Control method, device and equipment for refrigerator door

Abstract

The application discloses a refrigerator door control method, device and equipment, a radar array is arranged on the refrigerator door, the method comprises the following steps: determining the motion trajectory of a human target based on the radar array; in the case that the motion trajectory represents that the human target approaches the refrigerator door, determining the limb shape and the limb action of the human limb part detected by each radar in the radar array according to the real-time detection signal of each radar; determining the real-time human body shape of the human target according to the limb shape of the human limb part detected by each radar, and determining the real-time human body action of the human target according to the limb action of the human limb part detected by each radar; determining the intention recognition result of the human target; in the case that the intention recognition result represents that the human target has the intention to open the door, controlling the refrigerator door to open. The application improves the accuracy of the refrigerator door opening.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method, apparatus and device for controlling a refrigerator door. Background Technology

[0002] Currently, most refrigerators on the market use one of three automatic door opening modes: foot-activated sensing, infrared sensing, or camera-based image recognition. Foot-activated sensing requires the user to actively trigger the sensor; once detected, the device processes the logic. Infrared sensing uses an infrared wave sensing mechanism; when the infrared sensor detects an infrared wave, it automatically executes logic. Camera recognition primarily relies on image recognition; once image information is detected, logic is executed. Foot-activated sensing requires manual triggering, resulting in a less than ideal smart experience; infrared sensors are easily affected by external infrared waves, leading to false triggers; and camera recognition can pose cybersecurity risks. Summary of the Invention

[0003] This application provides a method, device, and equipment for controlling a refrigerator door, which can accurately control the automatic opening of the refrigerator door according to the intention of a human target, thereby improving the accuracy of refrigerator door opening.

[0004] On one hand, this application provides a method for controlling a refrigerator door, wherein a radar array is provided on the refrigerator door, the radar array including at least two radars, and the method includes: When the radar array detects a human target within the detection area around the refrigerator door, the movement trajectory of the human target is determined based on the radar array. When the motion trajectory indicates that the human target is approaching the refrigerator door, the limb shape and limb movement of the human limb detected by each radar in the radar array are determined according to the real-time detection signal of each radar; wherein, the human limb detected by each radar is different. Based on the limb shape of each human limb detected by each radar, the real-time human shape of the human target is determined, and based on the limb movement of each human limb detected by each radar, the real-time human movement of the human target is determined. Based on the real-time human form and real-time human movements of the human target, the intent recognition result of the human target is determined; If the intent recognition result indicates that the human target has the intent to open the door, the refrigerator door is controlled to open.

[0005] In one exemplary embodiment, the method further includes: The sample human body shape and sample human body action are obtained; the sample human body target is labeled with a sample intent tag; the sample intent tag indicates whether the sample human body target has the intention to open a door; The sample human body shape and sample human body movement are input into a preset model for intent prediction processing to obtain the sample intent prediction result of the sample human body target. Based on the difference between the sample intent prediction result and the sample intent label, the preset model is trained to obtain a human intent recognition model; Accordingly, determining the intent recognition result of the human target based on the real-time human morphology and real-time human movements includes: The real-time human form and real-time human movements of the human target are input into the human intent recognition model for intent recognition processing to obtain the intent recognition result of the human target.

[0006] In one exemplary embodiment, the step of inputting the sample human body shape and sample human body movement into a preset model for intent prediction processing to obtain the sample human body target's sample intent prediction result includes: The sample human body shape and sample human body movement are input into a preset model, and the trajectory prediction network based on the preset model is used to perform trajectory prediction processing on the sample human body shape to obtain the sample trajectory result of the sample human body target. The human motion prediction network based on the preset model performs dynamic prediction processing on the sample human body movements to obtain the sample dynamic result of the sample human body target; the sample dynamic result is a first result where the sample human body is close to the refrigerator door, a second result where the sample human body is in a stationary state, or a third result where the sample human body is far away from the refrigerator door. The intent prediction network based on the preset model performs intent prediction processing on the sample trajectory results and the sample dynamic results to obtain the sample intent prediction result of the human target.

[0007] In one exemplary embodiment, the preset model further includes a human category prediction network, and the method further includes: Based on the human body category prediction network, the human body morphology of the sample is subjected to human body category prediction processing to obtain the human body category of the sample. The intent prediction network based on the preset model performs intent prediction processing on the sample trajectory results and the sample dynamic results to obtain the sample intent prediction result of the human target, including: Based on the intent prediction network, the intent of the sample trajectory results is predicted to obtain the first sample intent; Based on the intent prediction network, the intent of the sample dynamic results is predicted to obtain the second sample intent; Based on the intent prediction network, the intent of the sample human body category is predicted to obtain the third sample intent; Based on the first sample intent, the second sample intent, and the third sample intent, the sample intent prediction result of the human target is determined.

[0008] In one exemplary embodiment, the radar array is disposed on the outer surface of the refrigerator door, and the radar array includes at least two radar groups, wherein each radar group includes at least two radars; the radars in the radar array are distributed on the outer surface of the refrigerator door, and the step of determining the real-time human form of the human target based on the limb form of the human limb parts detected by each radar, and determining the real-time human action of the human target based on the limb movements of the human limb parts detected by each radar, includes: Based on the limb shape of each human body part detected by each radar, the head shape, arm shape, and leg shape of the human target are determined. The real-time human form of the human target is determined based on the head shape, arm shape, and leg shape of the human target. Based on the limb movements of each human body part detected by each radar, the head movements, arm movements, and leg movements of the human target are determined. The real-time human motion of the human target is determined based on the head movement, arm movement, and leg movement of the human target.

[0009] In one exemplary embodiment, the radar array includes a first radar group and a second radar group, each radar group including two radars. The radars in the first radar group are distributed on both sides of the top of the refrigerator door, and the radars in the second radar group are distributed on both sides of the bottom of the refrigerator door. The step of determining the head shape, arm shape, and leg shape of the human target based on the limb shape detected by each radar includes: The head and arm shapes of the human target are detected by the first radar group. The second radar group detects the leg shape of the human target; The step of determining the head movement, arm movement, and leg movement of the human target based on the limb movements detected by each radar on each human limb part includes: The head and arm movements of the human target are detected by the first radar group. The second radar group detects the leg movements of the human target.

[0010] In one exemplary embodiment, when the motion trajectory indicates that the human target is approaching the refrigerator door, determining the limb shape and limb movement of each human limb detected by each radar in the radar array based on the real-time detection signal of each radar includes: When the motion trajectory indicates that the human target is approaching the refrigerator door, the real-time detection signal of each radar in the radar array is acquired; The real-time detection signal of each radar is filtered to remove noise and obtain a denoised signal. Based on the denoised signal corresponding to each radar, the limb shape and limb movement of the human limb parts detected by each radar are determined.

[0011] In one exemplary embodiment, a current loop is further provided on the outer side of the refrigerator door, and the method further includes: The resistance of the refrigerator door is detected by the current loop, and the presence of an obstacle is determined based on the resistance detection results. If an obstacle is detected in the refrigerator door, reverse braking is triggered against the obstacle within a preset time period.

[0012] On the other hand, a refrigerator door control device is provided, wherein a radar array is provided on the refrigerator door, the radar array including at least two radars, and the device includes: The motion trajectory determination module is used to determine the motion trajectory of the human target based on the radar array when the radar array detects the presence of a human target in the detection area around the refrigerator door. The limb information determination module is used to determine the limb shape and limb movement of each human limb detected by each radar based on the real-time detection signal of each radar in the radar array when the motion trajectory indicates that the human target is close to the refrigerator door; wherein, the human limb detected by each radar is different. The real-time human motion determination module is used to determine the real-time human form of the human target based on the limb form of each human limb detected by each radar, and to determine the real-time human motion of the human target based on the limb motion of each human limb detected by each radar. An intent recognition module is used to determine the intent recognition result of the human target based on the real-time human morphology and the real-time human movements of the human target. The control module is used to control the refrigerator door to open when the intent recognition result indicates that the human target has the intent to open the door.

[0013] In one exemplary embodiment, the apparatus further includes: The sample information acquisition module is used to acquire the human morphology and human actions of the human target; the human target is labeled with a sample intent tag; the sample intent tag indicates whether the human target has the intention to open a door; The sample intent prediction module is used to input the sample human body shape and sample human body action into a preset model for intent prediction processing, and obtain the sample intent prediction result of the sample human body target. The model training module is used to train the preset model based on the difference between the sample intent prediction result and the sample intent label to obtain a human intent recognition model. Accordingly, the intent recognition module is also used to input the real-time human form and the real-time human action of the human target into the human intent recognition model for intent recognition processing, so as to obtain the intent recognition result of the human target.

[0014] In one exemplary embodiment, the intent recognition module includes: The sample trajectory prediction unit is used to input the sample human body shape and sample human body movement into a preset model, and perform trajectory prediction processing on the sample human body shape based on the trajectory prediction network of the preset model to obtain the sample trajectory result of the sample human body target. The sample dynamic determination unit is used to perform dynamic prediction processing on the sample human body action based on the human body dynamic prediction network of the preset model to obtain the sample dynamic result of the sample human body target; the sample dynamic result is a first result in the sample human body being close to the refrigerator door, a second result in the sample human body being in a stationary state, or a third result in the sample human body being far away from the refrigerator door. The sample intent prediction unit is used to perform intent prediction processing on the sample trajectory results and the sample dynamic results based on the intent prediction network of the preset model, so as to obtain the sample intent prediction result of the human target.

[0015] In one exemplary embodiment, the preset model further includes a human category prediction network, and the device further includes: The sample category prediction module is used to perform human category prediction processing on the human body morphology of the sample based on the human body category prediction network to obtain the human body category of the sample. The sample intent prediction unit includes: The first intent prediction subunit is used to predict the intent of the sample trajectory result based on the intent prediction network to obtain the first sample intent. The second intent prediction subunit is used to predict the intent of the sample dynamic results based on the intent prediction network to obtain the second sample intent. The third intent prediction subunit is used to predict the intent of the sample human body category based on the intent prediction network to obtain the third sample intent. The sample intent determination subunit is used to determine the sample intent prediction result of the human target based on the first sample intent, the second sample intent, and the third sample intent.

[0016] In one exemplary embodiment, the radar array is disposed on the outer surface of the refrigerator door, and the radar array includes at least two radar groups, wherein each radar group includes at least two radars; the radars in the radar array are distributed on the outer surface of the refrigerator door, and the real-time human motion determination module includes: The limb shape determination unit is used to determine the head shape, arm shape, and leg shape of the human target based on the limb shape of each human limb part detected by the radar. A real-time human body shape determination unit is used to determine the real-time human body shape of the human body target based on the head shape, arm shape and leg shape of the human body target. The limb movement determination unit is used to determine the head movement, arm movement, and leg movement of the human target based on the limb movements of each human limb part detected by the radar. The real-time human motion determination unit is used to determine the real-time human motion of the human target based on the head motion, arm motion, and leg motion of the human target.

[0017] In one exemplary embodiment, the radar array includes a first radar group and a second radar group, each radar group including two radars. The radars in the first radar group are distributed on both sides of the top of the refrigerator door, and the radars in the second radar group are distributed on both sides of the bottom of the refrigerator door. The limb morphology determination unit includes: An arm shape detection subunit is used to detect the head shape and arm shape of the human target based on the first radar group. A leg shape detection subunit is used to detect the leg shape of the human target based on the second radar group; The limb movement determination unit includes: An arm movement detection subunit is used to detect the head and arm movements of the human target based on the first radar group. The leg movement detection subunit is used to detect the leg movements of the human target based on the second radar group.

[0018] In one exemplary embodiment, the limb information determination module includes: A real-time signal acquisition unit is used to acquire the real-time detection signal of each radar in the radar array when the motion trajectory indicates that the human target is approaching the refrigerator door; The noise reduction unit is used to filter the real-time detection signal of each radar, remove noise from the real-time detection signal, and obtain a noise-reduced signal. The limb movement determination unit is used to determine the limb shape and limb movement of each human limb part detected by each radar based on the denoised signal corresponding to each radar.

[0019] In one exemplary embodiment, a current loop is further provided on the outer side of the refrigerator door, and the device further includes: An obstacle detection module is used to detect the resistance of the refrigerator door through the current loop and determine whether an obstacle exists based on the resistance detection result. The reverse braking module is used to trigger reverse braking against the obstacle within a preset time period when the refrigerator door is detected to encounter an obstacle.

[0020] On the other hand, an electronic device is provided, the device including a processor and a memory, the memory storing at least one instruction or at least one program, the at least one instruction or the at least one program being loaded and executed by the processor to implement the refrigerator door control method as described above.

[0021] On the other hand, a computer storage medium is provided that stores at least one instruction or at least one program, which is loaded and executed by a processor to implement the refrigerator door control method described above.

[0022] On the other hand, a computer program product or computer program is provided, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the refrigerator door control method as described above.

[0023] The refrigerator door control method, device, and equipment provided in this application have the following technical advantages: This application includes a radar array on the refrigerator door, comprising at least two radars. When the radar array detects a human target within a detection area around the refrigerator door, it determines the human target's trajectory. Then, if the trajectory indicates the human target is approaching the refrigerator door, it determines the limb shape and movement of each detected limb based on the real-time detection signal from each radar in the array. Each radar detects a different limb. This allows for accurate determination of the human target's real-time limb shape and movement based on the limb shape detected by each radar. Furthermore, based on the human target's real-time limb shape and movement, the application quickly and accurately determines the human target's intent recognition result, improving the matching degree between the intent recognition result and the human target's true intent. If the intent recognition result indicates the human target intends to open the door, the application controls the refrigerator door to open. This achieves accurate automatic opening of the refrigerator door based on the human target's intent, improving the accuracy of the refrigerator door opening. Attached Figure Description

[0024] To more clearly illustrate the technical solutions and advantages in the embodiments or prior art of this specification, the drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a schematic diagram of the structure of a radar array for a refrigerator door provided in an embodiment of this specification; Figure 2 This is a flowchart illustrating a refrigerator door control method provided in an embodiment of this specification; Figure 3 This is a flowchart illustrating a method for determining the real-time human form and real-time human movements of the human target, as provided in an embodiment of this specification. Figure 4 This is a flowchart illustrating a method for determining the limb shape and limb movement of a human limb part detected by each radar based on the real-time detection signal of each radar in the radar array, as provided in the embodiments of this specification. Figure 5 This is a flowchart illustrating a method for training a human intent recognition model provided in an embodiment of this specification. Figure 6 This is a schematic diagram of the structure of a refrigerator door control system provided in the embodiments of this specification; Figure 7 This is a schematic diagram of the structure of a refrigerator door control device provided in the embodiments of this specification; Figure 8 This is a schematic diagram of the structure of a server provided in the embodiments of this specification. Detailed Implementation

[0026] The technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0027] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0028] In the embodiments of this application, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0029] like Figure 1 As shown, Figure 1 This embodiment provides a schematic diagram of the structure of a radar array for a refrigerator door, including a refrigerator door 01 and a human target 02. A radar array is provided on the refrigerator door 01, and the radar array includes at least two radars 03. For example, the radar array may include four radars 03, which are distributed at the top, bottom, left and right positions on the outer side of the refrigerator door 01.

[0030] The following describes a method for controlling a refrigerator door according to this application. Figure 2This is a flowchart illustrating a refrigerator door control method provided in an embodiment of this specification. This specification provides the operational steps of the method described in the embodiment or flowchart, but based on conventional or non-inventive labor, more or fewer operational steps may be included. The order of steps listed in the embodiment is merely one possible execution order among many and does not represent the only possible execution order. In actual system or server product execution, the method can be executed sequentially or in parallel (e.g., in a parallel processor or multi-threaded processing environment) as shown in the embodiment or drawings. A radar array is provided on the refrigerator door, and the radar array includes at least two radars, specifically as shown... Figure 2 As shown, the method may include: S201: When the radar array detects a human target within the detection area around the refrigerator door, the movement trajectory of the human target is determined based on the radar array. In the embodiments described in this specification, a 24GHz or 60GHz millimeter-wave radar can be used, which has high sensitivity, strong anti-interference capability, and can accurately detect human movement. The method of this embodiment can be applied to the main control module, which uses a low-power microcontroller to receive radar sensor signals, perform data processing, and make logical judgments. The radar array can detect in real time whether there is a human target in the detection area around the refrigerator door. When a human target is detected approaching, the signal is transmitted to the main control module.

[0031] The refrigerator door can be used as the center to determine the detection area of ​​the radar array; within the detection area, when a moving human target appears, the frequency of the echo received by the radar will change, and the Doppler frequency shift formula is: Fd=2Vr / λ Where Fd is the Doppler frequency shift; Vr is the radial velocity of the target relative to the radar; and λ is the wavelength of the radar wave. This formula can be used to determine whether there is a moving human target in the detection area.

[0032] If a human target is found to be present in the detection area, the movement trajectory of the human target can be determined based on the radar array; the movement trajectory is used to characterize whether the human target is close to the refrigerator door.

[0033] For example, the radial velocity of the human target can be calculated; the radial velocity is the component of the human target's actual velocity in the radar line-of-sight direction, and the calculation formula is: Vr=V*COS(θ) Where V: actual velocity; θ: angle between the target's direction of motion and the radar line of sight; This formula can be used to calculate the radial velocity of a human target within a preset time period, thereby analyzing whether the human target is close to the refrigerator door. If the radial velocity is greater than zero and the velocity direction is towards the refrigerator door, then it is determined that the human target is close to the refrigerator door.

[0034] S202: When the motion trajectory indicates that the human target is close to the refrigerator door, the limb shape and limb movement of the human limb detected by each radar are determined according to the real-time detection signal of each radar in the radar array; wherein, the human limb detected by each radar is different.

[0035] In the embodiments of this specification, when the motion trajectory indicates that the human target is approaching the refrigerator door, the limb shape and limb movement of each human limb detected by each radar can be further determined based on the real-time detection signal of each radar in the radar array.

[0036] Among them, limb shape refers to the shape corresponding to the limb parts detected by radar, which can be used to further determine the real-time human shape of the human target, thereby judging whether the human target continues to approach the refrigerator door and further improving the judgment accuracy; limb movement refers to the movement corresponding to the limb parts detected by radar, which can include micro-movements of the hands, legs, feet and other parts, and can be used to further determine the real-time human movement of the human target.

[0037] For example, each radar can calculate the horizontal distance between different parts of the human body and the refrigerator door by measuring the round-trip time of the signal, using the following formula: R=C*Δt / 2 Where R: distance, C: speed of light (approximately 3 x 10^8 m / s), Δt: round-trip time of the signal.

[0038] Then, by calculating the horizontal distance between different limbs of the human body and the refrigerator door, it can be determined whether the human target should continue to approach the refrigerator. The target distance for opening the refrigerator door can be preset according to the preset target distance. The refrigerator door needs to open if the distance between the human target and the refrigerator door is less than or equal to the target distance. For example, the target distance can be set to 0.5m.

[0039] For example, limb status can be determined by further calculating the echo signal power based on distance; the echo signal power is related to distance and radar cross-section (RCS). Pr=PtG2λ2σ / (4π)3R4 Where Pr: received power, Pt: transmitted power, G: antenna gain, σ: radar cross-section, and R: range.

[0040] This embodiment can determine the human body shape by power intensity, thereby determining the human body category corresponding to the human target. The larger the value, the stronger the power reflected back to the radar, indicating that the human target is more likely to be an adult. Conversely, the lower the value, the more likely the human target is a child. Since refrigerator users are usually adults, the condition for opening the refrigerator door is that the human target is an adult rather than a child.

[0041] For example, limb movements can be calculated using the micro-Doppler effect. The calculation formula is based on the localized periodic micro-movements (rotation, swinging, etc.) of the human target. The micro-Doppler frequency shift produced by the minute movements of the human target (such as arm swinging) is calculated using the following formula: fmD=2Vm / λ Where fmD is the micro-Doppler frequency shift and Vm is the velocity of the minute movement, thus enabling the determination of the limb movement of each human limb detected by the radar.

[0042] S203: Determine the real-time human form of the human target based on the limb shape of each human limb detected by each radar, and determine the real-time human movement of the human target based on the limb movement of each human limb detected by each radar.

[0043] In the embodiments of this specification, the real-time human form of the target can be determined based on the limb shape of each human limb detected by each radar. For example, it can be determined whether the target is close to the refrigerator door based on the direction the target's face is facing. If the target's face is facing the refrigerator door and the target is in motion, it is determined that the target continues to approach the refrigerator door. If the target's face is facing the refrigerator door and the target is in motion, it is determined that the target continues to approach the refrigerator door. If the target's face is facing away from the refrigerator door and the target is in motion, it is determined that the target is moving away from the refrigerator door.

[0044] Furthermore, based on the limb shape of the human target, the size of the limb area can be further determined, thereby facilitating the determination of the human category corresponding to the human target; human categories include adults, children, etc.

[0045] The real-time human actions of the target include actions such as the target approaching the refrigerator door, the target remaining near the refrigerator door, and the target moving away from the refrigerator door. For example, if a target's hand is detected reaching towards the refrigerator door, the action of the target approaching the refrigerator door can be determined; if a target's leg is detected moving towards the refrigerator door, the action of the target approaching the refrigerator door can be determined; if a target's leg is detected moving away from the refrigerator door, the action of the target moving away from the refrigerator door can be determined.

[0046] S204: Determine the intent recognition result of the human target based on the real-time human form and the real-time human movement of the human target.

[0047] In the embodiments of this specification, the intention recognition result of the human target can be determined based on the real-time human form and the real-time human movement of the human target; wherein, the intention recognition result of the human target includes the human target's intention to open the door and the human target's intention not to open the door.

[0048] S205: If the intent recognition result indicates that the human target has the intent to open the door, control the refrigerator door to open.

[0049] In the embodiments of this specification, the intent recognition result can be used to accurately determine whether a human target has the intention to open the door, and if it is determined that the human target has the intention to open the door, the refrigerator door can be controlled to open, thus realizing the precise control of the refrigerator door to open automatically according to the human target's intent.

[0050] In some embodiments, the radar array is disposed on the outer surface of the refrigerator door, and the radar array includes at least two radar groups, wherein each radar group includes at least two radars; as shown Figure 3 As shown, the radar array consists of radars distributed on the outside of the refrigerator door. The process of determining the real-time human form of the target based on the limb shape detected by each radar, and determining the real-time human movement of the target based on the limb movements detected by each radar, includes: S2031: Based on the limb shape of each human limb detected by the radar, determine the head shape, arm shape, and leg shape of the human target. S2032: Determine the real-time human form of the human target based on the head shape, arm shape, and leg shape of the human target; S2033: Based on the limb movements of each human body part detected by each radar, determine the head movement, arm movement, and leg movement of the human target. S2034: Determine the real-time human motion of the human target based on the head movement, arm movement, and leg movement of the human target.

[0051] In the embodiments of this specification, radars at different positions on the refrigerator door can detect the shape and movement of different limbs of a human target. For example, the radar array may include a first radar located above the refrigerator door, a second radar located in the middle of the refrigerator door, and a third radar located below the refrigerator door. The first radar can detect the shape and movement of the human target's head, the second radar can detect the shape and movement of the human target's arms, and the third radar can detect the movement and shape of the human target's legs. Therefore, based on the shape of the human target's head, arms, and legs, the real-time human shape of the human target can be determined; based on the head, arms, and legs, the real-time human movement of the human target can be determined. This facilitates subsequent determination of whether the human target intends to open the refrigerator door based on the real-time human shape and real-time human movement.

[0052] In some embodiments, the radar array includes a first radar group and a second radar group, each radar group including two radars. The radars in the first radar group are distributed on both sides of the top of the refrigerator door, and the radars in the second radar group are distributed on both sides of the bottom of the refrigerator door. Determining the head shape, arm shape, and leg shape of the human target based on the limb shape detected by each radar includes: The head and arm shapes of the human target are detected by the first radar group. The second radar group detects the leg shape of the human target; The step of determining the head movement, arm movement, and leg movement of the human target based on the limb movements detected by each radar on each human limb part includes: The head and arm movements of the human target are detected by the first radar group. The second radar group detects the leg movements of the human target.

[0053] In this embodiment, the radar array may include four radars, which are disposed on the outside of the refrigerator door and distributed around the perimeter of the refrigerator door. Two radars (the first radar group) are disposed on the top two sides of the refrigerator door, and the other two radars (the second radar group) are disposed on the bottom two sides of the refrigerator door. The first radar group can detect the head and arm shapes of the human target, and detect head and arm movements; the second radar group can detect the leg shapes and leg movements of the human target. The radar array of this embodiment achieves a micro-motion recognition accuracy of ±2cm for human targets.

[0054] In some embodiments, such as Figure 4As shown, when the motion trajectory indicates that the human target is approaching the refrigerator door, determining the limb shape and limb movement of each human limb detected by each radar in the radar array based on the real-time detection signal of each radar includes: S2021: When the motion trajectory indicates that the human target is approaching the refrigerator door, acquire the real-time detection signal of each radar in the radar array; S2022: Filter the real-time detection signal of each radar to remove noise from the real-time detection signal and obtain a denoised signal; S2023: Based on the denoised signal corresponding to each radar, determine the limb shape and limb movement of the human limb parts detected by each radar.

[0055] In the embodiments of this specification, when the motion trajectory indicates that the human target is approaching the refrigerator door, the real-time detection signal of each radar in the radar array is acquired. A Convolutional Neural Network (CNN) can be used to filter the real-time detection signal of each radar, separating the signal from the noise and obtaining a denoised signal. Finally, the denoised signal is used to determine the limb shape and movement of the human limb detected by each radar, thereby improving the accuracy of the limb shape and movement. Here, the signal refers to the signal corresponding to the shape and movement of the human target, and the noise may include irregular fluctuations of interference signals.

[0056] In some embodiments, using a convolutional neural network (CNN) to separate signal from noise is a deep learning-based end-to-end approach. Its core idea is to extract the effective components of the signal and suppress noise by automatically learning features.

[0057] 1. Data generation strategy: If a noisy signal and its corresponding clean signal already exist, they can be directly paired and used as a training set.

[0058] If there is no clean label, it needs to be generated by synthesizing data: adding noise of different intensities and types to the original signal (such as real recordings or medical images) to form noisy-clean pairs.

[0059] 2. Convolutional Autoencoder (CAE) Architecture CAE is a classic model for denoising tasks, combining the local feature extraction capabilities of CNNs with the reconstruction capabilities of autoencoders: (a) Encoder: Its function is to compress noisy signals into low-dimensional feature representations and gradually extract key information.

[0060] Typical structure: Convolutional layers: Use small convolutional kernels (such as 3×3 or 5×5) to capture local temporal / spatial features, and the activation function can be ReLU.

[0061] Batch Normalization: Accelerates training stability.

[0062] Max Pooling: Reduces dimensionality while preserving salient features.

[0063] Multi-layer stacking: Features are abstracted step by step through multiple encoder blocks.

[0064] (b) Decoder: Its function is to reconstruct a clean signal from low-dimensional features and gradually remove noise.

[0065] Typical structure: Upsampling layer: Recovers the signal length through transpose convolution or upsampling.

[0066] Convolutional layers: Use convolutional kernels symmetrical to the encoder to progressively refine the features.

[0067] Skip Connection: Concatenates features from intermediate layers of the encoder with corresponding layers of the decoder to enhance detail recovery (similar to U-Net).

[0068] Sigmoid / ReLU activation: Selected based on signal type (e.g., audio commonly uses Sigmoid output in the range of [-1,1]).

[0069] During training, noise (such as Gaussian noise or white noise) can be dynamically added to the clean signal to simulate real-world scenarios. Radar signal processing: Input: Radar echo signal containing spurious noise. Output: Target echo separation.

[0070] In some embodiments, such as Figure 5 As shown, the method further includes: S501: Obtain the human body shape and human body action of the human body target; the human body target is labeled with a sample intent tag; the sample intent tag indicates whether the human body target has the intention to open a door; S502: Input the sample human body shape and sample human body action into a preset model for intent prediction processing to obtain the sample intent prediction result of the sample human body target. S503: Based on the difference between the sample intent prediction result and the sample intent label, the preset model is trained to obtain a human intent recognition model.

[0071] Accordingly, determining the intent recognition result of the human target based on the real-time human morphology and real-time human movements includes: The real-time human form and real-time human movements of the human target are input into the human intent recognition model for intent recognition processing to obtain the intent recognition result of the human target.

[0072] In the embodiments of this specification, the sample human target may include a positive sample human target with the intention to open a door and a negative sample human target without the intention to open a door, and the sample intention label may include a first intention label with the intention to open a door and a second intention label without the intention to open a door.

[0073] A human intent recognition model can be obtained by training a preset model based on the human morphology and movements of the target human body; the preset model can be a machine learning model. Based on the difference between the predicted intent and the intent label, target loss data is determined, and the parameters of the preset model are adjusted according to the target loss data until a training termination condition is met. The preset model at the end of training is then identified as the human intent recognition model. The training termination condition can be determined based on at least one of the target loss data and the number of iterations; for example, the training termination condition can be that the target loss data is less than a preset value.

[0074] In the embodiments of this specification, after the model training is completed, the real-time human form and real-time human actions of the human target can be input into the human intention recognition model for intention recognition processing, thereby quickly and accurately obtaining the intention recognition result of the human target.

[0075] In some embodiments, the step of inputting the sample human body morphology and sample human body movements into a preset model for intent prediction processing to obtain the sample human body target's intent prediction result includes: The sample human body shape and sample human body movement are input into a preset model, and the trajectory prediction network based on the preset model is used to perform trajectory prediction processing on the sample human body shape to obtain the sample trajectory result of the sample human body target. The human motion prediction network based on the preset model performs dynamic prediction processing on the sample human body movements to obtain the sample dynamic result of the sample human body target; the sample dynamic result is a first result where the sample human body is close to the refrigerator door, a second result where the sample human body is in a stationary state, or a third result where the sample human body is far away from the refrigerator door. The intent prediction network based on the preset model performs intent prediction processing on the sample trajectory results and the sample dynamic results to obtain the sample intent prediction result of the human target.

[0076] In the embodiments of this specification, the sample trajectory results can be used to determine whether the motion trajectory of the human target is stable; during the training process, different sample dynamic results can be set for different sample human targets; for example, the first sample human target corresponds to the first result of the sample human being approaching the refrigerator door, the second result of the first sample human target corresponding to the sample human being in a stationary state, and the third result of the first sample human target corresponding to the sample human being moving away from the refrigerator door.

[0077] In some embodiments, the real-time human body shape of the human target can be used to determine whether the human target is located within a preset distance range from the refrigerator door. Real-time human movements can be used to determine whether the human target is close to the refrigerator door or the duration of its stay, and further determine the duration of the human target's stay within the preset distance range. If the human target is within the preset distance range, the duration of its stay within the preset distance range reaches the preset duration, and the trajectory of the human target is stable, an opening command can be triggered. The trajectory stability of the human target can be determined by the real-time human body shape and the real-time human movements. For example, the preset distance range can be 0.5-1 meter, and the preset duration can be set to 3 seconds. In some embodiments, when a user is detected to be continuously within the 0.5-1 meter range for 3 seconds and the trajectory is stable, a high-confidence opening command (response time <80ms) is triggered.

[0078] In some embodiments, the preset model further includes a human category prediction network, and the method further includes: Based on the human body category prediction network, the human body morphology of the sample is subjected to human body category prediction processing to obtain the human body category of the sample. The intent prediction network based on the preset model performs intent prediction processing on the sample trajectory results and the sample dynamic results to obtain the sample intent prediction result of the human target, including: Based on the intent prediction network, the intent of the sample trajectory results is predicted to obtain the first sample intent; Based on the intent prediction network, the intent of the sample dynamic results is predicted to obtain the second sample intent; Based on the intent prediction network, the intent of the sample human body category is predicted to obtain the third sample intent; Based on the first sample intent, the second sample intent, and the third sample intent, the sample intent prediction result of the human target is determined.

[0079] In the embodiments of this specification, the sample human body category may include categories such as adults and children; wherein, adults are the category that meets the door-opening condition, and children are the category that does not meet the door-opening condition. Intent prediction can be performed on the sample trajectory results based on the intent prediction network to obtain a first sample intent; intent prediction can be performed on the sample dynamic results based on the intent prediction network to obtain a second sample intent; intent prediction can be performed on the sample human body category based on the intent prediction network to obtain a third sample intent; and if the first sample intent, the second sample intent, and the third sample intent are all of the intent to open the door, the sample intent prediction result is determined to have an intent to open the door; if any one of the first sample intent, the second sample intent, and the third sample intent is not of the intent to open the door, the sample intent prediction result is determined to not have an intent to open the door, thereby improving the accuracy of the sample intent prediction result.

[0080] In some embodiments, a current loop is further provided on the outer side of the refrigerator door, and the method further includes: The resistance of the refrigerator door is detected by the current loop, and the presence of an obstacle is determined based on the resistance detection results. If an obstacle is detected in the refrigerator door, reverse braking is triggered against the obstacle within a preset time period.

[0081] In this embodiment, a current loop can be installed on the refrigerator door to detect whether the refrigerator door encounters an obstacle. If an obstacle is detected, the type of obstacle can be further determined. If the obstacle is determined to be non-living, reverse braking is triggered within a preset time period, thereby achieving a dynamic anti-collision function for the refrigerator door. The preset time period can be set according to actual conditions, for example, it can be set to 0-10ms. Reverse braking can be triggered within 10ms when the refrigerator door encounters an obstacle.

[0082] In some embodiments, controlling the opening of the refrigerator door may include: controlling the opening of the refrigerator door based on an S-shaped speed curve.

[0083] In the embodiments of this specification, when the intent recognition result indicates that the human target has the intent to open the door, an opening command can be triggered, and the refrigerator door can be controlled to open within a preset opening time range; controlling the opening of the refrigerator door based on the S-shaped speed curve can control the refrigerator door to take 1.5 seconds to travel from 0 to 90°, with a peak torque of 12 N•m, thus realizing the flexible start of the door. The preset door opening time range can be set to 0-80 milliseconds, thereby improving the response speed of the refrigerator door opening.

[0084] S-curve speed control for door opening is a smooth motion control technology that achieves smooth start and stop of door movement by planning the timing relationship between acceleration, speed and position, thereby reducing mechanical impact and energy loss.

[0085] An S-curve is essentially a continuous, smooth function of acceleration with respect to time, and its characteristics are as follows: Initial stage: Acceleration increases linearly from zero to its maximum value (acceleration stage).

[0086] Intermediate stage: Acceleration remains constant (uniform acceleration stage).

[0087] Final stage: Acceleration decreases linearly to zero (deceleration stage).

[0088] Final position: The velocity also smoothly returns to zero, avoiding oscillation.

[0089] The controller needs to implement three-loop control (position → velocity → acceleration) to ensure that the actual motion trajectory tracks an S-curve. The hardware includes drivers, sensors, and the controller itself. Drivers: servo motors or stepper motors. Sensors include encoders and current sensors. Encoders provide real-time feedback on the door's position; current sensors monitor motor torque to prevent overload. The controller: a programmable logic controller (PLC) or an embedded microcontroller unit (MCU).

[0090] In some embodiments, if no trigger operation on the refrigerator door is detected within a preset cumulative time period, the main control module is controlled to enter a deep sleep mode, thereby reducing the power consumption of the entire unit. For example, if no trigger operation on the refrigerator door is detected within 30 seconds, the main control module is controlled to enter a deep sleep mode, and the power consumption of the entire unit is reduced to 0.8W.

[0091] The sensor types in the current loop of this embodiment can include: electromagnetic current loop (Hall Effect Sensor): which determines the presence of an object by detecting changes in the magnetic field, suitable for detecting metallic objects; photoelectric sensor: which uses the principle of beam blocking for non-contact detection and has strong anti-interference capabilities; and capacitive sensor: which detects changes in object distance and is suitable for non-metallic materials. The detection distance is selected according to the door size (e.g., 5-30mm); the installation position can be on both sides of the door, for example, symmetrically installed on both sides of the refrigerator door to form a "light curtain" or "magnetic induction zone". Motion path coverage: ensure that the sensor covers the entire movement trajectory of the door to avoid blind spots.

[0092] Sliding window averaging: Averages N consecutive detection signals to filter transient fluctuations. Dynamic threshold adjustment: Automatically calibrates sensitivity based on changes in ambient light / temperature. State machine logic: Triggers reverse braking after a signal continuously exceeds a threshold for T seconds.

[0093] Specifically, in the embodiments of this specification, such as Figure 6 As shown, Figure 6 This is a schematic diagram of a refrigerator door control system, including a radar array and a main control module. The main control module includes an MCU algorithm processing unit, an electronic control unit, and a door control unit. The door control unit is used to control the opening and closing of the refrigerator door. The MCU algorithm processing unit receives signals detected by the radar array, performs signal filtering, and then uses it for human detection and judgment. When a human target is detected, the system further detects the human form and human movement, and then performs human intention recognition based on the human form and human movement. If the system recognizes that the human has the intention to open the door, the electronic control unit sends an opening command to the door control unit to open the refrigerator door.

[0094] As can be seen from the technical solutions provided in the embodiments of this specification above, the embodiments of this specification provide a radar array on the refrigerator door. The radar array includes at least two radars. When the radar array detects a human target within a detection area around the refrigerator door, the movement trajectory of the human target is determined based on the radar array. When the movement trajectory indicates that the human target is approaching the refrigerator door, the limb shape and limb movement of the human limb detected by each radar are determined based on the real-time detection signal of each radar in the radar array. Each radar detects a different human limb. Based on the limb shape of the human limb detected by each radar, the real-time human shape of the human target is determined, and based on the limb movement of the human limb detected by each radar, the real-time human movement of the human target is determined. Based on the real-time human shape and the real-time human movement of the human target, the intention recognition result of the human target is determined. When the intention recognition result indicates that the human target has the intention to open the door, the refrigerator door is controlled to open.

[0095] This specification also provides an embodiment of a refrigerator door control device, wherein the refrigerator door is equipped with a radar array, the radar array including at least two radars, such as... Figure 7 As shown, the device includes: The motion trajectory determination module 710 is used to determine the motion trajectory of the human target based on the radar array when the radar array detects the presence of a human target in the detection area around the refrigerator door. The limb information determination module 720 is used to determine the limb shape and limb movement of each human limb detected by each radar based on the real-time detection signal of each radar in the radar array when the motion trajectory indicates that the human target is close to the refrigerator door; wherein, the human limb detected by each radar is different. The real-time human motion determination module 730 is used to determine the real-time human form of the human target based on the limb form of each human limb part detected by each radar, and to determine the real-time human motion of the human target based on the limb motion of each human limb part detected by each radar. The intent recognition module 740 is used to determine the intent recognition result of the human target based on the real-time human shape and the real-time human action of the human target. The control module 750 is used to control the refrigerator door to open when the intent recognition result indicates that the human target has the intent to open the door.

[0096] In some embodiments, the apparatus further includes: The sample information acquisition module is used to acquire the human morphology and human actions of the human target; the human target is labeled with a sample intent tag; the sample intent tag indicates whether the human target has the intention to open a door; The sample intent prediction module is used to input the sample human body shape and sample human body action into a preset model for intent prediction processing, and obtain the sample intent prediction result of the sample human body target. The model training module is used to train the preset model based on the difference between the sample intent prediction result and the sample intent label to obtain a human intent recognition model. Accordingly, the intent recognition module is also used to input the real-time human form and the real-time human action of the human target into the human intent recognition model for intent recognition processing, so as to obtain the intent recognition result of the human target.

[0097] In some embodiments, the intent recognition module includes: The sample trajectory prediction unit is used to input the sample human body shape and sample human body movement into a preset model, and perform trajectory prediction processing on the sample human body shape based on the trajectory prediction network of the preset model to obtain the sample trajectory result of the sample human body target. The sample dynamic determination unit is used to perform dynamic prediction processing on the sample human body action based on the human body dynamic prediction network of the preset model to obtain the sample dynamic result of the sample human body target; the sample dynamic result is a first result in the sample human body being close to the refrigerator door, a second result in the sample human body being in a stationary state, or a third result in the sample human body being far away from the refrigerator door. The sample intent prediction unit is used to perform intent prediction processing on the sample trajectory results and the sample dynamic results based on the intent prediction network of the preset model, so as to obtain the sample intent prediction result of the human target.

[0098] In some embodiments, the preset model further includes a human category prediction network, and the apparatus further includes: The sample category prediction module is used to perform human category prediction processing on the human body morphology of the sample based on the human body category prediction network to obtain the human body category of the sample. The sample intent prediction unit includes: The first intent prediction subunit is used to predict the intent of the sample trajectory result based on the intent prediction network to obtain the first sample intent. The second intent prediction subunit is used to predict the intent of the sample dynamic results based on the intent prediction network to obtain the second sample intent. The third intent prediction subunit is used to predict the intent of the sample human body category based on the intent prediction network to obtain the third sample intent. The sample intent determination subunit is used to determine the sample intent prediction result of the human target based on the first sample intent, the second sample intent, and the third sample intent.

[0099] In some embodiments, the radar array is disposed on the outer surface of the refrigerator door, and the radar array includes at least two radar groups, wherein each radar group includes at least two radars; the radars in the radar array are distributed on the outer surface of the refrigerator door, and the real-time human motion determination module includes: The limb shape determination unit is used to determine the head shape, arm shape, and leg shape of the human target based on the limb shape of each human limb part detected by the radar. A real-time human body shape determination unit is used to determine the real-time human body shape of the human body target based on the head shape, arm shape and leg shape of the human body target. The limb movement determination unit is used to determine the head movement, arm movement, and leg movement of the human target based on the limb movements of each human limb part detected by the radar. The real-time human motion determination unit is used to determine the real-time human motion of the human target based on the head motion, arm motion, and leg motion of the human target.

[0100] In some embodiments, the radar array includes a first radar group and a second radar group, each radar group including two radars. The radars in the first radar group are distributed on both sides of the top of the refrigerator door, and the radars in the second radar group are distributed on both sides of the bottom of the refrigerator door. The limb morphology determination unit includes: An arm shape detection subunit is used to detect the head shape and arm shape of the human target based on the first radar group. A leg shape detection subunit is used to detect the leg shape of the human target based on the second radar group; The limb movement determination unit includes: An arm movement detection subunit is used to detect the head and arm movements of the human target based on the first radar group. The leg movement detection subunit is used to detect the leg movements of the human target based on the second radar group.

[0101] In some embodiments, the limb information determination module includes: A real-time signal acquisition unit is used to acquire the real-time detection signal of each radar in the radar array when the motion trajectory indicates that the human target is approaching the refrigerator door; The noise reduction unit is used to filter the real-time detection signal of each radar, remove noise from the real-time detection signal, and obtain a noise-reduced signal. The limb movement determination unit is used to determine the limb shape and limb movement of each human limb part detected by each radar based on the denoised signal corresponding to each radar.

[0102] In some embodiments, a current loop is further provided on the outer side of the refrigerator door, and the device further includes: An obstacle detection module is used to detect the resistance of the refrigerator door through the current loop and determine whether an obstacle exists based on the resistance detection result. The reverse braking module is used to trigger reverse braking against the obstacle within a preset time period when the refrigerator door is detected to encounter an obstacle.

[0103] The apparatus and method embodiments described herein are based on the same inventive concept.

[0104] This specification provides an electronic device including a processor and a memory. The memory stores at least one instruction or at least one program, which is loaded and executed by the processor to implement the refrigerator door control method provided in the above method embodiments.

[0105] Embodiments of this application also provide a computer storage medium, which can be disposed in a terminal to store at least one instruction or at least one program related to implementing a refrigerator door control method in the method embodiment. The at least one instruction or at least one program is loaded and executed by the processor to implement the refrigerator door control method provided in the above method embodiment.

[0106] Embodiments of this application also provide a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the refrigerator door control method provided in the above-described method embodiments.

[0107] Optionally, in the embodiments of this specification, the storage medium may be located at at least one of the multiple network servers in a computer network. Optionally, in this embodiment, the storage medium may include, but is not limited to, various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0108] The memory described in the embodiments of this specification can be used to store software programs and modules. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for the functions, etc.; the data storage area may store data created according to the use of the device, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, the memory may also include a memory controller to provide the processor with access to the memory.

[0109] The refrigerator door control method embodiments provided in this specification can be executed on a mobile terminal, computer terminal, server, or similar computing device. Taking running on a server as an example... Figure 8 This is a hardware structure block diagram of a server for a refrigerator door control method provided in an embodiment of this specification. (As shown...) Figure 8 As shown, the server 800 can vary significantly due to different configurations or performance. It may include one or more Central Processing Units (CPUs) 810 (CPUs 810 may include, but are not limited to, microprocessors (MCUs) or programmable logic devices (FPGAs), a memory 830 for storing data, and one or more storage media 820 (e.g., one or more mass storage devices) for storing application programs 823 or data 822. The memory 830 and storage media 820 may be temporary or persistent storage. The program stored in the storage media 820 may include one or more modules, each module may include a series of instruction operations on the server. Furthermore, the CPU 810 may be configured to communicate with the storage media 820 and execute the series of instruction operations stored in the storage media 820 on the server 800. Server 800 may also include one or more power supplies 860, one or more wired or wireless network interfaces 850, one or more input / output interfaces 840, and / or one or more operating systems 821, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0110] The input / output interface 840 can be used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of server 800. In one example, the input / output interface 840 includes a network interface controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the input / output interface 840 may be a radio frequency (RF) module used for wireless communication with the Internet.

[0111] Those skilled in the art will understand that Figure 8 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, server 800 may also include... Figure 8 The more or fewer components shown, or having the same Figure 8 The different configurations shown.

[0112] As can be seen from the embodiments of the refrigerator door control method, device, equipment, or storage medium provided in this application, this application provides a radar array on the refrigerator door, the radar array including at least two radars. When the radar array detects a human target within a detection area around the refrigerator door, the movement trajectory of the human target is determined based on the radar array. When the movement trajectory indicates that the human target is approaching the refrigerator door, the limb shape and limb movement of the human limb detected by each radar are determined according to the real-time detection signal of each radar in the radar array; wherein, the human limb detected by each radar is different. The real-time human shape of the human target is determined according to the limb shape of the human limb detected by each radar, and the real-time human movement of the human target is determined according to the limb movement of the human limb detected by each radar. The intention recognition result of the human target is determined according to the real-time human shape and the real-time human movement. When the intention recognition result indicates that the human target has the intention to open the door, the refrigerator door is controlled to open.

[0113] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments of this specification have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the embodiments and still achieve the desired result. Additionally, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0114] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0115] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer storage medium, such as a read-only memory, a disk, or an optical disk.

[0116] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A control method of a refrigerator door, characterized by, The refrigerator door is equipped with a radar array, the radar array comprising at least two radars, and the method includes: When the radar array detects a human target within the detection area around the refrigerator door, the movement trajectory of the human target is determined based on the radar array. When the motion trajectory indicates that the human target is approaching the refrigerator door, the limb shape and limb movement of the human limb detected by each radar in the radar array are determined according to the real-time detection signal of each radar; wherein, the human limb detected by each radar is different. Based on the limb shape of each human limb detected by each radar, the real-time human shape of the human target is determined, and based on the limb movement of each human limb detected by each radar, the real-time human movement of the human target is determined. Based on the real-time human form and real-time human movements of the human target, the intent recognition result of the human target is determined; If the intent recognition result indicates that the human target has the intent to open the door, the refrigerator door is controlled to open.

2. The method of claim 1, wherein, The method further includes: The sample human body shape and sample human body action are obtained; the sample human body target is labeled with a sample intent tag; the sample intent tag indicates whether the sample human body target has the intention to open a door; The sample human body shape and sample human body movement are input into a preset model for intent prediction processing to obtain the sample intent prediction result of the sample human body target. Based on the difference between the sample intent prediction result and the sample intent label, the preset model is trained to obtain a human intent recognition model; Accordingly, determining the intent recognition result of the human target based on the real-time human morphology and real-time human movements includes: The real-time human form and real-time human movements of the human target are input into the human intent recognition model for intent recognition processing to obtain the intent recognition result of the human target.

3. The method of claim 2, wherein, The step of inputting the sample human body shape and sample human body movement into a preset model for intent prediction processing to obtain the sample human body target's intent prediction result includes: The sample human body shape and sample human body movement are input into a preset model, and the trajectory prediction network based on the preset model is used to perform trajectory prediction processing on the sample human body shape to obtain the sample trajectory result of the sample human body target. The human motion prediction network based on the preset model performs dynamic prediction processing on the sample human body movements to obtain the sample dynamic result of the sample human body target; the sample dynamic result is a first result where the sample human body is close to the refrigerator door, a second result where the sample human body is in a stationary state, or a third result where the sample human body is far away from the refrigerator door. The intent prediction network based on the preset model performs intent prediction processing on the sample trajectory results and the sample dynamic results to obtain the sample intent prediction result of the human target.

4. The method of claim 3, wherein, The preset model also includes a human category prediction network, and the method further includes: Based on the human body category prediction network, the human body morphology of the sample is subjected to human body category prediction processing to obtain the human body category of the sample. The intent prediction network based on the preset model performs intent prediction processing on the sample trajectory results and the sample dynamic results to obtain the sample intent prediction result of the human target, including: Based on the intent prediction network, the intent of the sample trajectory results is predicted to obtain the first sample intent; Based on the intent prediction network, the intent of the sample dynamic results is predicted to obtain the second sample intent; Based on the intent prediction network, the intent of the sample human body category is predicted to obtain the third sample intent; Based on the first sample intent, the second sample intent, and the third sample intent, the sample intent prediction result of the human target is determined.

5. The method of claim 1, wherein, The radar array is disposed on the outer surface of the refrigerator door, and the radar array includes at least two radar groups, wherein each radar group includes at least two radars; the radars in the radar array are distributed on the outer surface of the refrigerator door; the step of determining the real-time human form of the human target based on the limb shape detected by each radar, and determining the real-time human action of the human target based on the limb movement detected by each radar, includes: Based on the limb shape of each human body part detected by each radar, the head shape, arm shape, and leg shape of the human target are determined. The real-time human form of the human target is determined based on the head shape, arm shape, and leg shape of the human target. Based on the limb movements of each human body part detected by each radar, the head movements, arm movements, and leg movements of the human target are determined. The real-time human motion of the human target is determined based on the head movement, arm movement, and leg movement of the human target.

6. The method of claim 5, wherein, The radar array includes a first radar group and a second radar group, each radar group including two radars. The radars in the first radar group are distributed on both sides of the top of the refrigerator door, and the radars in the second radar group are distributed on both sides of the bottom of the refrigerator door. The step of determining the head shape, arm shape, and leg shape of the human target based on the limb shape detected by each radar includes: The head and arm shapes of the human target are detected by the first radar group. The second radar group detects the leg shape of the human target; The step of determining the head movement, arm movement, and leg movement of the human target based on the limb movements detected by each radar on each human limb part includes: The head and arm movements of the human target are detected by the first radar group. The second radar group detects the leg movements of the human target.

7. The method of claim 1, wherein, When the motion trajectory indicates that the human target is approaching the refrigerator door, the limb shape and limb movement of each human limb detected by each radar in the radar array are determined based on the real-time detection signals of each radar, including: When the motion trajectory indicates that the human target is approaching the refrigerator door, the real-time detection signal of each radar in the radar array is acquired; The real-time detection signal of each radar is filtered to remove noise and obtain a denoised signal. Based on the denoised signal corresponding to each radar, the limb shape and limb movement of the human limb parts detected by each radar are determined.

8. The method of claim 1, wherein, A current loop is also provided on the outside of the refrigerator door, and the method further includes: The resistance of the refrigerator door is detected by the current loop, and the presence of an obstacle is determined based on the resistance detection results. If an obstacle is detected in the refrigerator door, reverse braking is triggered against the obstacle within a preset time period.

9. A control device of a refrigerator door, characterized by, The refrigerator door is equipped with a radar array, the radar array comprising at least two radars, and the device includes: The motion trajectory determination module is used to determine the motion trajectory of the human target based on the radar array when the radar array detects the presence of a human target in the detection area around the refrigerator door. The limb information determination module is used to determine the limb shape and limb movement of each human limb detected by each radar based on the real-time detection signal of each radar in the radar array when the motion trajectory indicates that the human target is close to the refrigerator door; wherein, the human limb detected by each radar is different. The real-time human motion determination module is used to determine the real-time human form of the human target based on the limb form of each human limb detected by each radar, and to determine the real-time human motion of the human target based on the limb motion of each human limb detected by each radar. An intent recognition module is used to determine the intent recognition result of the human target based on the real-time human morphology and the real-time human movements of the human target. The control module is used to control the refrigerator door to open when the intent recognition result indicates that the human target has the intent to open the door.

10. An electronic device, characterized in that, The device includes a processor and a memory, wherein the memory stores at least one instruction or at least one program, the at least one instruction or at least one program being loaded and executed by the processor to implement the refrigerator door control method as described in any one of claims 1-8.

Images