EP4368093A1 - Method for operating a cleaning device, method for training a neural network, and cleaning device - Google Patents

Abstract

The invention relates to a method for operating a cleaning device (100). The method comprises a step of reading in a door closing signal, a step of providing a control signal and a control step. The door closing signal represents a command for controlling the closing of a door (110) of the cleaning device (100). The control signal is provided for a motor (220) for driving a door (110) of the cleaning device (100), wherein the control signal is provided using a camera signal and a motor current signal. The camera signal is read in by a camera (125) for optically monitoring an interior (115) of the cleaning device (100) and/or a loading area of the cleaning device (100). The motor current signal is read in by a current sensor (235) of the motor (220). The control step is carried out in order to control the motor (220) using the control signal.

Description

The invention relates to a method for operating a cleaning device, a method for training a neural network and a cleaning device.

Household appliances such as dishwashers and ovens have a door or flap for loading, which can usually be opened and closed manually. Automatic door closing and opening devices for household appliances are also known. The opening or closing process can be triggered by activating an actuating device in the form of a sensor. The closing or opening process is triggered deliberately, which can require an additional handling step.

The EP 4 026 474 A1 discloses a method for operating a dishwasher with a motor-driven appliance door, in which the movement of the appliance door is monitored and the motor device is switched off when a blockage is detected.

The invention has for its object to provide an improved method for operating a cleaning device, an improved method for training a neural network and an improved cleaning device.

According to the invention, this object is achieved by a method for operating a cleaning device, a method for training a neural network and by a cleaning device having the features of the main claims. Advantageous embodiments and further developments of the invention emerge from the following subclaims.

The advantages that can be achieved with the invention are that a door of a cleaning device can close automatically in a safe and easy manner, which can increase user satisfaction.

A method for operating a cleaning device comprises a step of reading a door closing signal, a step of providing a control signal and a step of controlling. The door closing signal represents a command for controlling a closing of a door of the cleaning device. The control signal is provided for a motor for driving a door of the cleaning device, wherein the control signal is provided using a camera signal and a motor current signal. The Camera signal is read from a camera for optical monitoring of an interior of the cleaning device and/or a loading area of the cleaning device. The motor current signal is read from a current sensor of the motor. The step of controlling is carried out to control the motor using the control signal.

The cleaning device can be designed as a dishwasher. The camera can be arranged inside the door, for example, in order to be able to advantageously capture the interior and the loading area. The current sensor is arranged, for example, next to or in the motor and is designed, for example, to detect a current flow through the windings of the motor, which is designed as an electric motor. No additional actuators and sensors are required for the automatic insertion of the baskets, which means that additional costs and susceptibility to failure can be avoided. The motor can therefore be manufactured inexpensively and designed to be low-failure. Furthermore, only one drive is required to drive the door, which means that costs and installation space can be saved.

The approach presented here can be understood as a combined door and basket automation for a dishwasher. In other words, an automatic door closing system is disclosed that can, for example, automatically adapt to the freedom of movement of the baskets and push the baskets into the washing compartment if this is possible and can stop if the basket is too heavily loaded or is jammed in its guide. This makes it possible to ensure that the door is closed safely without supervision and to enable the door closing process to be adapted to the mobility of the baskets.

The door can be closed via remote start even if the baskets are pulled out. The door can be closed safely, taking into account typical operating states, for example "basket is pulled out". Any stiffness in the basket guides can be detected and compensated if necessary. If this is not possible, the closing process can be aborted to prevent damage to the components. The existing sensors can be used largely for the algorithm.

The step of providing can be carried out using a door sensor signal. The door sensor signal can be read by a door sensor. The door sensor signal can represent a closing angle of the door and additionally or alternatively a vibration that has occurred in the door. The door sensor can be designed as a vibration sensor that can detect vibrations in the area of the door. These vibrations can be caused by the retraction and extension of dish baskets and/or by the opening and closing of a dispensing compartment. The vibration sensor can be designed as a Acceleration sensor or as a microphone, wherein the acceleration sensor can be designed, for example, as a three-axis acceleration sensor.

In the step of providing, the door sensor signal can represent an object that can be located in a loading area and/or inside the cleaning device. The object can be, for example, a hand or generally a body part of an operator of the cleaning device or an animal. If the object is detected, the closing process can be stopped in order to avoid an accident, for example a hand being trapped.

In the control step, the control signal can be output in such a way that the motor can stop a closing process of the door, that the motor can be controlled at a lower speed and/or that the motor can exert a higher torque on the door. This enables the cleaning device to function reliably and safely.

During the control step, the motor can stop the door from closing or apply a higher torque to the door (compared to a previous movement) if a collision between the door and a load carrier is detected. In this way, safe operation of the cleaning device can be ensured.

In the control step, the motor can be controlled in such a way that the door is moved at a lower or reduced speed (compared to a previous movement) when the motor current exceeds a threshold. In this way, safe operation of the cleaning device can also be ensured.

In the reading step, the door closing signal can be read from an interface to a unit external to the cleaning device. This makes it possible to close the door of the cleaning device remotely, for example using a mobile device.

The steps of reading, providing and/or controlling can be carried out at least partially using an artificial intelligence unit, in particular a neural network. Advantageously, the automatic closing process can thus be carried out reliably and quickly.

A method for training a neural network for use in an embodiment of a method for operating a cleaning device mentioned herein comprises a step of reading in and a step of training. In the step of reading in, the door closing signal, the control signal, the camera signal and the Motor current signal is read in. In the training step, the neural network is trained using the door closing signal, the control signal, the camera signal and the motor current signal to control a closing process of the door of the cleaning device.

The approach presented here also creates a control device that is designed to carry out, control or implement the steps of a variant of one of the methods presented here in corresponding devices. This embodiment of the invention in the form of a control device can also solve the problem underlying the invention quickly and efficiently.

The control device can be designed to read in input signals and to determine and provide output signals using the input signals. An input signal can, for example, represent a sensor signal that can be read in via an input interface of the control device. An output signal can represent a control signal or a data signal that can be provided at an output interface of the control device. The control device can be designed to determine the output signals using a processing rule implemented in hardware or software. For example, the control device can comprise a logic circuit, an integrated circuit or a software module and can be implemented, for example, as a discrete component or be comprised of a discrete component.

A cleaning device has an embodiment of a control device mentioned herein. The cleaning device can be designed as a dishwasher, for example.

Also advantageous is a computer program product or computer program with program code that can be stored on a machine-readable carrier or storage medium such as a semiconductor memory, a hard disk memory or an optical memory. If the program product or program is executed on a computer or a device, the program product or program can be used to carry out, implement and/or control the steps of the method according to one of the embodiments described above.

Although the approach described is based on a household appliance, the methods and cleaning device described here can be used accordingly in connection with a commercial or professional device, for example a medical device, such as a cleaning or disinfection device, a small sterilizer, a large-scale disinfector or a container washing system.

Figur 1

eine Darstellung eines Ausführungsbeispiels eines Reinigungsgeräts;

Figur 2

eine Darstellung eines Ausführungsbeispiels eines Reinigungsgeräts;

Figur 3

eine Darstellung eines Ausführungsbeispiels eines Reinigungsgeräts;

Figur 4

ein Blockschaltbild zur Erläuterung eines Ausführungsbeispiels eines Verfahrens zum Trainieren eines neuronalen Netzes;

Figur 5

eine Darstellung eines Ausführungsbeispiels eines Reinigungsgeräts;

Figur 6

ein Diagramm zur Erläuterung eines Ausführungsbeispiels eines Verfahrens zum Betreiben eines Reinigungsgeräts;

Figur 7

ein Diagramm zur Erläuterung eines Ausführungsbeispiels eines Verfahrens zum Betreiben eines Reinigungsgeräts;

Figur 8

ein Diagramm zur Erläuterung eines Ausführungsbeispiels eines Verfahrens zum Betreiben eines Reinigungsgeräts;

Figur 9

eine Darstellung eines Ausführungsbeispiels eines Reinigungsgeräts;

Figur 10

ein Ablaufdiagramm eines Ausführungsbeispiels eines Verfahrens zum Betreiben eines Reinigungsgeräts;

Figur 11

ein Ablaufdiagramm eines Ausführungsbeispiels eines Verfahrens zum Betreiben eines Reinigungsgeräts;

Figur 12

ein Ablaufdiagramm eines Ausführungsbeispiels eines Verfahrens zum Betreiben eines Reinigungsgeräts;

Figur 13

ein Diagramm zur Erläuterung eines Ausführungsbeispiels eines Verfahrens zum Betreiben eines Reinigungsgeräts;

Figur 14

ein Diagramm zur Erläuterung eines Ausführungsbeispiels eines Verfahrens zum Betreiben eines Reinigungsgeräts;

Figur 15

ein Ablaufdiagramm eines Ausführungsbeispiels eines Verfahrens zum Betreiben eines Reinigungsgeräts;

Figur 16

ein Ablaufdiagramm eines Ausführungsbeispiels eines Verfahrens zum Trainieren eines neuronalen Netzes;

Figur 17

ein Blockschaltbild eines Ausführungsbeispiels einer Steuervorrichtung zum Betreiben eines Reinigungsgeräts; und

Figur 18

ein Blockschaltbild eines Ausführungsbeispiels einer Steuervorrichtung zum Trainieren eines neuronalen Netzes.

An embodiment of the invention is shown purely schematically in the drawings and is described in more detail below. It shows

Figure 1

a representation of an embodiment of a cleaning device;

Figure 2

a representation of an embodiment of a cleaning device;

Figure 3

a representation of an embodiment of a cleaning device;

Figure 4

a block diagram for explaining an embodiment of a method for training a neural network;

Figure 5

a representation of an embodiment of a cleaning device;

Figure 6

a diagram for explaining an embodiment of a method for operating a cleaning device;

Figure 7

a diagram for explaining an embodiment of a method for operating a cleaning device;

Figure 8

a diagram for explaining an embodiment of a method for operating a cleaning device;

Figure 9

a representation of an embodiment of a cleaning device;

Figure 10

a flow chart of an embodiment of a method for operating a cleaning device;

Figure 11

a flow chart of an embodiment of a method for operating a cleaning device;

Figure 12

a flow chart of an embodiment of a method for operating a cleaning device;

Figure 13

a diagram for explaining an embodiment of a method for operating a cleaning device;

Figure 14

a diagram for explaining an embodiment of a method for operating a cleaning device;

Figure 15

a flow chart of an embodiment of a method for operating a cleaning device;

Figure 16

a flowchart of an embodiment of a method for training a neural network;

Figure 17

a block diagram of an embodiment of a control device for operating a cleaning device; and

Figure 18

a block diagram of an embodiment of a control device for training a neural network.

Figure 1 shows an illustration of an embodiment of a cleaning device 100. The cleaning device 100 is designed, for example, as a dishwasher and is installed, for example, as a built-in appliance in a kitchen unit below a worktop 105.

The cleaning device has a door 110 for closing a washing interior 115. Within the washing interior 115, for example, a lighting device 120 is arranged, which is designed to illuminate the washing interior 115.

According to one embodiment, the door 110 has a camera 125 that is directed towards the interior of the dishwasher 115 and is designed to detect a loading state. For this purpose, for example, the lighting device 120 is activated, which illuminates the interior of the dishwasher 115 and the camera 125 detects the loading state.

Furthermore, the door 110 has, for example, a handle 130 with which a user can open the door 110 of the cleaning device 100. According to one embodiment, the cleaning device 100 has a motorized door opening 135 including a position sensor, which is designed to open the door 110 automatically.

Within the door 110, for example, a control unit 140 for controlling the door opening 135, the camera 125 and the lighting device 120 is arranged.

The control unit 140 is further designed to control a turbidity sensor 145, which is arranged in the region of a collecting pot 150 below the interior of the washing chamber 115. The turbidity sensor 145 is designed, for example, to detect a turbidity of a cleaning fluid and, in response, to initiate emptying of the collecting pot 150 when the turbidity exceeds a predetermined limit value.

Figure 2 shows an illustration of an embodiment of a cleaning device 100. The cleaning device 100 is designed, for example, as a dishwasher.

Similar to the cleaning device in Figure 1 The cleaning device 100 shown here has the camera 125, the control unit 140 inside the door 110 and the handle 130 on the door 110. The lighting device 120 is also arranged inside the interior of the washing chamber 115.

The cleaning device 100 additionally has, for example, a plurality of baskets 205, 210, 215. More specifically, an upper basket 210, a lower basket 215 and a cutlery drawer 205, wherein the upper basket 210 is arranged between the lower basket 215 and the cutlery drawer 205.

In addition, the cleaning device 100 shown here has a dosing device 200 in the door 110, which is designed, for example, to hold a cleaning agent. The dosing device 200 is connected, for example, to the control unit 140 in a way that is capable of transmitting signals. According to one embodiment, a knock sensor 230 is arranged within the control unit 140.

The cleaning device 100 has, for example, a motor 220 below the lower basket 215, which can also be referred to as a door closing device 220, which is connected to the door 110 in order to open or close it when a corresponding signal is detected by a cleaning device sensor 225, which can also be referred to as a door sensor, above the cutlery drawer 205. The motor 220 is also connected to a current sensor 235, which is designed to drive the motor 220.

In other words, the cleaning device 100 has the interior of the washing machine 115, which can also be referred to as the washing chamber, and which can be closed with the door 110 and opened using the door handle 130. The cleaning device 100 is usually installed as a built-in device in the kitchen unit below the worktop 105. The control unit 140, which can also be referred to as the device control, is arranged inside the door 110 and controls the actuators (not shown here) of the cleaning device 100, as well as the dosing device 200. The camera 125, which takes pictures of the interior of the washing machine 115, is arranged inside the door 110 on the washing chamber side. To support the image recording, the lighting device 120 illuminates the interior of the washing machine 115. A basket system is arranged inside the cleaning device 100, consisting of the cutlery drawer 205, the upper basket 210 and the lower basket 215 for holding items to be washed. The motor 220 is provided for opening and/or closing the door 110. The door closing process is triggered, for example, by the cleaning device sensor 225 arranged in the area of the dishwasher interior 115 or via a mobile phone.

The door 110 is opened, for example, by activating the knock sensor 230 located in the door 110. Knocking on the door 110 activates an automatic door opening mechanism.

Figure 3 shows a representation of an embodiment of a cleaning device 100. In contrast to the previous figures, the cleaning device 100 is shown open, the door 110 is completely open. Similar to the cleaning device in Figure 2 the cleaning device 100 shown here has the cutlery drawer 205, the upper basket 210 and the lower basket 215, with the lower basket 215 shown pulled out and arranged completely on the inside of the opened door 110. The lower basket 215 has, for example, two rollers 300, 305 on which the lower basket 215 is movably mounted. In the exemplary embodiment shown here, items to be washed 315 are shown as an example, which are arranged in the lower basket 215. An arrow 320 shows, for example, a direction in which the lower basket 215 can be pushed in.

The door 110 of the cleaning device 100 is connected, for example, to the motor 220, wherein the motor 220 can be driven by the current sensor 235. The motor 220 is designed to to drive the door 110 to close it in response to a command to close the door 110. The command may be issued, for example, by a user.

According to the embodiment shown here, the lower basket 215 is pulled out. If the command to close the door 110 is issued in this state, the door 110 is driven in such a way that the lower basket 215 is automatically pushed into the interior of the dishwasher 115 during the closing process. An arrow 330 shows, for example, a door closing angle α. A door sensor 325 is arranged, for example, in the door 110 and is designed to detect vibrations of the door 110.

In other words, the approach presented here uses a neural network to evaluate several process variables of the motor 220, which can also be referred to as automatic door closing, in particular a load-dependent motor parameter, and the camera 125 to ensure that the door closes securely when the baskets 205, 210, 215 are pushed in, in particular when started remotely or without supervision by the user. The door sensor 325, which can also be referred to as a vibration sensor, can be designed as a three-axis acceleration sensor or as a microphone. The load-dependent motor parameter can be, for example, the motor current, the speed or the slip.

The Figure 3 illustrates the structure of the cleaning device 100 with the baskets 205, 210, 215, the camera 125 and the door sensor 325, which can also be referred to as a knock sensor, for example, and is located in the control unit 140 or the door 110. At least one lighting device 120 is located within the washing chamber 115. The camera 125 is used to record the states of the baskets 205, 210, 215, i.e. whether they are outside the washing chamber 115 and, if so, how far away.

The door sensor 325 detects not only load-specific information of the baskets 205, 210, 215 caused, for example, by the insertion of wash ware 315, which can also be referred to as a piece of crockery, but also the mechanical vibrations, structure-borne noise, which arise when the lower basket 215 is moved back and forth on the appliance door 110 or when the lower basket 215 hits a rear wall 335 of the wash chamber 115 when rolling into the wash chamber 115 and the door 110 hits an extended 205, 210, 215 when closing automatically. The door sensor 325 also detects the closing angle α of the door 110.

In order to determine the status of the baskets 205, 210, 215, which is characterized by the loading state and the extension distance, before starting and during the door closing process, the camera 125, the door sensor 325 and the current sensor 235 communicate with a neural network, which in the following Figure 4 shown and described.

Figure 4 shows a block diagram to explain an embodiment of a method for training a neural network 400.

According to one embodiment, the neural network 400 reads camera signals 405 from the camera, motor current signals 425 from the current sensor and vibration signals 410 from the vibration sensor and processes them. The neural network 400, which can also be referred to as a classifier, then outputs an information signal 415 to the control unit 140, which can also be referred to as a device controller. The control unit 140 then outputs a control signal 420 to the lighting device 120 as an example.

In other words, the neural network 400 is fed by the camera, the vibration sensor and the current sensor, whereby the neural network 400 records, assesses and evaluates information specific to the load and extension width and transmits the information about the loading state and the extension width to the control unit 140. The neural network 400 also contains information about the characteristic vibration or sound pattern that occurs when the door collides with one of the three baskets when closing or when the baskets hit the rear wall when being pushed in. In addition, the neural network 400 uses the value of the current sensor to determine whether the basket may have become jammed in its guide when it collided with the door or whether the operator may have interfered with the closing process with his hand when closing. In a training environment, for example a laboratory, an experimental kitchen, test households in the field, etc., the neural network 400 is trained to recognize the states using the images from the camera, the signals 410, 425 from the vibration sensor and the current sensor. The processing of several different signals with a single neural network, as in this example, is also referred to as multimodal learning.

Figure 5 shows a representation of an embodiment of a cleaning device 100. The cleaning device 100 is similar or corresponds to the cleaning device from Figure 3 . Figure 5 shows a possible constellation that the upper basket 210 is still pulled out when the door closing process was initiated, for example, by a remote start. The signal sizes that can be evaluated show the following Figures 6 to 8 . An object 500 is shown purely as an example, which is located in the loading area of the door 110. The object 500 is designed as a hand and triggers a stop of the closing process when it is detected in the loading area of the door 110.

Figure 6 shows a diagram to explain an embodiment of a method for operating a cleaning device. This can be the cleaning device described in the previous figures. If the door hits the upper basket, the collision is detected by the vibration sensor as a vibration pattern or signal bundle 600.

Figure 7 shows a diagram to explain an embodiment of a method for operating a cleaning device. This can be the cleaning device described in the previous figures.

From the fully open position, the door can initially close unhindered, which is preferably characterized by a linearly increasing angle signal α 700. Using a control system not shown in detail, the voltage on the motor is now preferably increased, which allows the motor to deliver a higher torque to push the upper basket into the wash cabinet. If the closing angle α 700 then does not increase, there may be a stiffness in the basket guide system and the door closing process is stopped. If the upper basket can be pushed into the wash cabinet, the vibration sensor detects the collision of the upper basket with the rear wall of the wash cabinet. The door is then completely closed and the closing process is finished.

Figure 8 shows a diagram to explain an embodiment of a method for operating a cleaning device. This can be the cleaning device described in the previous figures. The diagram shows an increase in the current consumption 800 of the drive of the door closing device, since the resistance to movement of the door increases due to the extended upper basket.

Figure 9 shows a representation of an embodiment of a cleaning device 100. The lower basket 215 is pulled out as an example. If the door closing process starts in this state, the control system (not described in detail) preferably increases the voltage of the motor 220 until the door 110 initiates the rolling of the lower basket 215 into the washing chamber 115. As the closing angle α increases, the lower basket 215 then slowly rolls into the washing chamber 115. The door sensor 325 detects the collision of the lower basket 215 with the rear wall 335 of the washing chamber 115, the lower basket 215 is in the washing chamber. The advantageous algorithm according to the invention or its essential steps for closing the door 110 is shown in the following Figure 10 .

Figure 10 shows a flow chart of an embodiment of a method for operating a cleaning device. The method can be carried out, for example, in connection with a cleaning device as described with reference to the previous figures.

A step 1000 marks a state in which the cleaning device is ready for operation and the door is open. Then, in a step 1005, the camera determines a basket status, for example the loading status and/or an extension width. Then, in a step 1010, it is checked whether the lower basket is extended. If this is the case, the system jumps to a step 1015, in which the door drive is switched on and the Current sensing is queried. If necessary, the torque is also increased or the process is aborted. The program then jumps to step 1020, which checks whether the lower basket has been retracted. If this is not the case, the program jumps back to step 1015. If the lower basket has been retracted, the program jumps to step 1025, which checks whether a collision with an upper basket has occurred. If it is previously detected in step 1010 that the lower basket has not been extended, the program jumps to step 1030, which switches on the door drive and queries the door sensor. The program then jumps directly to step 1025. If a collision with the upper basket is detected in step 1025, the program jumps to step 1035. Step 1035 queries the current sensing, if necessary, the torque is increased or the process is aborted. The program then checks in step 1040 whether the upper basket has been retracted. If this is not the case, the program jumps back to step 1035. If the upper basket has been retracted, the program jumps to step 1045, which checks for a collision with a cutlery drawer. If it is previously detected in step 1025 that there is no collision with the upper basket, the door drive is switched on in step 1050 and the vibration sensor is checked. The program then jumps directly to step 1045. If a collision with the cutlery drawer is detected in step 1045, the program jumps to step 1055, which checks the current sensing, increases the torque if necessary or aborts it and closes the door. The program then jumps to step 1060, which represents the end of the closing process. If no collision with the cutlery drawer is previously detected in step 1045, the door is closed in step 1065 and then the program jumps to step 1060.

In other words, at the start of the process, the neural network in conjunction with the camera determines the position of the baskets. The camera detects how far the baskets are outside the washing chamber. In addition, the camera images provide information about the degree of loading of the baskets. If the lower basket is extended, step 1010, the door closing process is started and the current sensor is queried, step 1015. The level of the current shows, for example, how heavily the basket is loaded. If the basket does not start to roll into the washing chamber, step 1020, the control increases the torque of the door closing device. If the lower basket has already been pushed in, step 1020, the door closing device is also switched on and the door is tilted until the vibration sensor detects the collision with the possibly pulled out upper basket or the cutlery drawer, step 1025, step 1045. If, for example, the upper basket cannot be pushed into the washing cabinet even with a higher torque, step 1035, the upper basket may have become jammed in its basket guide and the door closing process is aborted, step 1035. The door sensor always detects whether the respective basket is completely in the washing cabinet by means of the collision of the respective basket with the rear wall of the washing cabinet.

The approach presented here advantageously creates the possibility of ensuring automatic closing of the door, for example by remote start, of pushing any baskets that are still pulled out into the washing compartment with the closing door, or of aborting the closing process if, for example, one of the baskets has become jammed in its basket guide.

In other words, the camera image created in step 1005 records the loading status of the baskets and the cutlery drawer. If the camera in conjunction with the neural network detects a high loading level, a higher start current setpoint is specified for the control system, since a higher torque is required or expected to push in the baskets. If the door sensor detects a collision between the door and the baskets, the motor current control is switched on with the start current setpoint I _start , triggering, and the current closing angle α is determined by the door sensor. During the switch-on phase of the motor current control, the neural network or the device control uses the door sensor to determine the angular velocity Δα/Δt of the device door. If this exceeds the limit value 1 (Δα/Δt > w1), no further increase in the current setpoint is necessary. If the angular velocity does not change or is low or falls below limit value 2 (Δα/Δt < ω2), the current setpoint is increased to limit value 3 (I _max ). If limit value 3 is reached, the closing process is aborted. A system error can be assumed, for example the basket may be jammed.

The measurement of the angular velocity ω and the motor current I is necessary in order to reliably record the level of resistance in the system, i.e. the load on the baskets, stiffness, and tilting of the baskets. This would not be possible using an imaging process alone. This is especially true since it is difficult or even impossible to determine the weight using an image.

If all parameters of the closing movement - that is, I, w and "Image" - are within the nominal range, all baskets are automatically pushed into the washing chamber when the door is closed. If all baskets have already been pushed in, for example by the operator himself, the door is automatically closed without the algorithm described above.

Figure 11 shows a flow chart of an embodiment of a method for operating a cleaning device. The method can be carried out, for example, in connection with a cleaning device as described with reference to the previous figures.

Step 1100 marks a state in which the cleaning device is ready for operation and, for example, the door is open. In a step 1105, it is checked whether a remote start is activated. If this is not the case, the program jumps back to step 1100. If remote start is activated, the program jumps to step 1110, which switches on the automatic door closing system. A step 1115 then checks whether a collision is detected. If a collision is detected, the automatic door closing system is stopped in step 1120. And then the program jumps back to step 1100. If no collision is detected in step 1115, the program jumps to step 1125. There it checks whether the camera detects a hand or a pet. If so, the program jumps back to step 1120 and then to step 1100. If the camera does not detect a hand or pet in step 1125, a step 1130 checks whether the motor current exceeds a threshold. If this is the case, the automatic door closing is stopped in a step 1135, the speed is reduced and/or the closing of the door is continued. The system then jumps back to step 1100.

If it is detected in step 1130 that the motor current does not exceed a threshold, the system jumps to step 1140, which switches on the automatic door closing system and continues closing the door. The system then jumps back to step 1100.

Figure 12 shows a flow chart of an embodiment of a method for operating a cleaning device. The method can be carried out, for example, in connection with a cleaning device as described with reference to the previous figures.

Step 1200 marks a state in which the cleaning device has been started, or more precisely, the door is open. The program then jumps to step 1205, in which it is checked whether the vibration sensor detects the open door. If this is not the case, it jumps back to step 1200. If the vibration sensor detects the open door, it jumps to step 1210. In step 1210 it is checked whether the basket has been pulled out. If the basket has not been pulled out, it jumps to step 1215, which activates a time recording device, which can also be referred to as a timer. The duration of the time recording device is only 30 seconds, for example. After the time recording device has been activated, it jumps to step 1220. In step 1220 it is checked whether a vibration signal, which can also be referred to as a knocking signal, is detected. The vibration signal is generated, for example, when a dispensing compartment is operated. If a vibration signal has been detected, the system jumps to step 1225, which restarts the time recording device and then jumps back to step 1220. If no vibration signal is detected in step 1220, the system jumps to step 1230, which checks whether the 30 seconds have been reached. If the 30 seconds have not been reached, the system jumps back to step 1220. If, however, the 30 seconds have been reached, the system jumps to step 1235, which represents and executes automatic closing of the door. The system then jumps back to step 1200 and executes the process again, for example.

If an extended basket is registered in the aforementioned step 1210, the system jumps to step 1240, which starts the time recording device, which also has a time duration of 30 seconds, just as an example. The system then jumps to step 1245, which checks whether a vibration signal is detected. If the vibration signal is detected, the system jumps to step 1250, which restarts the time recording device. The system then jumps back to step 1245. If the vibration signal is not detected in step 1245, the system jumps to step 1255, which checks whether the 30 seconds have been reached. If the 30 seconds have not been reached, the system jumps back to step 1245. If, however, the 30 seconds have been reached, the system jumps to step 1260, which represents and carries out an automatic closing of the door. The system then jumps back to step 1200 and the process is carried out again, for example.

Figure 13 shows a plurality of temporal pulse curves to explain an embodiment of a method for operating a cleaning device.

The first pulse curve 1305 occurs, for example, when a basket is loaded with washware. The second pulse curve 1310 represents, for example, an unloading of the basket and the third pulse curve 1315 shows, for example, a partial loading or unloading pause.

Figure 14 shows a pulse curve to explain an embodiment of a method for operating a cleaning device. The pulse curve results, for example, from loading the lower basket and/or operating a dosing flap of the dosing device. The x-axis 1400 represents, for example, the temporal progression, the y-axis 1405 represents, for example, a progression or the amplitude of the pulse signal.

In other words, the loading events exhibit typical signal curves, which are shown as examples in the diagram.

Heavy crockery items have a higher signal amplitude A3, a lower frequency, i.e. a sound, and a longer signal duration Δt3 than light crockery items. Each signal peak represents a loading event, the sum of all signal peaks represents the current loading status of the lower basket, starting from the value zero after the program has ended.

In addition, the vibration sensor records handling signals on the dosing device. For example, opening the dosing flap of the dosing device leads to signal peaks similar to those Figure 14 shows.

Figure 15 shows a flow chart of an embodiment of a method 1500 for operating a cleaning device.

The method 1500 comprises a step 1505 of reading in a door closing signal, a step 1510 of providing a control signal and a step 1515 of controlling. The door closing signal represents a command for controlling the closing of a door of the cleaning device. According to one embodiment, the step 1505 of reading in the door closing signal is read in from an external interface. The control signal is provided for a motor for driving a door of the cleaning device, wherein the control signal is provided using a camera signal and a motor current signal. The step 1510 of providing is carried out, for example, using a door sensor signal.

The camera signal is read in by a camera for optically monitoring an interior of the cleaning device and/or a loading area of the cleaning device. The motor current signal is read in by a current sensor of the motor. The control step 1515 is carried out to control the motor using the control signal. In the control step 1515, the motor stops the closing process of the door, for example, or applies a higher torque to the door if a collision of the door with a load carrier is detected. Alternatively, in the control step 1515, the motor is controlled at a lower speed if the motor current exceeds a threshold.

The steps 1505, 1510, 1515 are carried out, for example, using an artificial intelligence unit, in particular a neural network.

Figure 16 shows a flow chart of an embodiment of a method for training a neural network for a method for operating a cleaning device, as described for example in Figure 15 described.

The method 1600 comprises a reading step 1605 and a training step 1610. In the reading step, the door closing signal, the control signal, the camera signal and the motor current signal are read. In the training step, the neural network is trained using the door closing signal, the control signal, the camera signal and the motor current signal in order to carry out a closing operation of the door of the cleaning device.

Figure 17 shows a block diagram of an embodiment of a control device 1700 for operating a cleaning device. The control device 1700 is designed to carry out the method of Figure 15 or to carry out a similar method. The control device 1700 has a unit 1705 for reading a door closing signal 1720, a unit 1710 for providing a control signal 1725 and a unit 1715 for controlling the motor. The unit 1710 for providing provides the control signal 1725, for example, using a door sensor signal 1730.

Figure 18 shows a block diagram of an embodiment of a control device 1800 for training a neural network. The control device 1800 is designed to implement the method of Figure 16 or a similar method. The control device 1800 has a unit 1805 for reading in the door closing signal 1720, the control signal 1725, the camera signal 405 and the motor current signal 425. Furthermore, the control device 1800 has a unit 1810 for training the neural network using the door closing signal 1720, the control signal 1725, the camera signal 405 and the motor current signal 425 in order to carry out a closing process of the door of the cleaning device.

Claims (13)

Method (1500) for operating a cleaning device (100), the method (1500) comprising the following steps: Reading (1505) a door closing signal (1720), wherein the door closing signal (1720) represents a command for controlling a closing of a door (110) of the cleaning device (100); Providing (1510) a control signal (1725) for a motor (220) for driving a door (110) of the cleaning device (100), wherein the control signal (1725) is provided using a camera signal (405) and a motor current signal (425), wherein the camera signal (405) is read in by a camera (125) for optically monitoring an interior (115) of the cleaning device (100) and/or a loading area of the cleaning device (100) and the motor current signal (425) is read in by a current sensor (235) of the motor (220); and Controlling (1515) the motor (220) using the control signal (1725). Method (1500) according to claim 1, wherein the step (1510) of providing is performed using a door sensor signal (1730), wherein the door sensor signal (1730) is read in by a door sensor (325) and represents a closing angle of the door (110) and/or a vibration occurring in the door (110). Method (1500) according to claim 2, wherein in the step of providing the door sensor signal (1730) represents an object (500) which is located in a loading area and/or in the interior area (115) of the cleaning device (100). Method (1500) according to one of the preceding claims, wherein in the step (1515) of controlling, the control signal (1725) is output such that the motor (220) stops a closing operation of the door (110), that the motor (220) is controlled at a lower speed and/or that the motor (220) exerts a higher torque on the door (110). Method (1500) according to claim 4, wherein in the step (1515) of controlling, the motor (220) stops the closing process of the door (110) or exerts a higher torque on the door (110) when a collision of the door (110) with a load carrier (205, 210, 215) is detected. Method (1500) according to claim 4, wherein in the step (1515) of controlling the motor (220) is controlled such that the door (110) is opened with a lower or reduced speed when the motor current exceeds a threshold. Method (1500) according to one of the preceding claims, wherein in the step (1505) of reading in, the door closing signal (1720) is read in from an interface to a unit external to the cleaning device. Method (1500) according to one of the preceding claims, wherein at least one of the steps of reading (1705), providing (1710), and/or controlling (1715) is carried out using an artificial intelligence unit, in particular a neural network (400). Method (1600) for training a neural network (400) for use in a method (1500) for operating a cleaning device (100) according to one of the preceding claims 1 to 8, wherein the method (1600) comprises the following steps: Reading (1605) the door closing signal (1720), the control signal (1725), the camera signal (405) and the motor current signal (425); and Training (1610) the neural network (400) using the door closing signal (1720), the control signal (1725), the camera signal (405) and the motor current signal (425) to control a closing operation of the door (110) of the cleaning device (100). Control device (1700; 1800) which is designed to carry out and/or control the steps of one of the methods (1500; 1600) according to one of claims 1 to 8 and/or according to claim 9 in corresponding units. Cleaning device (100) with a control device (1700; 1800) according to claim 10. Computer program product with program code for carrying out the method (1500; 1600) according to one of the preceding claims 1 to 8 and/or according to claim 9, when the computer program product is executed on a control device (1700; 1800) according to claim 10. A machine-readable storage medium on which the computer program according to claim 12 is stored.

Images