The present disclosure relates to a waste sorting robot for sorting waste objects and a method for detecting faults.

In the waste management industry, industrial and domestic waste is increasingly being sorted in order to recover and recycle useful components. Each type of waste, or “fraction” of waste can have a different use and value. If waste is not sorted, then it often ends up in landfill or incineration which may have an undesirable environmental and economic impact.

It is known to sort industrial and domestic waste using a waste sorting robot. The waste sorting robot may pick objects with a suction gripper which uses negative pressure for sucking and gripping an object to be sorted. A problem with existing suction grippers is that the waste sorting robot is used in an environment with a significant amount of variability. For example, waste sorting environment has a significant amount of dust and debris and many waste objects to be sorted are different shapes and sizes.

This means that the information received from sensors may be used to generate an incorrect assessment in respect of waste sorting robot malfunctions e.g. false positives. This reduces the efficiency of the waste sorting robot because the waste sorting robot must be taken offline whilst unneeded maintenance and inspections are carried out.

Examples described hereinafter aim to address the aforementioned problems.

In a first aspect, there is provided a method of cleaning a waste sorting robot having a manipulator moveable within a working area and a suction gripper connected to the manipulator and arranged to selectively grip a waste object in the working area, the method comprising: determining one or more operational parameters of the suction gripper over the plurality of suction gripper operations; and detecting a fault in the suction gripper based on the determined one or more operational parameters; and suppling a solvent to the suction gripper for cleaning the suction gripper in response to detecting the one or more faults in the suction gripper.

Optionally, the method comprises determining a gripping rate of the suction gripper operations over a plurality of suction gripper operations and the detecting the fault in the suction gripper based on the based on the determined one or more operational parameters and the gripping rate of the suction gripper operations. Optionally, the determining the gripping rate of the suction gripper operations comprises determining that the gripping rate of the suction gripper operations drops below a predetermined threshold.

Optionally, the gripping rate of the suction gripper operations is determined over a predetermined number of previous suction gripper operations.

Optionally, the average gripping rate of the suction gripper operations is determined over a previous 10, 50 or 100 suction gripper operations.

Optionally, the determining one or more operational parameters of the suction gripper comprises determining one or more pressure parameters of the suction gripper are outside a normal operating range.

Optionally, the determining one or more operational parameters of the suction gripper comprises determining a highest maximum vacuum pressure of the suction gripper over a predetermined number of previous suction gripper operations.

Optionally, the determining one or more operational parameters of the suction gripper comprises determining that the maximum vacuum pressure is outside a maximum vacuum pressure operating range.

Optionally, wherein the determining one or more operational parameters of the suction gripper comprises determining an average minimum air supply pressure supplied to the suction gripper over a predetermined number of previous sorting operations.

Optionally, the determining one or more operational parameters of the suction gripper comprises determining that the minimum air supply pressure is within a minimum air supply pressure operational range.

Optionally, the method comprises generating an alert in dependence of the detecting one or more faults.

Optionally, the method comprises determining the type of the one or more faults in dependence on the determined gripping rate and the determined parameters and including the type of the one or more faults in the alert. Optionally, the detecting a fault in the suction gripper comprises determining that the suction gripper comprises a build-up of residue.

Optionally, the method comprises varying the flow rate of the solvent supplied to the suction gripper in dependence of the determined one or more parameters and the determined gripping rate of suction gripper operations.

Optionally, the varying the flow rate of the solvent comprises increasing the flow rate of the suction gripper in dependence of the rate of decrease of the highest maximum vacuum pressure and I or the gripping rate of suction gripper operations.

Optionally, the supplying the solvent comprises dosing an airflow in the suction gripper with the solvent.

Optionally, the supplying comprises actuating a valve in fluid connection between a solvent outlet mounted on the suction gripper and a solvent supply.

In a second aspect of the disclosure, there is provided a computer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of the method according to the first aspect.

In a third aspect of the disclosure, there is provide a waste sorting robot comprising: a manipulator moveable within a working area; a suction gripper connected to the manipulator and arranged to selectively grip a waste object in the working area; a solvent outlet mounted on the suction gripper and in fluid communication with a solvent valve arranged to selectively supply solvent to the solvent outlet; and a controller configured to: determine one or more operational parameters of the suction gripper over a plurality of suction gripper operations; detect one or more faults with the suction gripper based on the determined one or more operational parameters; and actuate the valve configured to supply the solvent to the suction gripper for cleaning the suction gripper in response to detecting the one or more faults in the suction gripper.

Various other aspects and further examples are also described in the following detailed description and in the attached claims with reference to the accompanying drawings, in which:

Figure 1 shows a perspective view of a waste sorting robot; Figure 2 shows a schematic front view of a waste sorting robot;

Figure 3 shows a perspective view of a suction gripper;

Figure 4 shows a cross-sectional view of a suction gripper;

Figure 5 shows a schematic view of a waste sorting robot;

Figures 6a, 6b, 6c, 7a, 7b, and 7c show graphs of different parameters of the waste sorting robot in different operational scenarios; and

Figures 8 and 9 show flow diagrams for operation of a waste sorting robot.

Figure 1 shows a perspective view of a waste sorting robot 100. In some examples, the waste sorting robot 100 can be a waste sorting gantry robot 100. In other examples other types of waste sorting robots can be used. For the purposes of brevity, the examples will be described in reference to waste sorting gantry robots but the examples described below can be used with other types of robot such as robot arms or delta robots. In some other examples, the waste sorting robot 100 is a Selective Compliance Assembly Robot Arm (SCARA).

The waste sorting robot 100 comprises a controller 200 (schematically shown in Figure 2) for sending control and movement instructions to a manipulator 104 for interacting with a waste object 106 to be sorted. For the purposes of clarity, only one waste object 106 is shown in Figure 1 but there can be any number of waste objects 106 moving past the waste sorting robot 100. The controller 200 may be implemented on hardware, firmware or software operating on one or more processors or computers. A single processor can operate the different functionalities or separate individual processors, or separate groups of processors can operate each functionality.

The combination of the controller 200 sending control instructions to the manipulator 104 can also be referred to as a “robot”. The controller 200 is located remote from the manipulator 104 and in some examples is housed in first and second cabinets 112, 116. In other examples, the controller 200 can be integral with the manipulator 104 and / or a gantry frame 102. In some examples, part of the gantry frame 102 is housed in the first and second cabinets 112, 116 for shielding one or more components of the waste sorting robot 100.

The manipulator 104 physically engages and moves the waste object 106 that enters a working area 108 in order to sort the waste object 106. The working area 108 of a manipulator 104 is an area within which the manipulator 104 is able to reach and interact with the waste object 106. The working area 108 as shown in Figure 1 is a cross hatched area beneath the manipulator 104. The manipulator 104 is configured to move at variable heights above the working area 108. In this way, the manipulator 104 is configured to move within a working volume defined by the height above the working area 108 where the robot can manipulate the waste object 106. The manipulator 104 comprises one or more components for effecting relative movement with respect to the waste object 106. The manipulator 104 will now be described in further detail.

As shown in Figure 1 , the manipulator 104 is configured to move within the working volume. The manipulator 104 comprises one or more servos, pneumatic actuators or any other type of mechanical actuator for moving the manipulator 104 in one or more axes. For the purposes of clarity, the servos, pneumatic actuators or mechanical actuators are not shown in Figure 1. Movement of the manipulator 104 is known and will not be discussed any further. A suction gripper 120 is coupled to the manipulator 104 and suction gripper 120 is discussed in further detail below.

The servos, pneumatic actuators or mechanical actuators are connectively connected to the controller 200 and the controller 200 is configured to issue instructions for actuating one or more of the servos, pneumatic actuators or mechanical actuators to move the manipulator 104 within the working area 108. Connections (not shown) between the servos, pneumatic actuators or mechanical actuators and the controller 200 can comprise one or more data and I or power connections. The control of servos, pneumatic actuators or mechanical actuators to move of the manipulator 104 is known and will not be discussed any further.

The waste object 106 is moved into the working area 108 by a conveyor belt 110. The path of travel of the conveyor belt 110 intersects with the working area 108. The direction of the conveyor belt 110 is shown in Figure 1 by two arrows. This means the waste object 106 moving on the conveyor belt 110 will pass through the working area 108. The conveyor belt 110 can be a continuous belt, or a conveyor belt formed from overlapping portions. The conveyor belt 110 can be a single belt or alternatively a plurality of adjacent moving belts (not shown).

In other examples, the waste object 106 can be conveyed into the working area 108 via other conveying means. The conveyor belt 110 can be any suitable means for moving the waste object 106 into the working area 108. For example, the waste object 106 are fed under gravity via a slide (not shown) to the working area 108.

The waste object 106 can be any type of industrial waste, commercial waste, domestic waste or any other waste which requires sorting and processing. Unsorted waste material comprises a plurality of fractions of different types of waste. Industrial waste can comprise fractions, for example, of metal, wood, plastic, hardcore and one or more other types of waste. In other examples, the waste can comprise any number of different fractions of waste formed from any type or parameter of waste. The fractions can be further subdivided into more refined categories. For example, metal can be separated into steel, iron, aluminium etc. Domestic waste also comprises different fractions of waste such as plastic, paper, cardboard, metal, glass and I or organic waste. A fraction is a category of waste that the waste can be sorted into by the waste sorting gantry robot 100. A fraction can be a standard or homogenous composition of material, such as aluminium, but alternatively a fraction can be a category of waste defined by a customer or user.

The waste sorting robot 100 is arranged to sort the waste object 106 into fractions according to one or more parameters of the waste object 106. The controller 200 receives information from the at least one sensor (not shown) corresponding to the waste object 106 on the conveyor belt 110. The at least one sensor is positioned in front of the manipulator 104 so that detected measurements of the waste object 106 are sent to the controller 200 before the waste object 106 enters the working area 108. In some examples, the at least one sensor can be any sensor suitable for determining a parameter of the waste object 106 e.g. one or more of a RGB camera, an infrared camera, a metal detector, a hall sensor, a temperature sensor, visual and I or infrared spectroscopic detector, 3D imaging sensor, terahertz imaging system, radioactivity sensor and / or a laser e.g. LIDAR. Additionally or alternatively, the at least one sensor is configured to detect the waste object 106 and send signals to the controller 200 when the waste object 106 enters or is in the working area 108.

The controller 200 determines instructions for moving the manipulator 104 based on the received information according to one or more criteria. Various information processing techniques can be adopted by the controller 200 for controlling the manipulator 104. Such information processing techniques are described in WO2012/089928, WO2012/052615, WO2011/161304, W02008/102052 which are incorporated herein by reference. Techniques for sorting the waste object 106 are known and will not be discussed any further.

Once the manipulator 104 has received instructions from the controller 200, the manipulator 104 executes the commands and moves the suction gripper 120 to pick the waste object 106 from the conveyor belt 110. The process of selecting and manipulating the waste object 106 on the conveyor belt 110 is known as a “pick”. Once a pick has been completed, the manipulator 104 drops or throws the waste object 106 into a chute 114 adjacent to the conveyor belt 110. The mix of waste products means that the environment of the waste sorting robot 100 can be particularly dirty. For example the conveyor belt 110 can be dusty and be covered with debris. This means that the waste sorting robot 100 operates in a challenging environment and maintenance must be regularly carried out on parts of the waste sorting robot 100 such as the manipulator 104. Furthermore, often such types of waste objects can comprise organic matter. For example, domestic waste objects can comprise residual waste food. This is often sticky and can adhere to parts of the waste sorting robot 100. Mitigation of the dirt contaminating the waste sorting robot 100 will described in more detail below.

A waste object 106 dropped into the chute 114 is considered to be a successful pick. In order to achieve a successful pick, the waste sorting robot 100 must also perform a successful gripping operation. A successful gripping operation is an operation performed by the suction gripper 120 whereby by the waste object 106 is gripped and then moved to the intended destination e.g. the chute 114. In some other examples, the intended destination can be another conveyor belt (not shown), a pile of other waste objects (not shown), a bin or any other location for receiving sorted waste objects 106. The manipulator 104 can move the waste object 106 to the intended destination by using any suitable technique e.g. throwing, blowing, moving, or placing etc the waste object 106. A controller 200 determines whether a successful gripping operation has occurred in dependence of a signal received from a sensor on the suction gripper 120 e.g. the first and second pressure sensors 408, 410 (as discussed in reference to Figures 4 below.) In some examples, a successful gripping operation is determined when the controller determines that a maximum vacuum pressure in the suction gripper is achieved.

If the suction gripper 120 fails to grip and move the waste object 106 to the intended destination then this is an unsuccessful gripping operation. An unsuccessful gripping operation can include failing to lift the waste object 106 off the conveyor belt 110 or dropping the waste object 106 before moving the waste object 106 to the chute 114. . In this case the controller 200 receives a signal that there is no vacuum pressure or vacuum pressure has been lost too soon during a gripping operation.

The % gripping rate R of the gripping operations is calculated as follows: 100 where g _s is the number of successful gripping operations, gt is the number of failed gripping operations and g _s + gr is the total number of gripping operations.

Whilst clogging of the suction gripper 120 is likely to decrease the actual picking success rate, the % gripping rate R may not reflect the picking success rate. Since R is a derivative of the first pressure sensor 408, the result of a clog could indicate that:

1 ) the % gripping rate R is 100% because the first pressure sensor 408 detects gripping the clogged object;

2) none of the pick attempts are successful and the % gripping rate R is 0%;

3) or the % gripping rate R is between 0% to 100%.

In this way the % gripping rate R is not a measure of the true picking success rate, but an indication of the operational performance of the waste sorting robot 100. The % gripping rate R will be a reliable indicator of the picking success rate only when there is no interference such as objects stuck in the suction gripper 120.

Clogging of the suction gripper 120 is likely to decrease the actual pick success rate of the waste sorting robot 100. However the % gripping rate R which is derived from the first pressure sensor 408 of the suction gripper 120 may not show the decrease in actual pick success rate. Accordingly, one or more other operational parameters are used to infer operational performance of the waste sorting robot 100 in addition to the % gripping rate R.

In some examples, the controller 200 comprises a statistical module 250 configured to compute statistical information relating to one or more parameters of the waste sorting robot 100, the suction gripper 120 and the operation thereof. Similar to the controller 200, the statistical module 250 may be implemented on hardware, firmware or software operating on one or more processors or computers. A single processor can operate the different functionalities or separate individual processors, or separate groups of processors can operate each functionality. The statistical module 250 as shown in Figure 2 is part of the controller 200, although in other examples, the statistical module 250 can be a separate remote processor (not shown).

In some examples, the controller 200 determines whether a picking operation comprises a successful gripping operation or not. In some examples, the controller 200 determines the nature of the gripping operation based on received sensor information. This will be discussed in more detail below. In other examples, the controller 200 receives information relating to the nature of the gripping operation from another source e.g. another controller (not shown) or from an operator.

The controller 200 is connected to a first pressure sensor 408 (as shown in Figure 4) via a communication line 218. The first pressure sensor 408 is arranged to detect the vacuum pressure in the suction cup 220 and the suction tube 400. Accordingly, if the suction gripper 120 fails to successfully grip the waste object 106, the first pressure sensor 408 will send pressure measurement information to the controller 200 indicating that there is no or insufficient vacuum pressure in the suction cup 220. This indicates that the suction cup 220 has not achieved making a seal against the surface of the waste object 106. This means that the suction gripper 120 is not able to grip, lift and move the waste object 106.

The controller 200 can receive pressure measurement information from the first pressure sensor 408 that there is no or insufficient vacuum pressure in the suction cup 220 whilst the manipulator 104 is moving or about to move. In this case, the controller 200 can determine that the waste object 106 was not lifted off the conveyor belt 110 or the waste object 106 fell off the suction gripper 120 during a gripping operation. In some examples, the controller 200 sends information relating to the nature of the gripping operation to the statistical module 250. In some examples, the statistical module 250 determines the % gripping rate R of the gripping operations.

The waste sorting robot 100 will now be described in reference to Figure 2. Figure 2 shows a schematic front view of the waste sorting robot 100. The suction gripper 120 comprises a suction cup 220 for physically engaging with a surface of the waste object 106.

The suction gripper 120 is in fluid communication with a pneumatic system 222. The pneumatic system 222 comprises at least a first air hose 202 for connecting the suction gripper 120 to a compressed air supply. For the purposes of clarity, only the first air hose 202 is shown in Figure 2 connecting the suction gripper 120 to the compressed air supply but there can be any number of air hoses connected between the suction gripper 120 and the compressed air supply. For example, there can optionally be at least a second air hose connecting the suction gripper 120 to the compressed air supply. In this way, a second source of air is provided to the suction gripper 120 for operating a blow tube 402 (discussed in reference to Figure 4 below).

In some examples, the first air hose 202 can be connected to a plurality of downstream supply air hoses 500, 502 (as shown in Figure 5) for supplying compressed air to a plurality of pneumatic components in the pneumatic system 222. For example, the first air hose 202 is a single, unitary air hose mounted on the manipulator 104. By providing only the first air hose 202 which is mounted on the manipulator 104 to the suction gripper 120, installation and maintenance of the waste sorting robot 100 can be simplified. The first air hose 202 is flexible and mounted to the gantry frame 102 and I or the manipulator 104. The first air hose 202 is sufficiently flexible to move and flex so as to change shape as the manipulator 104 moves without impeding the movement of the manipulator 104.

The pneumatic system 222 comprises an air compressor 206 for generating a source of compressed air. Optionally, the pneumatic system 222 can also comprise an air storage tank (not shown) for compressed air. Furthermore, the pneumatic system 222 can also comprise one or more pneumatic valves 204 for selectively providing air to the suction gripper 120. In this way, the pneumatic system 222 comprises air supply such as air compressor 206 in fluid connection to the suction gripper 120 configured to generate an airflow along an airflow path between the air supply e.g. the air compressor 206 and the suction gripper 120. In other examples, the air supply can be provided by any suitable source of compressed air or compressed gas.

The pneumatic system 222 is schematically shown as being located within the first cabinet 112. However, in other examples the pneumatic system 222 can be partially or wholly located remote from the waste sorting robot 100. For example, there may be a plurality of waste sorting robots 100 on a sorting line (not shown) each of which require a source of air. In this way, a single air compressor 206 can be connected to a plurality of waste sorting robots 100 via a plurality of air hoses. Accordingly, the pneumatic system 222 may be located between waste sorting robots 100.

Operation of the pneumatic system 222 is controlled by the controller 200. The controller 200 is connected via pneumatic control lines 208, 210 to the pneumatic system 222, the air compressor 206 and the pneumatic valve 204. The controller 200 is configured to send control instructions to the pneumatic system 222, the air compressor 206, and the pneumatic valve 204. This means that the controller 200 can selectively operate e.g. the air compressor 206 or the pneumatic valve 204 to deliver a supply of air to the suction gripper 120.

The waste sorting robot 100 as shown in Figure 2 also comprises a solvent supply such as solvent tank 212 for dissolving organic matter or other dirt dried onto the suction gripper 120 or other parts of the pneumatic system 222. In some other examples, the solvent supply is alternatively or additionally a pipe (not shown) in fluid communication with the solvent outlet 214. For example the pipe can be a pipe for feeding solvent e.g. a mains water pipe. This means that the waste sorting robot 100 always connected to a solvent supply and the solvent tank 212 does not have to be replenished. The waste sorting robot 100 can be installed in a remote location and it may not be possible to connect the waste sorting robot 100 to a mains water supply.

In some examples, the solvent is water, but in other examples the solvent can be ethanol, methanol, ammonia, acetone or any other suitable solvent for dissolving organic matter or other dirt dried to the waste sorting robot 100. In a preferred example, the solvent is water because it is easier for the operator to handle water than other solvents. However, there may be certain types of waste objects that contaminate the waste sorting robot 100 with dirt that is not easily removed with water. For example, silicone sealant tubes may contaminate the waste sorting robot 100 with silicone sealant. Silicone sealant may require another solvent other than water for successful removal. The term “solvent” will be used to describe the examples and refer to any suitable fluid for dissolving or removing dirt and debris stuck to the surfaces of the waste sorting robot 100.

In some examples, the solvent can optionally comprise one or more additives. In some examples one or more of a disinfectant, surfactant, detergent, dispersant is added to the solvent to help removal of dirt from the waste sorting robot 100.

The waste sorting robot 100 comprises a solvent outlet 214 which is in fluid connection with the solvent tank 212. The solvent outlet 214 is positioned along the airflow path of the suction gripper 120 and configured to dose the airflow with the solvent. As shown in Figure 2, the solvent outlet 214 is mounted on the suction gripper 120. In other examples (not shown), the solvent outlet 214 can be mounted on one or more of the first air hose 202, or downstream supply air hoses 500, 502 and I or the pneumatic valve 204. In this way, the solvent outlet 214 is arranged to dose the airflow remote from the suction gripper 120.

Similar to the pneumatic system 222, the waste sorting robot 100 comprises a solvent hose 228 in fluid communication between the solvent outlet 214 and the solvent tank 212. In some examples, the solvent hose 228 is mounted on the manipulator 104. The solvent hose 228 is flexible and mounted to the gantry frame 102 and I or the manipulator 104. The solvent hose 228 is sufficiently flexible to move and flex so as to change shape as the manipulator 104 moves without impeding the movement of the manipulator 104. In some examples, the solvent tank 212 can comprise a pump (not shown) for urging the solvent from the solvent tank 212 to the solvent outlet 214. In some examples, the solvent tank 212 can be pressurised and no pump is required. For example, a third air hose (not shown) and another pneumatic valve (not shown) can be coupled to the solvent tank 212 for pressurising the solvent in the solvent tank 212. In this way, the controller 200 can selectively control pressurising the solvent tank 212 to control the flow of the solvent to the solvent outlet 214. Alternatively, the solvent tank 212 is mounted on the gantry frame 102 in a position above the suction gripper 120. This means that the solvent will be fed to the solvent outlet 214 via gravity alone.

In some examples, the solvent tank 212 is mounted in the first cabinet 112 or the second cabinet 116 and easily accessible to an operator. The solvent tank 212 can optionally have a transparent window in a wall of the solvent tank 212 or the wall of the solvent tank 212 can be translucent. This means that the operator can visually inspect the amount of solvent left in the solvent tank 212.

As shown in Figure 2, there is a solvent valve 216 arranged to selectively control a flow of solvent to the solvent outlet 214. Operation of the solvent dosing is controlled by the controller 200. The controller 200 is connected via solvent control line 230 to the solvent valve 216. The controller 200 is configured to send control instructions to solvent valve 216. This means that the controller 200 can selectively operate the solvent valve 216 to deliver a supply of solvent to the solvent outlet 214 mounted on the suction gripper 120.

Optionally, there is no solvent valve 216 and the solvent is selectively fed to the solvent outlet 214 by the controller 200 selectively pressurizing the solvent tank 212 or selectively controlling a solvent pump (not shown).

In some examples, the controller 200 can actuate the solvent valve 216 to modify the flow rate of the solvent to the solvent outlet 214. This means that the controller 200 can adjust the solvent flow rate if the waste sorting robot 100 is particularly dirty. In some examples, the controller 200 can adjust the flow rate of the solvent in dependence on a dirt parameter of the waste sorting robot 100. In some examples, the controller 200 can adjust the flow rate of the solvent in dependence of determining pressure parameters of the suction gripper 120. This is discussed in further detail below.

Optionally, an operator can manually input information relating to the dirt parameter relating to the cleanliness of the waste sorting robot 100. Additionally or alternatively, the controller 200 can optionally receive and analyse images of the suction cup 220 to determine whether the suction cup 220 is soiled with dried dirt e.g. dried organic matter.

An example of the suction gripper 120 will now be discussed in reference to Figures 3 and 4. Figure 3 shows a perspective view of the suction gripper 120 without the suction cup 220. Figure 4 shows a cross-sectional side view of the suction gripper 120. As mentioned previously, the suction gripper 120 comprises a suction cup 220 (as shown in Figure 4). The suction cup 220 as shown in Figure 4 has a cup shape e.g. an approximate hemispherical shape. However, other known suction cups can be used instead e.g. a ribbed cylindrical suction cup 506 as shown in Figure 5.

The suction gripper 120 as shown in Figure 4 comprises an integrated suction tube 400 and blow tube 402 for carrying out grip I pick operations and throwing operations. This is known and will not be discussed in any further detail.

The suction gripper 120 comprises a suction tube air supply inlet 300 which is in fluid communication with the first air hose 202 (not shown in Figure 3). The suction tube air supply inlet 300 introduces a fast, high pressure source of air into the suction tube 400 which creates a vacuum pressure in the suction tube 400 represented by the arrows in Figure 3. The vacuum pressure is also created in the suction cup 220 since the suction cup 220 is in fluid communication with the suction tube 400.

As shown in Figure 4, the suction gripper 120 also comprises a blow or “sneezing” tube 402 connected to the suction tube 400. The blow tube 402 is essentially the same as the suction tube 400 but reversed in orientation to generate a positive air pressure rather than a negative air pressure (e.g. a vacuum pressure).

Similar to the suction tube 400, the blow tube 402 comprises a blow tube air supply inlet 302 which is in fluid communication with the first air hose 202. Accordingly, the blow tube air supply inlet 302 introduces a second air supply into the suction gripper 120.

In some examples the first air hose 202 is coupled between the air compressor 206 and a pneumatic valve 204. In some examples the pneumatic valve 204 which is a three-way valve 504 (as best shown in Figure 5). The three-way valve 504 is configured for selectively providing an air flow to either the suction tube 400 or the blow tube 402. In some examples, the suction tube 400 comprises a first opening 404 to receive the first pressure sensor 408 to measure the vacuum pressure in the suction gripper 120. In some examples, the first pressure sensor 408 is configured to detect the maximum vacuum pressure p _v max in the suction gripper 120.

Likewise, the blow tube 402 comprises a second opening 406 to receive a second pressure sensor 410 to measure the positive pressure when the suction gripper 120 operates in a blow mode. The first and second pressure sensors 408, 410 are connected to the controller 200 and send signals to the controller 200. Only the communication line 218 between the first pressure sensor 408 and the controller 200 is shown for the purposes of clarity in Figure 2.

The first pressure sensor 408 is configured to measure the pressure in the suction tube 400 and the suction cup 220. In some examples, the controller 200 can receive pressure measurement information from the first pressure sensor 408. The controller 200 is configured to determine the maximum vacuum pressure p _{v m} ax of the suction tube 400.

The vacuum pressure p _v of the suction tube 400 defined as follows:

Pv Patm Pabs

Wherein p _atm is the atmospheric pressure and p _a bs is the absolute pressure in the suction gripper 120. Absolute pressure is the pressure in the suction gripper 120 measured with respect to a hard vacuum (e.g. a pressure of 0 bar).

In this way, the maximum vacuum pressure p _{v m} ax of the suction tube 400 is the greatest difference between atmospheric pressure and the absolute pressure of the suction tube 400. In other words, this measures the ability of the pneumatic system 222 to create a partial vacuum in the suction tube 400. The maximum vacuum pressure p _{v m} ax of the suction gripper 120 is an important parameter of the suction gripper 120 because it relates to the maximum gripping force of the suction gripper 120. For example, maximum vacuum pressure p _{v m} ax of the suction gripper 120 relates to the maximum weight of the waste object 106 that can be lifted by the suction gripper 120. The maximum vacuum pressure p _{v m} ax of the suction gripper 120 also relates to the combined maximum acceleration and weight of the waste object 106 that can be lifted by the suction gripper 120.

The maximum vacuum pressure p _vm ax is also important because not every gripping operation will achieve the maximum vacuum pressure p _{v m} ax. For example, the waste object 106 can have an irregular shape and surface texture so a good seal may not be possible in every gripping operation. Accordingly, the suction gripper 120 may need to generate a certain maximum vacuum pressure p _vm ax to pick the waste object 106 with an imperfect seal between the suction gripper 120 and the waste object 106.

In addition, the second pressure sensor 410 sends pressure information to the controller 200. This means that the controller 200 can determine the positive sneeze pressure p _sn eeze of the blow tube 402.

The pneumatic system 222 also comprises an air supply pressure sensor 224. The air supply pressure sensor 224 is connected to the controller 200 via a communication line 226. The air supply pressure sensor 224 is configured to measure the pressure of the compressed air supply to the suction gripper 120. In some examples the air supply pressure sensor 224 is mounted in the first cabinet 112. In some other examples, the air supply pressure sensor 224 is mounted on the suction tube 400, for example mounted at the suction tube air supply inlet 300 of the suction tube 400. In some other examples, the air supply pressure sensor 224 is mounted on the first air hose 202, for example a gauge (not shown). In this way, the air supply pressure sensor 224 sends pressure information to the controller 200. The controller 200 is configured to determine the minimum pressure p _{as m} in of the air supplied to the suction gripper 120.

The minimum air supply pressure p _{as m} in is an important parameter of the suction gripper 120 because it relates to whether suction gripper 120 is operational for a specified gripping performance.

As shown in Figures 4 and 5, the solvent hose 228 is attached to the solvent outlet 214 in the same way as described in reference to Figure 2. The position of the solvent outlet 214 with respect to the blow tube 402 means that the solvent outlet 214 introduces the solvent into the positive pressure airflow path. Accordingly, the solvent is entrained in the airflow (as indicated by the arrow in Figure 4) in the blow tube 402 and the solvent is blown out of the suction gripper 120 and the suction cup 220. Since the solvent is introduced in the blow tube 402 part of the suction gripper 120, the solvent is not sucked into the pneumatic system 222. Even if the suction tube 400 is operational, the solvent will be ejected from the blow tube 402 at the open end of the blow tube 402, e.g. the blow tube first air inlet 412.

Whilst Figures 4 and 5 shows the solvent outlet 214 is positioned opposite second opening 406 to receive a second pressure sensor 410, the solvent outlet 214 can be positioned at any position along the suction gripper 120 (aligned along axis A-A). The solvent outlet 214 can be mounted in the suction cup 220, the suction tube 400 or any other component of the pneumatic system 222.

Control of the solvent dosing operation will be discussed in reference to the controller 200 and the statistical module 250 determining operational parameters of the suction gripper 120 as discussed in reference to Figures 6a, 6b, 6c, 7a, 7b, 7c. In this way, the controller 200 and the statistical module 250 can analyse the performance of the suction gripper 120 and issue control signals to supply solvent to the suction gripper 120. This means that the controller 200 can determine when the suction gripper 120 becomes clogged with sticky residue and automatically clean the suction gripper 120 with a solvent dosing operation.

Turning to Figures 6a, 6b, 6c, operation of the waste sorting robot 100 will be discussed in further detail. Figures 6a, 6b, 6c show graphs of different parameters of the waste sorting robot 100 normal operational scenarios.

Figures 6a, 6b, 6c show normal operation of the waste sorting robot 100. Figure 6a shows a graph of the % gripping rate R of gripping operations over time, Figure 6b shows a graph of the maximum vacuum pressure p _{v m} ax (mbar) over time, and Figure 6c shows a graph of the minimum air supply pressure p _{as m} in (bar) over time.

Figures 6a, 6b, 6c show a series of four picking operations over time. The different series of four picking operations are separated indicating that there is a period of time between the series of picking operations where the waste sorting robot 100 was not in operation.

As shown in Figure 6a, in normal operation the % gripping rate R of the gripping operations is generally above a predetermined threshold. The normal R range 600 is shown by a rectangle which represents a % gripping rate R of between 75% to 100%. In some examples, the normal R range 600 of the % gripping rate R can be varied to any other suitable ranges or combination thereof e.g. between 85% to 100%, 90% to 100%, 95% to 100% etc.

A below normal R range 602 is shown by rectangle which represents a % gripping rate R of between 50% to 75%. In some examples, if the gripping rate R of the gripping operations remains in or lower than the below normal R range 602, then the statistical module 250 sends a signal to the controller 200. This can indicate a fault with the waste sorting robot 100 or the suction gripper 120 and the controller 200 can generate an alert to the operator. In some examples, the below normal R range 602 of the % gripping rate R can be varied to any other suitable ranges or combination thereof e.g. between 60% to 85%, 65% to 90%, 70% to 95% etc.

The % gripping rate R is determined as previously mentioned. As can be seen from Figure 6a, there is some variation in the % gripping rate R of the gripping operations. The variation in the % gripping rate R of the gripping operations is because different types of waste objects 106 have different % gripping rates R. For example, some types of waste objects 106 are easier to successfully pick than other types of waste objects 106. In this way, the % gripping rate R of the gripping operations can be lowered temporarily due to external factors such as the type of waste being sorted, but nevertheless, the waste sorting robot 100 and the suction gripper 120 are operating normally.

Since there is inherent variability in the % gripping rate R of the gripping operations during normal operations, a moving % gripping rate R of the gripping operations (rather than a cumulative % gripping rate) is more indicative of whether there is a fault with the waste sorting robot 100 and I or the suction gripper 120. The moving % gripping rate R of the gripping operations is calculated as previously discussed. This means that the % gripping rate R of the gripping operations is calculated based on a number n of the most recent gripping operations. In some examples, the moving % gripping rate R is reset every time the waste sorting robot 100 is turned on.

In some examples, the statistical module 250 determines the moving % gripping rate R of the gripping operations. The statistical module 250 determines the moving % gripping rate R of the gripping operations over a predetermined number n of previous operations. In some examples, the statistical module 250 is configured to determine the moving % gripping rate R of the gripping operations over the previous n 10, 50, 100, 200, 500, or 1000 suction gripper operations. In some examples, the statistical module 250 is configured to determine the moving % gripping rate R of the gripping operations over any number of previous gripping operations.

The number n of previous gripping operations can be varied depending on the required sensitivity for detecting changes in R. However, the fewer the number n of suction gripper operations used to calculate R, the more likely R is to be affected by false positives. In contrast, the greater the number n of suction gripper operations used to calculate R, the more accurate R. However, with a greater number n of suction gripper operations used to calculate R, the slower R will change when the waste sorting robot 100 malfunctions. In some examples, the controller 200 sends a signal to the statistical module 250 to change the number n in order to increase the accuracy of R or decrease n to increase the sensitivity of R.

Figure 6b shows the maximum vacuum pressure p _v max over time. Figure 6b shows the maximum vacuum pressure p _{v m} ax as the instantaneous maximum vacuum pressure detected in the suction gripper 120 represented by thick line 612.

At the same time, for n consecutive suction gripper operations, the statistical module 250 records the highest maximum vacuum pressure p _v max which is referred to as p _v high_max. hereinafter.

By measuring maximum vacuum pressure p _{v m} ax and highest maximum vacuum pressure p _v hi _g h_max operational parameters of the waste sorting robot 100 can easily be determined from the first pressure sensor 408. These operational parameters can easily indicate the performance of the waste sorting robot 100 without detecting that a pick has been successful i.e. the waste object 106 has been placed or thrown into a chute.

In some examples, the statistical module 250 determines the highest maximum vacuum pressure p _v high_max. In some examples, the controller 200 sends a signal to the statistical module 250 to change the number n in order to increase the accuracy of p _V high_max or decrease n to increase the sensitivity of p _V high_max.

As shown in Figure 6b, in normal operation the instantaneous maximum vacuum pressure p _v max is generally above a predetermined threshold. The predetermined threshold of the maximum vacuum pressure p _{v m} ax is an operational specification maximum vacuum pressure Pvmax_ _sp ec of the waste sorting robot 100. That is, the designed maximum vacuum pressure p _v max for the waste sorting robot 100. As shown in Figure 6b, the predetermined threshold is represented as a range 604 which reflects an operational tolerance in the variability of the maximum vacuum pressure p _{v m} ax during operation. In some examples, the predetermined threshold can be represented on the graph in Figure 6b as a straight line 614 representing the specification maximum vacuum pressure p _V max_spec without any operational tolerance. Figure 6b shows four separate waste sorting operations and each has a highest maximum vacuum pressure p _V high_max within a normal range p _V high_max range 604.

The normal p _V high_max range 604 is shown by a rectangle which represents a range between 600 to 800 mbar. A below normal p _V high_max range 606 of the is shown by rectangle which represents 500 to 600 mbar. During normal operations as shown in scenario 1 , the highest maximum vacuum pressure p _V high_max lies within the normal p _v high_max range 604. In some examples, the normal p _v high_max range 604 can be varied to any other suitable ranges or combination thereof e.g. between 650 to 850 mbar, 700 to 900 mbar, 800 to 950 mbar. In some examples, the below normal p _V high_max range 606 can be varied to any other suitable ranges or combination thereof e.g. between 550 to 550 mbar, 600 to 700 mbar, 700 to 850 mbar etc.

In some examples, if the highest maximum vacuum pressure p _V high_max remains in or lower than the below normal p _V high_max range 606, then the statistical module 250 sends a signal to the controller 200. This can indicate a fault with the waste sorting robot 100 or the suction gripper 120 and the controller 200 can generate an alert to the operator. Use of the below normal p _V high_max range 606 is optional and in other examples, the statistical module 250 sends a signal to the controller 200 when the highest maximum vacuum pressure pvhigh_max \\s below and indicates a fault with the suction gripper 120 and I or the waste sorting robot 100

Similar to % gripping rate R of the gripping operations, there is also some variation in the highest maximum vacuum pressure p _V high_max . The variation in the highest maximum vacuum pressure p _V high_max is because different types of waste objects 106 have different properties. For example, the suction gripper 120 can make good seals against smooth surfaces but not against rough or crumpled surfaces.

The statistical module 250 is configured to compare the maximum vacuum pressure p _{v m} ax, highest maximum vacuum pressure p _v high_max, and the specification maximum vacuum pressure p _V max_spec- If the statistical module 250 determines that the highest maximum vacuum pressure p _V high_max is lower than the specification maximum vacuum pressure p _{v m} ax_spec then, the statistical module 250 may determine that there were no “good” e.g. no suitable grippable objects. Due to the inherent variability of waste objects, some waste objects are good for gripping and some waste objects are bad for gripping. For example, a good waste object for gripping may be hard, smooth, and I or solid surface against which a high vacuum can be generated in the suction gripper 120. For example, a bad waste object for gripping may be porous, rough and I or flexible against which a low vacuum can only be generated in the suction gripper 120.

By using the highest maximum vacuum pressure p _v high_max to assess the operational performance of waste sorting robot 100, it is possible to assess whether there is a fault with the waste sorting robot 100 rather than variability in the type of waste objects 106. For example, if the highest maximum vacuum pressure p _V high_max remains high e.g. close to the specification maximum vacuum pressure p _V max_spec but a lower % gripping rate R, then there is a degree of confidence that the is not a problem with the waste sorting robot 100. Nevertheless, a gradual degradation in operational performance will be shown as a downward slope for the highest maximum vacuum pressure p _v high_max. At some point, the highest maximum vacuum pressure p _V high_max will not be high enough for the suction gripper 120 to generate a high enough vacuum pressure to grip even the “good” waste objects. In other words, a decreasing highest maximum vacuum pressure p _v high_max indicates that there is probably a problem with the waste sorting robot 100.

Alternatively, the statistical module 250 may determine that the waste sorting robot 100 e.g. the suction gripper 120 is unable to achieve a specified performance. As the statistical module 250 analysis the parameters of the suction gripper 120 and the waste sorting robot 100 over a greater number n of operations, the parameters determined by the statistical module 250 become more reliable metrics of performance of the waste sorting robot 100.

In some examples, the statistical module 250 is configured to compare the maximum vacuum pressure p _v max and the highest maximum vacuum pressure p _v high_max. The determined difference between the maximum vacuum pressure p _{v m} ax and the highest maximum vacuum pressure p _V high_max can be an indicator of the operational performance of the waste sorting robot 100. In particular, if the difference between maximum vacuum pressure p _{v m} ax and the highest maximum vacuum pressure p _V high_max is increasing, then this is an indicator that the performance of the suction gripper 100 is worsening. This can be an indication that there is a fault in the suction gripper 120.

Similarly, a falling highest maximum vacuum pressure p _V high_max can also be an indication that there is a fault in the suction gripper 120.

Figure 6c shows the minimum air supply pressure p _aS min over time. In some examples, the minimum air supply pressure p _{as m} in is the instantaneous air supply pressure detected in the suction gripper 120, the first air hose 202 or another component in the pneumatic system 222 suppling the compressed air to the suction gripper 120. In other examples, the minimum air supply pressure p _as min is a minimum air supply pressure moving average p _aS min. The minimum air supply pressure moving average p _aS min over the previous n gripping operations is calculated as follows: n

Pas min -1V / Pa ¹ s min i = l

In some examples, the statistical module 250 determines the minimum air supply pressure moving average pas min. In some examples, the controller 200 sends a signal to the statistical module 250 to change the number n in order to increase the accuracy of pas min or decrease n to increase the sensitivity of pas min.

In some other examples, instead of determining the minimum air supply pressure moving average pas min the statistical module 250 determines the lowest minimum air supply pressure Pas iow_min. The statistical module 250, determines the lowest minimum air supply pressure p _as iow_min. for n consecutive suction gripper operations. In some examples, the lowest minimum air supply pressure p _a s iow_min is used instead of minimum air supply pressure moving average Pas min. This is because the lowest minimum air supply pressure p _as iow_min may change more rapidly during operation and changes in the air supply pressure may be easier to detect. By analyzing e.g. 10, 100 or 1000 gripping operations, the natural variation in the waste objects can be reliably filtered out.

As can be seen from Figure 6c, in normal operation there is limited variation in the instantaneous minimum air supply pressure Pas min. For the purposes of clarity, the minimum air supply pressure moving average pas min has not been plotted on Figure 6c.

In normal operation the instantaneous minimum air supply pressure p _a s min is generally above a predetermined threshold. In normal operation the minimum air supply pressure moving average p _as min is above a predetermined threshold. The normal pas min range 608 is shown by rectangle which represents a range between 5 to 7 bar. A below normal pas min range 610 is shown by rectangle which represents 4 to 5 bar. In some examples, if the minimum air supply pressure moving average pas min remains in or lower than the below normal pas min range 610, then then the statistical module 250 sends a signal to the controller 200. This can indicate a fault with the waste sorting robot 100 or the suction gripper 120 and the controller 200 can generate an alert to the operator. In some examples, the normal pas min range 608 can be varied to any other suitable ranges or combination thereof e.g. between 6 to 8 bar, 7 to 9 bar, 8 to 10 bar. In some examples, the below normal pas min range 610 can be varied to any other suitable ranges or combination thereof e.g. between 5 to 6 bar, 5 to 7 bar, 6 to 8 bar etc. Use of the below normal p _aS min range 510 is optional and in other examples, the statistical module 250 sends a signal to the controller 200 when the minimum air supply pressure moving average p _as min falls below and indicates a fault with the suction gripper 120 and I or the waste sorting robot 100.

Turning to Figures 7a, 7b, 7c another scenario will now be discussed. Figures 7a, 7b, 79c show graphs of different parameters of the waste sorting robot 100 in according to a fault scenario.

Another problem that may occur during sorting waste objects 106 is that the suction gripper 120 and other components of the pneumatic system 222 acquire a buildup of dirt and sticky residue on their internal surfaces. This can be due to organic matter present on the waste object 106. Excessive build-up of dirt and sticky residue will reduce the airflow and ultimately block the airflow.

Here the highest maximum vacuum pressure p _V high_max will decrease into and then lower than the below normal p _V high_max range 606 as shown by curve 702 and at the same and the % gripping rate R of the gripping operations will decrease as shown by curve 700 into or lower than the below normal R range 602. The decrease may be gradual and not detectable over a short time period. Therefore, a long-term moving average for one or more parameters of the waste sorting robot 100 and the suction gripper 120 may be required for determining that there is fault.

Operation of the controller 200 and the statistical module 250 will now be discussed in reference to Figure 8 when the highest maximum vacuum pressure p _v high_max and the % gripping rate R are determined to be outside normal operating parameters. Figure 8 shows a flow diagram of operation of the waste sorting robot 100 and fault detection.

The waste sorting robot 100 starts at letter “A” and proceed to function in a normal mode of operation as shown in step 1000. Periodically, the statistical module 250 determines the % gripping rate R of the gripping operations as shown in step 1002. Step 1002 may be carried out after every gripping operation so that the % gripping rate R is kept current. In some other examples the statistical module 250 determines the % gripping rate R of the gripping operations by sampling a number of gripping operations and extrapolating the % gripping rate R from the sample. The statistical module 250 determines in step 1004 whether the % gripping rate R of the gripping operations is within the normal R range 600. If the statistical module 250 determines that the % gripping rate R is normal, then the controller 200 determines that the waste sorting robot 100 is operating normally and returns to step 1000. However, as discussed above, when the % gripping rate R is normal, there still may be a fault in the suction gripper 120. This means the statistical module 250 may perform steps 1006, 1008, 1010 and 1012 as discussed below. The dotted arrow 1018 indicates that the statistical module 250 performs other steps before returning to step 1000. In some examples, step 1004 is always performed before steps 1006, 1008, 1010 and 1012.

If the statistical module 250 determines that the % gripping rate R of the gripping operations is below the normal R range 600 or lower than the below normal R range 602, then the statistical module 250 determines the current instantaneous maximum vacuum pressure p _vm ax in step 1006 and the current instantaneous minimum air supply pressure p _{as m} in in step 1008. In some examples, step 1004 is carried out before step 1006 and step 1008. It may be preferable for the statistical module 250 to perform step 1004 first because if the % gripping rate R is within the normal range 500, then it is likely that there are no faults with the waste sorting robot 100. However, in other examples, step 1004 can be carried out in parallel with steps 1006 and 1008.

In a less preferred examples, the statistical module 250 can omit step 1004. This is because a poor highest maximum vacuum pressure p _v high_max or the minimum air supply pressure moving average p _as min will cause a drop in the % gripping rate R of the gripping operations. However, not determining the % gripping rate R of the gripping operations, the detection of faults in the suction gripper 120 will be less successful.

The statistical module 250 then determines whether the highest maximum vacuum pressure p _v high_max is within or lower than the below normal p _V high_max range 606 as shown in step 1010. As the same time the statistical module 250 then determines whether the minimum air supply pressure moving average p _as min is within or lower than the below normal p _aS min range 610 as shown in step 1012.

The controller 200 then determines in step 1014 that there is a fault in the suction gripper 120 if the % gripping rate R of the gripping operations and the highest maximum vacuum pressure p _v high_max or the minimum air supply pressure moving average p _as min are outside operational parameters. In some examples the statistical module 250 may only perform step 1010 or step 1012. In other words, the statistical module 250 only analyses the highest maximum vacuum pressure Pv high_max or the minimum air supply pressure moving average pas min . This may be desirable if only a single type of fault is needed to be detected.

In some examples, the statistical module 250 optionally classifies the fault determined in step 1014 and the statistical module 250 optionally sends information of the probable fault to the controller 200. For example, the statistical module 250 uses stored information to determine what type of fault is experienced by the waste sorting robot 100. In this way, the statistical module 250 uses the characteristics of the % gripping rate R of the gripping operations, the highest maximum vacuum pressure p _V high_max, and the minimum air supply pressure moving average pas min to identify the type of fault. In some examples, the controller 200 is configured to determine the type of fault instead of the statistical module 250.

In some examples, the statistical module 250 determines that the % gripping rate R of the gripping operations and the highest maximum vacuum pressure p _V high_max, have gradually dropped whilst the minimum air supply pressure moving average p _as min has remained constant. The statistical module 250 determines that the rate of change of the % gripping rate R of the gripping operations and the highest maximum vacuum pressure p _V high_max, and the constant air supply pressure moving average pas min corresponds to a fault caused by sticky residue building up in the suction gripper 120.

Accordingly, the statistical module 250 sends a signal to the controller 200 indicating that the statistical module 250 has determined that there is a fault. The controller 200 then optionally issues an alert or alarm to the operator as shown in step 1016. In some examples, the controller 200 includes the probable fault in the alert that the suction gripper 120 has become partially clogged with the sticky residue. In some examples, the alert can be a message issued on a control panel (not shown).

In some other examples, the statistical module 250 determines that highest maximum vacuum pressure p _V high_max is below the operational specification maximum vacuum pressure pv max_spec- This means that the waste sorting robot 100 and the suction gripper 120 are not operating within the required specification. As mentioned previously, if the suction gripper 120 generates a highest maximum vacuum pressure p _V high_max which is too low, this can affect the operational performance of the waste sorting robot 100. Similarly, the statistical module 250 sends a signal to the controller 200 indicating that the statistical module 250 has determined that there is a fault because the highest maximum vacuum pressure p _V high_max is too low. The controller 200 then optionally issues an alert or alarm to the operator as shown in step 1016. Additionally or alternatively, the controller 200 carries out the steps as described in Figure 9 with starts at “B”. Figure 9 shows a flow diagram view of a method used by the waste sorting robot 100 to clean the suction gripper 120 following a determination that the suction gripper 120 is clogged with sticky residue.

As mentioned above, the statistical module 250 sends a signal to the controller 200 and this triggers the solvent dosing operation for cleaning the suction gripper 120 as shown in step 1018.

When the controller 200 initiates the solvent dosing operation, the controller 200 sends a control instruction to actuate the solvent valve 216 according to step 1020. The solvent outlet 214 then does the airflow in the suction gripper 120 with the solvent according to step 1022 as discussed in reference to the previous examples. The waste sorting robot 100 then returns to “A” and proceeds to function in the normal operation step 1000.

Whilst the controller 200 initiates the solvent dosing operation, the controller 200 can modify the flow rate of the solvent at the solvent outlet 214 and modify the duration of the solvent dosing operation. The controller 200 can modify the solvent dosing operation in dependence of one of more parameters of the waste sorting robot 100. For example, the statistical module 250 can send a signal to the controller 200 that the rate of change of the % gripping rate R of the gripping operations, the highest maximum vacuum pressure p _V high_max is decreasing rapidly or slowly. Accordingly, the controller 200 can change the amount of solvent used in the solvent dosing operation in dependence on the rate of change of the % gripping rate R of the gripping operations and I or the maximum highest maximum vacuum pressure p _v high_max e.g. an indication of how fast the sticky residue is building up on the suction gripper 120.

For example, if the waste objects 106 are particularly dirty, the controller 200 can perform the solvent dosing operation in response to the level of contamination that the suction gripper 120 is experiencing.

By periodically introducing solvent into the airflow of the suction gripper 120, the build-up of dirt and debris can be reduced. This means that the waste sorting robot 100 requires less maintenance and is more efficient. By determining the more parameters of the suction gripper 120 over the plurality of suction gripper operations. The controller 200 can detect a fault in the suction gripper 120 and automatically clean the suction gripper 120 with the solvent. In this way partial information on how picks are actually succeeding can be used to measure the performance of the waste sorting robot 100. A signal (e.g. % gripping rate R of the gripping operations) which correlates strongly with successful picks is used to determine the performance of the waste sorting robot 100. However the signal relating to the performance of the waste sorting robot 100 is prone to faults. The inventors have realized that by applying their experience and external knowledge, the signal can be effectively used through statistical analysis for determining the performance of the waste sorting robot 100. The inventors have realized that the suction gripper 120 will achieve a good grip on at least some of the some of the objects and over time, it's virtually guaranteed that such objects will be sorted by the suction gripper 120.

In another example two or more examples are combined. Features of one example can be combined with features of other examples.

Examples of the present disclosure have been discussed with particular reference to the examples illustrated. However it will be appreciated that variations and modifications may be made to the examples described within the scope of the disclosure.