement device (of any appropriate type, such as those discussed above) if a height of the sample relative to the location of the level indicator is unchanging.

[0050] Detecting if the height of the sample relative to the location of the surface of the liquid is unchanging allows the method to decide if the sample is not sinking. This may comprise detecting that the height of the sample relative to the location of the level indicator is unchanging. The method may further comprise alerting an operator that the sample is not sinking. Alternatively or in addition, the method may report a sinking time of zero, or an arbitrarily large number.

[0051] The method may further comprise automatically removing an unsinking sample from the container.

[0052] The detection may take place over a series of measurements of the height of the sample. The sample height may be averaged and compared to the location of the level indicator, which may also be averaged. There may be a minimum threshold for sinking that must be met, such as 1 cm per 10 seconds (s). There may be a cut-off time at which the sample must have sunk below the location of the level indicator, such as 600 seconds, 60 seconds, or 10 seconds.

[0053] If the level indicator comprises a float or is designed to sit on the surface of the liquid, it may move around inside the container. This may be due to the introduction of samples disturbing the surface of the liquid. The level indicator moving may make detection of the level indicator using the one or more measurement devices more complicated or unreliable. To this end, the level indicator may be coupled to a support, such that the level indicator is constrained to move along an axis perpendicular to (or with a component perpendicular to) the surface of the liquid at rest. In this way, the level indicator is permitted to move up and down with the surface of the liquid, whilst being restrained from moving freely across the surface of the liquid.

[0054] In some examples, the support comprises walls of the container. In other examples, the support may be a rail or bar, which may be affixed to the container. In some examples, the coupling may comprise a physical tether or link, though the coupling may also or alternatively comprise a magnetic or electrostatic attraction between the level indicator and the support. Additionally or alternatively, the level indicator may be configured to fit partially or wholly within or around the support.

[0055] In order for the calculated sinking time to be comparable to other measurements and used to determine properties of the sample, it is important that other variables that affect the sinking time are controlled. Accordingly, in some examples, the method may further comprise the measurement of a temperature of the liquid. The temperature of the liquid may affect the time that the same sample takes to sink. The temperature may be recorded alongside the sinking time and factored in to comparisons of the sinking time.

[0056] Alternatively or additionally, the temperature of the liquid may be monitored so that it can be regulated or kept constant. Keeping a constant liquid temperature allows for simple comparison of the sinking times without requiring additional calculation to compensate for temperature changes. In some examples, the liquid will be kept at a temperature of 20 ± 1 degrees Celsius.

[0057] Further, in order to allow for easier comparison of the sinking time with known prior measurements, the method may further comprise the calculation of the sinking time further dependent on the temperature of the liquid. In this way, the sinking time is used as a metric for absorbency and not as a literal time. If the temperature of the liquid is higher than a known benchmark temperature and therefore the sinking time would be different than at the benchmark temperature, the sinking time may be adjusted to compensate. According to a second aspect, there is provided an apparatus for assessing liquid absorption of a sample, the apparatus comprising: a container holding a liquid; one or more measurement devices; and a processor configured to: identify, using the at least one of the one or more measurement devices a location of a surface of the liquid; detect, using at least one of the one or more measurement devices and based on the location of the surface of the liquid, contact of the sample with a surface of the liquid and a submersion of the sample in the liquid; record a first time corresponding to the contact and a second time corresponding to the submersion; and calculate a sinking time based on the first time and the second time. It will be appreciated that the above discussion of the first aspect can be applied to the second aspect and its features. Further the second aspect may have features allowing the method according to the first aspect to be implemented.

[0058] BRIEF DESCRIPTION OF DRAWINGS

[0059] An example process and an example apparatus are described in detail herein with reference to the accompanying figures, in which:

[0060] Figure 1 shows a schematic of an example apparatus;

[0061] Figure 2 shows a schematic of the steps of an example method;

[0062] Figures 3a and 3b show an example implementation before the sample has been introduced to the liquid;

[0063] Figures 4a and 4b show an example implementation a short time after the sample has been introduced to the liquid;

[0064] Figures 5a and 5b show an example implementation after the submersion of the sample;

[0065] Figure 6 shows a schematic of automatic removal of the sample from the container;

[0066] Figures 7a to 7f show example training data for a computer vision system; and Figure 8 shows a schematic of an example apparatus. DETAILED DESCRIPTION

[0067] An example apparatus for measuring the absorbency of a sample 40 is generally illustrated in Figure 1. The apparatus comprises a container with a first chamber 10 and a second chamber 20, which are connected and partially filled with a liquid 30. In other examples the container has (only) a single chamber.

[0068] In various examples, the first chamber 10 and second chamber 20 comprise transparent walls, which allows a camera 70 to see inside them. In some examples the walls are plastic. In some examples, a different measurement device, such as a LIDAR sensor or rangefinder sensor, or any suitable measurement device, such as those set out above, may be used instead of or in addition to a camera.

[0069] In Figure 1 , the liquid in the container, in this case water, is shown with a surface 32, which has been disturbed by the introduction of a sample 40 into the container by a gripper 60. This causes the sample to not be level. In a number of examples, the liquid is non-disturbed, so is flat and / or level. This is also the case in some examples when a sample is introduced into the container.

[0070] In the example shown in Figure 1 , the gripper 60 comprises two parallel surfaces, mechanically actuated to hold the sample on two opposing sides. In some examples, the surfaces have an angled bottom which the sample sits on, to hold the sample securely. In various examples, the gripper is powered electronically, though pneumatic gripping systems could also be used to introduce the sample.

[0071] In the second chamber 20, there is located a level indicator 50, attached to a vertical support 56. In some examples, the level indicator comprises cork, such that it can float on the surface 32 of the water. In other examples, no level indicator is used, and the location of the surface of the liquid is determined directly.

[0072] Also shown in the example in Figure 1 are possible locations 52, 54 for the level indicator 50. These possible locations correspond to lower water levels in the container, which would be detected by the system accordingly. The apparatus further comprises a camera 70 with field of view 72, containing both the sample 40 and the level indicator 50. The camera is connected to a processor 74. The camera is able to be used to capture one or more images as a video or (stills) picture feed. In some examples, the camera is arranged such that it is level with the surface of the liquid. This can include an arrangement in which the camera is positioned in front of the container. In other examples, the placement of the camera may be different, and the calculation of the liquid level includes compensation for the position of the camera relative to the level indicator and the sample. The images captured by the camera are transmitted to the processor. In the example shown in Figure 1 , the transmission of the images is carried out using a wired connection. In other examples, the transmission may include or may be carried out fully by a wireless connection. In further examples, the processor is included in the camera and transmission outside of the camera is not required.

[0073] In the example shown in Figure 1 , the level indicator 50 and the sample 40 are in the field of view 72 of the camera 70. In this way, the system is able to make measurements of the location of the level indicator using the images captured by the camera 70 on the processor 74, elaborated upon below.

[0074] In Figure 2, an example schematic is shown, detailing the steps of one example process for performing a sink test. The process begins with a level indicator location step S100, in which the computer vision system is used to locate the level indicator. In this example, the centroid of the level indicator is tracked over time using a machine learning algorithm trained with appropriate training data. The specifics of this algorithm and the training data used are discussed below. After locating the level indicator, the processor calculates the location of the liquid surface by extrapolating from the location of the level indicator perpendicular to the axis along which gravity acts, assuming the liquid is constrained under gravity. The processor is aware of the axis along which gravity acts. In this example, no calibration process or external frame of reference is required as the locations of the level indicator and of the sample are measured relative to each-other. In other examples, the locations of the level indicator and of the sample may be measured relative to a third point of reference.

[0075] Next, in some examples, the temperature of the liquid is measured in the temperature measurement step S102. The temperature measurement is carried out using an analogue or digital thermometer or thermocouple, but in this example is carried out using a thermal camera. The temperature of the liquid can be used to regulate the measured sinking time as a function of temperature, in order to make the measured sinking time comparable to sinking times measured at different temperatures.

[0076] Before the sample can be introduced, in some examples, the stability detection step S104 is carried out. The location of the level indicator is measured at intervals to monitor how much the level indicator moves. If the level indicator moves less than, for example, 2 cm from the position measured in the level indicator locations step S100 over, for example, three measurements e.g. five seconds apart, the liquid surface is determined to be adequately stable. If the liquid surface is not stable enough, in various examples, the stability detection step is repeated until the liquid surface is stable.

[0077] The next step of the process is the sample introduction step S106. In some examples the sample is introduced to the liquid by automatic means, e.g. using an electronic gripper, which reduces the amount of manual input required from an operator. In various implementations, the sample is introduced to the liquid in a consistent way, such that the resultant sinking time can be accurately compared between measurements of different samples. While the sample may be introduced from a great height or at speed, minimizing the disturbance of the surface of the liquid will increase the reliability of the measurement.

[0078] In the contact detection step S108, the computer vision system acknowledges the introduction of the sample into the liquid. The sample has been detected and its location and dimensions (as visible to the computer vision system) measured, which is compared to the known location of the liquid surface. When the detected sample intersects with the detected liquid surface, the processor establishes that the sample has contacted the surface of the liquid. The system only detects the first instance of the sample contacting the surface of the liquid, as once the sample has contacted the surface it typically remains there until submersion.

[0079] After the contact detection step, the first time recordal step S110 takes place and the processor records a first time corresponding to the contact between the sample and the liquid surface. In some examples, the time recordal happens with minimal delay from the contact detections step S108, to ensure the accuracy of the sinking time calculation.

[0080] It is possible that the sample that has been introduced to the liquid will not sink. Optionally, in the sinking detection step S112, the height of the sample is repeatedly measured and analysed to determine if the sample is sinking into the liquid. If the sample is not sinking into the liquid, the process cannot continue as there will be no submersion. In this case, an operator is alerted, or the sample is removed from the container, such as in the sample removal step S120. Otherwise, if it is determined the sample is sinking, the process continues.

[0081] As the sample is now sinking into the liquid, the next step of the process is the submersion detection step S114. A submerged sample, in the context of this application, is intended to mean a sample where the whole of the sample is under a liquid surface and the sample is therefore wholly immersed in a liquid. The computer vision system continues to measure the location and dimensions of the sample, and these are compared to the location of the liquid surface as calculated by the processor. After some time, the sample will pass under the liquid surface and become submerged. At this point, there will no longer be an intersection between the calculated location of the liquid surface and the measured bounds of the sample, which will lie wholly beneath. At this point, the processor establishes that the sample has become submerged in the liquid. The system only detects the first instance of the sample becoming submerged, as once the sample is submerged it will, typically, continue to be so until removed.

[0082] After the submersion detection step, the second time recordal step S116 takes place and the processor records a second time corresponding to the submersion. In various examples, the time recordal happens with minimal delay from the submersion detection step S114, to ensure the accuracy of the sinking time calculation.

[0083] As the sample has now sunk into the liquid, the sinking time calculation step S118 occurs and the processor calculates the sinking time based on the first time, second time and, optionally, the temperature of the liquid. The calculation of the sinking time could, in some examples, further comprise the transmission of the sinking time over a wired or wireless connection to another device, such as an edge device, server, or Programmable Logic Controller (PLC).

[0084] Finally, in some examples, the sample is removed from the container in the sample removal step S120. In this example, the sample is removed using a mechanically- actuated fork, which lifts the sample from the container and deposits it outside of the container. More detail about the sample removal is given below.

[0085] The process shown in Figure 2 has various optional steps. These are included above, and include at least steps S102, S104, S112 and S120.

[0086] In Figure 3a and Figure 3b, an example of the system is shown, used to detect the position of a sample 40 and a level indicator 50. In Figure 3a, the container comprises a single chamber 12, made of transparent plastic, which can be seen housing a volume of water 30 with surface 32.

[0087] In the example shown in Figure 3a and Figure 3b, the sample 40 is being introduced manually by the operator with their hand 62 and there is no optional gripper 60. In this example, the sample is being held above the surface of the water and has not yet been introduced. The sample comprises mineral wool, also known as rock wool, formed from fiberised rock. The sample is in the form of a block in this example, and in this example is packaged with a plastic exterior on some sides. In other examples, other materials, such as other porous products, are used, and / or other packaging materials are used, such as fabric or organic sheets and / or different shapes are used. Further, the packaging could be present on some, all, or no sides of the sample. In some examples, the packaging is a different colour to the level indicator, which can make detection of the sample and the level indicator by the computer vision system more accurate.

[0088] In Figure 3b, an annotated version of the images captured by the camera 70 is shown on a computer display. The sample 40, chamber 12, level indicator 50, water 30, and water surface 32 are visible in the image. Also visible are sample annotation 80, level indicator annotation 82, liquid level annotation 84 and status annotation 86, which denote the quantities measured and calculated by the computer vision system.

[0089] In various examples, the sample annotation 80 shows where the computer vision has determined the sample to be, in this case labelled “cube”. The annotation also contains a certainty, in this case “0.95”, informing the operator that the system is 95% confident the annotated region indeed contains the sample.

[0090] In some examples, the level indicator annotation 82 shows where the computer vision system has determined the level indicator to be, in this case labelled “cork”. The annotation also contains a certainty, in this case “0.88”, informing the operator that the system is 88% confident the annotated region indeed contains the level indicator.

[0091] The liquid level annotation 84 differs from the first two annotations discussed in that it is calculated by the computer vision system based on the location of the level indicator. It comprises a line running along the screen denoting the location of the surface of the liquid. In a number of examples, the liquid level annotation is set to be perpendicular to the axis along which gravity acts and intersects with the centre of mass or centre of gravity of the level indicator 50 based on the identification, by the computer vision system, of the level indicator. In other examples, the liquid level annotation is set to intersect with the mid-point of the height of the level indicator with height measured along an axis parallel to the axis along which gravity acts and / or a mid-point of the width of the level indicator with the width measured along an axis perpendicular to the axis along which gravity acts. In various examples, the status annotation 86, meanwhile, uses the measured location of the sample and the calculated location of the liquid surface to determine whether the sample is above the surface, on the surface but not submerged, or submerged. In the example shown in Figure 3b, it can be seen to read “fully above water level”.

[0092] In Figure 4a and 4b, an example of the system is shown. This corresponds to the system shown in Figures 3a and 3b as described above, but with the sample 40 has now been introduced into the water 30. In Figure 4a, the sample can be seen sitting on the surface of the water, partially immersed but not fully below the water surface 32.

[0093] Correspondingly, in Figure 4b, the location of the level indicator is detected and given an annotation 82, along with the sample annotation 80 and the liquid level annotation 84. By comparison of the detected location of the sample and the calculated liquid level, the system infers the conclusion that the sample is sitting on the surface of the water. This is displayed in the text of the status annotation 86, which reads “has just crossed the water level”.

[0094] In Figure 5a and 5b, an example of the system is shown. As with Figures 4a and 4b, this corresponds to the system shown in Figures 3a and 3b as described above, but with the sample 40 now fully submerged below the water level 32. The sample can be seen sitting at the bottom of the chamber 12, though, of course, the same result would be achieved if the sample was shown anywhere below the water level.

[0095] In this case, as can be seen in Figure 5b, the location of the sample annotation 80 is below the surface of the water. By comparison with the calculated location of the water level, the system calculates that the sample is below the level. Accordingly, the status annotation 86 reads “below the water level”. Additionally, a time annotation 88 is shown on the display, informing the operator of the sinking time, in this example “18.2175 seconds”. In Figure 6, a schematic of an example of automatically removing the sample 40 from the chamber 10 is shown. A fork 90a is shown, in a first example position, below the surface 32 of the liquid 30, arranged such that the sample, upon submersion, comes to rest on the fork. The fork provides a wide surface for the sample to fall upon.

[0096] In various examples, the tines 92 of the fork 90a are made of stainless steel. In some example, the tines of the fork are of 5 mm (millimetre) width or diameter.

[0097] The fork comprises six tines in a number of examples. In some examples the tines are spaced with a 25mm gap between them. In other examples, any means, such as a net or platform could be used to remove the sample from the chamber.

[0098] The fork provides the advantage that it does not capture liquid when removed, lifting only the sample from the chamber. This is demonstrated in Figure 6 by the fork 90b also being shown, in a second example position, removed from the liquid 30, removing the sample with it.

[0099] In Figure 7a, an example of training data is shown. This training data is used to train the machine learning algorithms used in the computer vision system in this example to recognise the location of the level indicator and the sample. The machine learning algorithms are of an image classifier type, capable of identifying elements in an image with a given certainty (a function of how closely the element matches the elements in the training data, as a percentage, for example).

[0100] These algorithms typically require a large corpus of annotated training data, where pairs of images and the desired output from the algorithm for that image are given, and statistical mappings between the training data images and the desired outputs are made. Once this training process is complete, the machine learning algorithms are trained and no longer need access to the corpus of training data. Training data including anomalies such as the presence of a human hand helps make the algorithm resilient to their presence in the images.

[0101] In this example, a camera is used to capture the image, shown unannotated in the left panel. In this case, the training data corresponds to the method before the introduction of the sample. In the right panel, the image is shown processed for training, with annotations provided to the computer vision shown (in the figures as filed) as rectangles of different colours. For instance, in the right panel of Figure 7a, the location of the level indicator is annotated with a level indicator annotation 750.

[0102] In Figures 7b to 7d, further example training data is shown. In these cases, both the level indicator and sample are present and are floating on the surface of the liquid. In the right panel, level indicator annotations 750 and sample annotations 740 are shown, teaching the computer vision system to recognise these elements.

[0103] In Figures 7e and 7f, further example training data is shown. In these cases, the level indicator floats on the surface of the liquid, whilst the sample has sunk below the surface and is submerged.

[0104] In each of Figures 7a to 7f it can be seen that the sample has a variety of orientations. Multiple orientations and positions of the sample were provided during the training in order to include a greater variety of data within the training data corpus.

[0105] An example apparatus for measuring the absorbency of a sample 40 is generally illustrated in Figure 8. The reference numerals therein have the same meaning as in the above description, unless stated otherwise. This example details the use of multiple sensors, which may be combined with any other example, or used independently of one another.

[0106] Figure 8 details schematically the measurement of the location of the surface of the liquid (via detection of the liquid level) using a float level 102. The float level measures the liquid level before and after the insertion of the sample into the liquid, detecting a change.

[0107] The location of the surface of the liquid is also detected via the use of a LIDAR range-finding sensor 100. In other examples, other types of range-finding sensor could be used. The location of the surface of the liquid is detected via the emission of radiation 112, which is then detected back at the range-finding sensor 100. Also shown is the use of the LIDAR sensor 100 to detect the location of the sample 40. In other examples, the LIDAR sensor, or any other range-finding sensor, could also be used to locate the level indicator 50. An ultrasonic sensor 104 is used to determine the location of the level indicator 50 via emission and detection of ultrasound 114. Of course, in other examples, an ultrasonic sensor could be used to determine the location of the surface of the liquid in addition to or instead of detecting the location of the level indicator.

Claims

CLAIMS1. A method of assessing liquid absorption of a sample, the method comprising: identifying using at least one of one or more measurement devices, a location of a surface of a liquid inside a container; introducing the sample to the liquid; identifying, using at least one of the one or more measurement devices, based on the location of surface of the liquid, a contacting of the surface of the liquid by the sample and a submersion of the sample in the liquid; recording a first time corresponding to the contacting and a second time corresponding to the submersion; and calculating a sinking time based on the first time and the second time.

2. The method of claim 1 , wherein the one or more measurement devices is a single measurement device.

3. The method of any preceding claim, wherein the one or more measurement devices comprises a computer vision system.

4. The method of any one of the preceding claims, wherein the one or more measurement devices comprises a range-finding sensor.

5. The method of any one of the preceding claims, wherein the one or more measurement devices comprises a float level sensor.

6. The method of any one of the preceding claims, wherein identifying the location of the surface of the liquid comprises identifying the location of a level indicator.

7. The method of claim 6, wherein the level indicator is at least partially immersed in the liquid.

8. The method of any one of the preceding claims, wherein the identifying a contacting of the surface of the liquid comprises detecting, using at least one of the one or more measurement devices, the sample touching the liquid.

9. The method of any one of the preceding claims, wherein the method further comprises, before the step of introducing the sample to the liquid, detecting that the location of the surface of the liquid changes less than a stabilising threshold over a stabilising time.

10. The method of any one of the preceding claims, wherein the sample is introduced to the liquid automatically.

11. The method of claim 10, wherein before the sample is introduced, the sample is held by a gripper, wherein introducing the sample to the liquid comprises the gripper releasing the sample allowing the sample to fall into the container.

12. The method of any one of the preceding claims, wherein the level indicator is located inside a first chamber of the container, the sample being introduced in a second chamber separated from the first chamber.

13. The method of any one of the preceding claims, wherein the detection of contacting of the surface of the liquid by the sample and detection of submersion of the sample comprise detection of a location of a corner of the sample and comparison of the location of the corner with the location of the level indicator.

14. The method of any one of the preceding claims, further comprising an automatic removal of the sample from the container after a sinking time has been calculated.

15. The method of any one of the preceding claims, further comprising detecting, using one of the one or more measurement devices, if a height of the sample relative to the location of the surface of the liquid is unchanging.

16. The method of any one of the preceding claims, wherein the level indicator is coupled to a support such that the level indicator is constrained to move along an axis perpendicular to the surface of the liquid at rest.

17. The method of any one of the preceding claims, further comprising a temperature measurement step, comprising measurement of a temperature of the liquid.

18. The method of claim 17, wherein the calculation of the sinking time is further dependent on the temperature of the liquid.

19. An apparatus for assessing liquid absorption of a sample, the apparatus comprising: a container holding a liquid; one or more measurement devices; and a processor configured to: identify, using a at least one of the one or more measurement devices, a location of the liquid; identify, using at least one of the one or more measurement devices and based on the location of the surface of the liquid, contacting of the surface of the liquid by the sample and a submersion of the sample in the liquid; record a first time corresponding to the contact and a second time corresponding to the submersion; and calculate a sinking time based on the first time and the second time.

Images