A gesture recognition-based intelligent hair dryer control system and method

Abstract

The application relates to the technical field of intelligent household appliances, and discloses an intelligent hair dryer control system and method based on gesture recognition, the system comprising a visual acquisition module, which is used for acquiring image data streams covering an air outlet area and containing depth information and color information; a main control module; a mechanical arm driving module, which is used for converting the spatial pose instruction into an electric signal to change the position and orientation of the machine body; a wind control module and a nozzle adjusting module, the method comprising the following steps: collecting environmental image data; sending an air flow suppression instruction; controlling the rotating speed of a hair dryer motor; identifying features in the image; mode determination; and interactive end recovery. Through the active air flow suppression mechanism of motion trend detection, the hair blocking caused by high-speed air flow and the light distortion caused by heat convection are physically eliminated, so that the technical problems of many depth sensor data noise points and difficult feature extraction in the strong wind working state are solved, and the accurate input of man-machine interaction instructions in the strong convection environment is ensured.

Description

Technical Field

[0001] This invention relates to the field of smart home appliance technology, specifically to a smart hair dryer control system and method based on gesture recognition. Background Technology

[0002] In daily hair care and styling, users typically need to complete both blow-drying and styling simultaneously. To achieve the desired look, the user often needs to hold styling tools such as curling combs or straighteners in one hand to comb or straighten the hair, while the other hand holds the hairdryer and moves it to provide a continuous flow of hot air. This two-handed operation requires a high degree of coordination, and because hairdryers are usually quite heavy, holding them for extended periods can easily lead to arm muscle fatigue, making it difficult to maintain a stable airflow angle and distance, thus affecting the styling result.

[0003] To address the issue of hand fatigue, various hair dryer stands or upright hair dryers have emerged as existing technological solutions. While these devices physically free the user's hands, their air outlet positions are usually fixed or only allow for simple mechanical oscillation. When the user changes head posture or moves styling tools, the fixed airflow cannot follow the operating area in real time, forcing the user to constantly adjust their body position to adapt to the airflow, reducing the flexibility and comfort of use.

[0004] With the development of computer vision technology, some home appliances have begun to incorporate gesture recognition technology to achieve contactless control. However, directly applying visual recognition technology to hair dryers presents unique engineering challenges. Hair dryers generate high-speed airflow during operation, causing hair to move erratically. This high-frequency dynamic obstruction severely interferes with the camera's ability to capture key hand points or facial features. Furthermore, the hot airflow from the vent causes changes in local air density, resulting in uneven light refraction and distortion of the acquired image data. Most existing visual control solutions do not consider the interference of this strong convection environment on sensor data, leading to a significant drop in gesture recognition accuracy when the hair dryer is on, and even false triggering, failing to meet the continuous and precise control requirements of real-world styling scenarios. Finally, existing nozzle adjustments are discrete and lack air pressure coordination, easily causing motor overload and airflow noise, impacting the user experience. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent hair dryer control system and method based on gesture recognition, which solves the problems that the visual recognition of hair dryer control systems in strong convection environments is easily interfered with by flying hair and hot airflow, and that it is difficult to achieve accurate and coordinated following when styling complex hair.

[0006] To achieve the above objectives, the first aspect of the present invention provides an intelligent hair dryer control system based on gesture recognition, including a vision acquisition module, a main control module, a robotic arm drive module, a wind control module, and a nozzle adjustment module.

[0007] The visual acquisition module is used to acquire image data streams containing depth and color information covering the air outlet area. The main control module is connected to the other modules and is used to detect motion indicators based on the image data stream, send airflow suppression commands when the motion indicators exceed the limits, and generate spatial pose commands and shape control commands based on extracted hand or tool features under airflow suppression conditions. The robotic arm drive module is connected to the support structure and is used to convert the spatial pose commands into motor electrical signals to change the position and orientation of the robot. The wind control module is connected to the fan and heating components and is used to adjust the speed and power, and execute a sudden drop in wind speed in response to the airflow suppression commands. The nozzle adjustment module is connected to the variable cross-section mechanism and is used to convert the shape control commands into driving actions to change the effective cross-sectional diameter of the air outlet.

[0008] To address the issue of strong airflow interfering with visual recognition, the system employs an active airflow suppression mechanism. The main control module performs differential calculations on consecutive frame images to obtain the motion indicators. The wind control module is configured with variable frequency anti-interference logic: upon receiving the airflow suppression command, it controls the fan motor to reduce its operating speed to a maintenance speed. The maintenance speed is defined as the speed at which the motor maintains the minimum rotational inertia without generating enough airflow to disturb the hair. When the main control module determines that the interaction has ended, the wind control module controls the fan motor's speed to increase from the maintenance speed back to the operating speed according to an S-shaped speed planning curve, thus smoothly restoring the operating state.

[0009] To improve the quality of visual data, the visual acquisition module uses a pre-calibrated extrinsic parameter matrix to map the acquired depth image frames to the coordinate system of the two-dimensional color image frames, generating an aligned image. Subsequently, the system performs statistical outlier removal filtering on the point cloud data of the aligned image. This filtering process includes calculating the average distance between each pixel and a preset number of its nearest neighbors, and marking pixels whose average distance is greater than the sum of the global average distance and the standard deviation as noise points and removing them.

[0010] In terms of control strategy, the main control module performs multimodal logic arbitration: when only the hand feature is detected in the field of view, or the tool feature is detected but no locking gesture is detected, it enters the direct gesture guidance mode; when the hand feature and the tool feature are detected in the field of view at the same time, and the spatial distance between the two is less than the association threshold or the hand presents a directional gesture, it enters the tool cooperative following mode.

[0011] In the tool-coordinated following mode, the system can automatically accompany the hair dryer to the styling tool. The main control module performs six-degree-of-freedom pose estimation on the styling tool corresponding to the tool features to obtain its current pose matrix. Subsequently, the main control module calculates the target pose matrix at the end of the robotic arm drive module by performing matrix multiplication between the current pose matrix and a preset relative pose offset matrix, so as to maintain the relative position lock between the hair dryer and the styling tool.

[0012] Based on the calculated target pose matrix (including target position and target attitude), the robotic arm drive module calls the inverse kinematics solver to construct a set of kinematic equations and solve for the joint angle vectors of each joint motor. During the solution process, if multiple solutions exist, the system selects the solution with the smallest difference from the current joint angle as the final control command based on the principle of minimum movement cost.

[0013] In the direct gesture guidance mode, the system controls the size of the air outlet through gesture patterns. The main control module extracts the geometric center of the palm and the fingertips from the hand features and calculates the gesture expansion degree.

[0014] The main control module uses a linear mapping algorithm to calculate the target physical diameter corresponding to the shape control command. This algorithm establishes a positive correlation between the target physical diameter and the gesture expansion degree; that is, when the gesture expansion degree increases, the target physical diameter increases proportionally.

[0015] In addition, to ensure airflow stability, the system is also equipped with wind pressure coordinated control logic: when the shape control command drives the air outlet to perform a contraction action, the main control module synchronously generates a control signal to reduce the speed of the fan motor; when the shape control command drives the air outlet to perform an expansion action, the main control module synchronously generates a control signal to increase the speed of the fan motor.

[0016] A second aspect of the present invention provides a smart hair dryer control method based on gesture recognition, which is applied to the above-mentioned system and includes the following steps:

[0017] The visual acquisition module continuously collects environmental image data and transmits it to the main control module;

[0018] The main control module calculates motion indicators in the image data, and when the motion indicators exceed the action threshold, it sends an airflow suppression command to the wind control module.

[0019] In response to the airflow suppression command, the wind control module controls the fan motor speed to decrease to a maintenance speed, thereby eliminating airflow interference and providing a stable visual acquisition environment;

[0020] During the period of speed reduction, the main control module identifies hand and tool features in the image;

[0021] If the system is determined to enter the direct gesture guidance mode, the main control module controls the nozzle adjustment module to change the cross-sectional area of ​​the air outlet according to the gesture expansion of the hand features, and controls the robotic arm drive module to adjust the position of the machine body according to the spatial displacement of the hand features.

[0022] If the tool collaborative following mode is entered, the main control module calculates the current pose matrix of the modeling tool and solves the target pose matrix of the end effector of the robotic arm according to the preset offset matrix, and controls the robotic arm drive module to follow the movement of the modeling tool.

[0023] After the interaction ends, the main control module sends a recovery command to the wind control module, which then controls the wind turbine motor speed to increase.

[0024] This invention provides a smart hair dryer control system and method based on gesture recognition. It has the following beneficial effects:

[0025] 1. This invention employs an active airflow suppression mechanism based on motion trend detection, enabling the system to instantly reduce the fan speed to a maintenance speed when it detects an interactive action trend. This physically eliminates hair shading caused by high-speed airflow and light distortion caused by thermal convection, thereby solving the technical problems of high noise in depth sensor data and difficulty in feature extraction in existing visual hair dryer systems under strong wind conditions. This ensures accurate input of human-computer interaction commands in strong convection environments.

[0026] 2. This invention uses a visual acquisition module to perform six-degree-of-freedom pose estimation on the styling tool and combines this with inverse kinematics calculations of the robotic arm to achieve spatial collaborative following between the hair dryer and the styling tool. Simultaneously, by utilizing a relative pose offset matrix, the air outlet of the hair dryer is logically locked to a specific operation vector of the tool, achieving automated control of the airflow following the comb. This solves the problem of users having to hold both a heavy hair dryer and styling tool simultaneously when performing complex styles, resulting in high operational difficulty and difficulty in maintaining the optimal blowing distance, thus improving operational convenience.

[0027] 3. This invention establishes a linear mapping model between the gesture expansion degree and the nozzle opening and closing parameters, and configures wind pressure coordinated control logic to achieve non-contact continuous adjustment of airflow pattern. This allows for real-time calculation of the target air outlet cross-section based on the degree of hand opening, and simultaneous fine-tuning of motor speed commands based on the fluid model while driving the variable cross-section mechanism. This solves the control problem that simply changing the physical nozzle cross-section can lead to sudden changes in wind resistance, as well as motor overload or airflow whistling. Under the premise of ensuring fluid dynamic stability, a smooth transition from diffused air to concentrated air is achieved, ensuring the continuity of operation. Attached Figure Description

[0028] Figure 1 This is a system module architecture diagram of the present invention;

[0029] Figure 2 This is a flowchart of the method steps of the present invention.

[0030] Among them, 10 is the vision acquisition module; 20 is the main control module; 30 is the robotic arm drive module; 40 is the wind control module; and 50 is the air nozzle adjustment module. Detailed Implementation

[0031] Please see the appendix Figure 1 This invention provides an intelligent hair dryer control system based on gesture recognition, including: a vision acquisition module 10, a main control module 20, a robotic arm drive module 30, a wind control module 40, and a nozzle adjustment module 50.

[0032] The main control module 20 serves as the data processing and command distribution center, establishing signal connections with the vision acquisition module 10, the robotic arm drive module 30, the wind control module 40, and the air nozzle adjustment module 50, respectively. It receives image data from the vision acquisition module 10 and generates control commands based on a preset algorithm, sending these commands to the robotic arm drive module 30, the wind control module 40, and the air nozzle adjustment module 50, respectively.

[0033] The visual acquisition module 10 is mounted on the working end of the hair dryer or on an associated support structure. Its acquisition range covers the target area in the direction of the hair dryer's airflow. It is used to acquire image data streams containing depth and color information in real time and transmit the image data streams to the main control module 20. The visual acquisition module 10 uses a depth camera component or a time-of-flight sensor component to acquire spatial three-dimensional data.

[0034] The robotic arm drive module 30 is connected to the multi-degree-of-freedom mechanical structure that supports the hair dryer body. It receives position and attitude signals sent by the main control module 20 and converts them into electrical signals to drive the motor, thereby changing the position coordinates of the hair dryer body in the spatial coordinate system and the direction of the air outlet.

[0035] The air control module 40 is connected to the fan motor and heating element inside the blower body, and is used to adjust the speed of the fan motor and the power of the heating element. The air control module 40 receives airflow control commands sent by the main control module 20, which include the wind speed setpoint under normal operating conditions and the instantaneous suppression signal at the interaction time.

[0036] The nozzle adjustment module 50 is connected to the variable cross-section mechanism at the air outlet of the blower. The nozzle adjustment module 50 receives shape control commands sent by the main control module 20 and drives the variable cross-section mechanism to perform physical opening and closing movements to change the geometric diameter or air outlet cross-sectional area.

[0037] The specific workflow of the smart hair dryer control system based on gesture recognition, constructed using the above modules, is as follows:

[0038] The visual acquisition module 10 is in monitoring mode, continuously transmitting environmental image data to the main control module 20. The main control module 20 performs differential calculations or optical flow calculations on consecutive frame images to detect the displacement trend of moving targets in the images.

[0039] When the main control module 20 detects a moving object within the field of view that meets a preset speed threshold, it determines that an interactive trigger state has been established. At this time, the main control module 20 prioritizes sending an airflow suppression command to the wind control module 40. In response to the airflow suppression command, the wind control module 40 controls the fan motor speed to decrease to a preset low-energy consumption threshold or to pause, eliminating the interference of strong airflow on hair and light within the field of view and constructing a clear visual acquisition window.

[0040] During the visual acquisition window, the main control module 20 extracts features from the images acquired by the visual acquisition module 10. The main control module 20 identifies the target types in the images, including human hand features and tool features.

[0041] If the main control module 20 recognizes human hand features but not the features of the shaping tool, or if the recognized gesture command is an independent control command, the main control module 20 calculates the spatial position and gesture shape of the hand features. Based on the gesture shape, the main control module 20 calculates the corresponding nozzle opening and closing parameters and sends a shape control command to the nozzle adjustment module 50; simultaneously, based on the spatial displacement of the gesture, it calculates the target pose of the robotic arm and sends a motion control command to the robotic arm drive module 30.

[0042] If the main control module 20 simultaneously recognizes both human hand features and styling tool features, and the gesture command is determined to be a collaborative locking command, the main control module 20 calculates the spatial position vector and main axis direction vector of the styling tool. Based on a preset following strategy, the main control module 20 generates relative pose parameters and sends a following control command to the robotic arm drive module 30, ensuring that the blower outlet remains locked in relative position with the styling tool.

[0043] When the robotic arm drive module 30 and the nozzle adjustment module 50 complete their actions, and the main control module 20 determines that the interaction has ended, the main control module 20 sends a recovery command to the wind control module 40. The wind control module 40 then controls the fan motor speed to return to the user-set operating value.

[0044] The visual acquisition module 10, serving as the information sensing front end of the intelligent hair dryer control system, is responsible for constructing a three-dimensional digital environment for human-computer interaction. Specifically, the visual acquisition module 10 employs an RGB-D depth camera component, which integrates a color image sensor and a depth sensor. In specific implementations, the depth sensor can be a time-of-flight sensor or a structured light sensor to adapt to different lighting conditions in bathrooms or barbershops. The specific photoelectric conversion principles and hardware circuit structures of the aforementioned sensors are well-known to those skilled in the art and will not be elaborated upon here.

[0045] The vision acquisition module 10 is fixedly mounted at the end of the robotic arm of the blower system or at a high point on an independent support, with its optical axis pointing towards the user's work area. To ensure the accuracy of subsequent control commands, the vision acquisition module 10 performs the following data acquisition and preprocessing procedures after the system starts:

[0046] Step S110: The visual acquisition module 10 simultaneously triggers the color image sensor and the depth sensor to acquire the two-dimensional color image frame and the depth image frame of the current field of view, respectively.

[0047] Two-dimensional color image frames contain texture and color information, which are used for subsequent object recognition and semantic segmentation; depth image frames contain physical distance information of each pixel in the field of view from the camera, which are used for spatial positioning.

[0048] Step S120: Due to the baseline distance between the color sensor and the depth sensor in their physical locations, there is a deviation in their viewing angles. The vision acquisition module 10 uses a pre-calibrated extrinsic parameter matrix to map the depth pixels in the depth image frame to the coordinate system of the two-dimensional color image frame, generating an RGB-D aligned image. This process ensures that the same pixel coordinate in the image simultaneously possesses the correct color and depth attributes.

[0049] Step S130: To achieve precise guidance of the robotic arm, the pixel coordinates of the image plane need to be converted into three-dimensional spatial coordinates in the camera coordinate system. The vision acquisition module 10 or the main control module 20 connected to it performs reverse calculation based on the pinhole imaging model. For any point in the image pixel coordinate system... Its corresponding depth value Then the point is in the camera coordinate system spatial coordinates below The calculation is as follows:

[0050] ;

[0051] In the formula, The physical distance of the target point along the optical axis of the camera; and These are the physical coordinates of the target point along the horizontal and vertical axes of the camera coordinate system, respectively. The depth reading is the depth value directly measured by the depth sensor. and These are the x and y coordinates in the pixel coordinate system; and The principal point coordinates of the camera are the projection center of the optical axis onto the image plane. and This refers to the equivalent focal length of the camera in both the horizontal and vertical directions. (The above...) , , , These are all intrinsic parameters of the camera, obtained from factory calibration.

[0052] Step S140: Considering that the working environment of the hair dryer may contain water mist or hot airflow disturbances, causing speckle noise or edge abrupt changes in the depth data, the visual acquisition module 10 begins to perform statistical outlier removal filtering on the generated point cloud data. It calculates the outlier removal values ​​for each point and its neighborhood. If the average distance between the nearest neighbors is greater than a preset multiple of the sum of the global average distance and the standard deviation, then the point is marked as a noise point and removed.

[0053] This process effectively removes floating noise caused by light refraction or sensor errors, improving the robustness of subsequent gesture recognition.

[0054] After the above processing, the visual acquisition module 10 outputs to the main control module 20 RGB-D data streams or reconstructed 3D point cloud data that have been registered and denoised. This data stream constitutes the basic information source for the system to understand user intentions and environmental states. Furthermore, the visual acquisition module 10 is also equipped with a dynamic region of interest setting function. Based on the positions of the human head and hands detected in the previous frame, it crops out a local area containing the upper body of the human body in the current frame for focused scanning, thereby reducing data transmission bandwidth and improving the real-time response speed of the system.

[0055] The air control module 40 is the execution unit in the intelligent hair dryer control system responsible for airflow regulation and heat management. Physically, the air control module 40 is located between the main control module 20 and the brushless DC motor and heating wire assembly inside the hair dryer. It is not only responsible for constant airflow control and PID temperature control during routine drying operations, but its core function lies in its active airflow suppression mechanism in response to interactive requests, addressing the technical problem of visual recognition failure in strong convection environments.

[0056] The wind control module 40 has a built-in motor drive circuit and power modulation circuit.

[0057] For the motor drive circuit, this embodiment uses sinusoidal vector control or pulse width modulation technology to drive the brushless motor to achieve millisecond-level speed response; for the power modulation circuit, thyristor voltage regulation or solid-state relay is used to control the on / off duty cycle of the heating element.

[0058] The hardware topology of the aforementioned basic drive circuit is well-known in the field. However, in this control system, the wind control module 40 operates with specific frequency conversion anti-interference logic, and specifically executes the following control flow:

[0059] Step S210: The risk control module 40 listens in real time to the interaction status flag from the main control module 20. When the main control module 20 detects that the motion index within the field of view exceeds a preset action threshold based on visual data, it sends an airflow suppression command to the risk control module 40. This action threshold represents the trend of the user's intention to send gesture commands.

[0060] Step S220: In response to the airflow suppression command, the wind control module 40 immediately cuts off or reduces the drive voltage output to the motor, so that the fan speed drops rapidly from the current operating speed to the maintenance speed.

[0061] The sustaining speed is defined as the speed at which the motor maintains its minimum moment of inertia without generating a weak airflow sufficient to disturb a hair, or, in a specific mode, it is zero.

[0062] Specifically, maintaining the rotational speed setting must satisfy the following two physical constraints:

[0063] First, the lower limit constraint: the speed needs to be slightly higher than the dead zone speed of the motor, that is, the minimum continuous rotational speed that can overcome the static friction of the motor bearings and the cogging torque. It is usually set to 3% to 5% of the rated speed of the motor to ensure that the motor is in a hot standby state and avoid the time lag of hundreds of milliseconds caused by starting from zero.

[0064] Secondly, the upper limit constraint: the air velocity generated at the outlet at this speed should be less than the hair-starting air velocity threshold. According to fluid dynamics experiments, this air velocity threshold is usually set to 0.5m / s to 1.2m / s (depending on the cross-sectional area of ​​the outlet), and the corresponding motor speed is usually no more than 10% to 15% of the rated speed.

[0065] For example, for a high-speed brushless motor with a rated speed of 110,000 rpm, the maintaining speed can be set to 5,000 rpm to 8,000 rpm. Within this range, the airflow is in a laminar state and cannot overcome the gravity and static friction of the hair, thus ensuring that the hair strands in the field of view are in a relatively static state.

[0066] To quantify the control logic, the risk control module 40 dynamically adjusts the target rotational speed based on the received motion indicators. The control function is described below:

[0067] ;

[0068] In the formula, The target speed command is output by the wind control module 40 to the motor drive circuit. This is a preset low-energy-consumption, anti-interference speed, and its value is typically 5% to 10% of the rated speed. The normal drying speed is preset by the user via gestures or a physical knob; The motion index of the current frame image is calculated and input by the main control module 20; A threshold for determining whether a user is making large gestures.

[0069] Based on the above logic, the system forcibly reduces the wind speed the moment it detects the user's waving intention, physically eliminating two types of interference sources:

[0070] First, the disorderly flying of hair caused by high-speed airflow obscures the face and hands;

[0071] Secondly, the change in the refractive index of ambient light and the accumulation of water mist caused by hot air.

[0072] This process creates a brief moment of clear vision, enabling the visual acquisition module 10 to acquire clear and stable depth images for subsequent high-precision gesture calculation.

[0073] Step S230: Simultaneously with the sudden drop in airflow, the logic circuit inside the air control module 40 triggers a thermal linkage protection mechanism. To prevent the reduced airflow from causing the heating element to overheat and burn out, the air control module 40 proportionally reduces the heating power of the heating wire based on the current target rotation speed, or directly cuts off the heating circuit. The linkage between heating power and airflow satisfies thermal balance constraints, ensuring that the outlet temperature does not experience a momentary overshoot.

[0074] Step S240: After the main control module 20 completes the recognition and execution of the gesture command (such as completing the adjustment of the nozzle or the movement of the robotic arm), it will reset the interaction status flag. Upon detecting the flag reset, the wind control module 40 controls the fan motor speed from... Restore to To avoid sudden changes in wind speed that could startle users, the speed recovery process uses an S-shaped speed planning curve for smooth loading, allowing the wind speed to gently rise to the set value within a preset time window (e.g., 0.5 to 1.0 seconds).

[0075] The wind control module 40 also features a timeout forced recovery function. If the system enters an airflow suppression state for more than a preset safety time limit (e.g., 3 seconds) and fails to complete effective gesture recognition, the wind control module 40 will forcibly restore the wind speed to prevent prolonged shutdown due to accidental triggering from affecting drying efficiency. This mechanism ensures the system's continuous operation under extreme interference conditions.

[0076] The main control module 20 is the computing and decision-making center of the intelligent hair dryer control system. It is mainly responsible for processing visual data, interpreting user intentions, and generating control commands for lower-level actuators. At the hardware level, the main control module 20 is built on a high-performance embedded system-on-chip (SoC), which integrates a central processing unit and a neural network acceleration unit to support real-time image processing and deep learning inference tasks.

[0077] The main control module 20 internally runs parallel processing logic, specifically including a motion trend detection subroutine, a feature extraction and classification subroutine, and a multimodal logic arbitration subroutine. The specific operation steps of this processing logic are as follows:

[0078] Step S310: During standby or routine monitoring, the main control module 20 performs inter-frame difference calculation on two consecutive grayscale images input from the vision acquisition module 10. This step aims to quickly assess the degree of dynamic change within the field of view without involving specific object recognition, thereby ensuring extremely low computational latency.

[0079] The main control module 20 calculates global motion metrics, and the calculation logic is as follows:

[0080] ;

[0081] In the formula, The number of pixels representing the image width; The number of pixels representing the image height; For the current moment In pixel coordinates The grayscale value at that location; For the previous moment Gray values ​​at the same coordinates; This is a global motion index.

[0082] When the calculation results When the preset action threshold is exceeded, the main control module 20 determines that there is a clear action trend and immediately sends the highest priority airflow suppression command to the wind control module 40 to trigger the aforementioned wind speed drop logic, thereby creating an interference-free environment for subsequent high-precision identification.

[0083] Step S320: After airflow suppression takes effect and the field of view is clear, the main control module 20 calls a pre-trained deep neural network model to perform semantic analysis on the RGB-D data of the current frame. This analysis process outputs two sets of feature data in parallel:

[0084] The first set consists of hand skeleton features (hand features), which is a set of key points extracted by the system from the human hand within the field of view, including the coordinates of the wrist, palm, and the knuckles of the five fingers. For the detection algorithm of hand key points, an hourglass network (or a similar human pose estimation model) can be used. The set of key points is represented as a three-dimensional coordinate sequence:

[0085] ;

[0086] In the formula, This is a set of features representing the skeletal structure of the hand. Indicates the first The three-dimensional coordinates of each hand joint in the camera coordinate system .

[0087] The second group consists of tool object features (tool features). The system uses object detection algorithms (such as YOLO or SSD network) to identify whether there are styling tools from the pre-set library within the field of view, such as curling combs, straighteners, or hairsprays. If a styling tool is detected, the main control module 20 outputs the bounding box information of the tool and the spatial position of the tool center and the principal axis direction vector.

[0088] Step S330: The main control module 20 performs logical arbitration between the "direct gesture guidance mode" and the "tool collaborative follow mode" based on the feature combination extracted in step S320.

[0089] If only valid hand skeletal features are detected in the field of view, and no shaping tool is detected, or if a shaping tool is detected but the hand does not make a specific locking gesture, the main control module 20 enters the direct gesture guidance mode. In this mode, the system maps the spatial displacement of the hand to the displacement command of the robotic arm, and maps the shape change of the gesture (such as opening or clenching a fist) to the nozzle adjustment command.

[0090] If both hand skeletal features and styling tool features are detected simultaneously in the field of view, and the spatial distance between the hand and the styling tool is less than a preset association threshold, or the hand displays a specific directional gesture (e.g., the index finger points to the tool), the main control module 20 determines that the user's intention is to use the tool for styling, and thus switches to the tool-assisted following mode. In this mode, the main control module 20 ignores the hand's displacement commands and instead locks the position and direction of the styling tool, calculating the ideal following trajectory of the hair dryer's air outlet relative to the styling tool.

[0091] Through the above logic, the main control module 20 realizes the transformation from a single remote control logic to an intelligent assistant logic. It can automatically adjust the control strategy according to whether there are tools in the scene and the user's interactive gestures, and output the corresponding posture control parameters to the robotic arm drive module 30 and the shape parameters to the nozzle adjustment module 50.

[0092] The nozzle adjustment module 50 is the actuator in the system used to change the physical shape of the airflow. Its control structure is located at the air outlet of the fan and employs a conventional electrically operated variable cross-section structure, such as an electrically operated aperture adjustment mechanism or an electrically operated guide vane adjustment mechanism. The nozzle adjustment module 50 receives digital control signals from the main control module 20 and converts these signals into stepping pulses or voltage signals for the drive motor, thereby driving the aforementioned conventional variable cross-section structure to physically open and close, changing the effective cross-sectional diameter of the air outlet.

[0093] The nozzle adjustment module 50, main control module 20, and air control module 40 form a linkage control loop, achieving real-time mapping between user gestures and physical nozzle shapes through the following control logic:

[0094] Step S410: In direct gesture guidance mode, the main control module 20 locates the geometric center of the user's palm and the positions of the fingertips in real time based on the 3D point cloud data input from the vision acquisition module 10. To convert qualitative gesture shapes (such as open or clenched fists) into quantitative control variables, the system calculates the gesture expansion degree. The calculation logic is as follows:

[0095] ;

[0096] In the formula, This refers to the gesture expansion, and its unit is consistent with the unit of the spatial coordinate system. The number of valid fingers identified is typically 5; For the first The three-dimensional coordinate vectors of the fingertips in the camera coordinate system; The three-dimensional coordinate vector of the geometric center of the palm; Let represent the Euclidean norm of the vector. This formula numerically characterizes the degree of hand opening by calculating the average radial distance between the fingertips and the palm.

[0097] Step S420: The main control module 20 calculates the gesture expansion degree. The target opening and closing parameters of the nozzle adjustment module 50 are calculated using a linear mapping algorithm. To create an intuitive interactive experience, the system sets a positive correlation between the gesture expansion degree and the air outlet diameter; that is, the larger the palm is opened, the larger the air outlet cross-section. The formula for calculating the target diameter is:

[0098] ;

[0099] In the formula, The target physical diameter that the nozzle adjustment module 50 needs to achieve; This is the proportional coefficient, used to adjust the sensitivity of the control. For gesture expansion; The preset minimum gesture expansion threshold corresponds to the user's fist-clenching state; This refers to the foundation diameter or minimum mechanical diameter of the variable cross-section structure. If the calculated value is... Exceeding the physical limit of the nozzle adjustment module by 50 These represent the maximum and minimum values ​​that the nozzle adjustment module 50 needs to adjust, respectively, and the system then truncates them to the corresponding limit values.

[0100] Step S430: Drive execution and wind pressure coordination. The main control module 20 will calculate the target diameter. The corresponding motor control commands are converted and sent to the nozzle adjustment module 50. The nozzle adjustment module 50 then drives the actuator to change the outlet size. Simultaneously, considering that changes in the outlet cross-sectional area will cause changes in fluid back pressure and flow velocity (Bernoulli's principle), the system implements coordinated air pressure control to maintain a consistent blowing experience and protect the fan. When the air outlet shrinks (i.e., the air outlet contracts), the main control module 20 synchronously sends a speed adjustment command to the air control module 40 based on a preset fluid model, appropriately reducing the power output of the fan motor to prevent motor overload or wind noise caused by increased airflow resistance. Conversely, when the air outlet expands, the fan speed is appropriately increased to maintain a constant airflow. This automatic coupling adjustment of the nozzle shape and fan speed ensures the system's operational stability when the physical structure changes.

[0101] The robotic arm drive module 30 is the actuator in the intelligent hair dryer control system that realizes spatial pose adjustment. The robotic arm drive module 30 can independently control the rotation angle of the motors supporting each joint of the robotic arm. The robotic arm drive module 30 is communicatively connected to the main control module 20 and receives data including the target position coordinates. and target attitude angle The spatial pose command is obtained, and the Cartesian space command is converted into angular displacement command for each joint space using an inverse kinematics algorithm.

[0102] To transform the hair dryer from a passive tool into an active styling assistant, the robotic arm drive module 30, under the control of the main control module 20, executes two different path planning strategies based on visual feedback: a basic gesture following strategy and a tool-coordinated following strategy. The tool-coordinated following strategy is the core control logic of this system and specifically includes the following processing steps:

[0103] Step S510: When the system enters the tool collaborative following mode, the main control module 20 no longer simply tracks the hand position, but instead performs real-time six-degree-of-freedom pose estimation on the styling tools (such as curling combs and straighteners) within the field of view of the vision acquisition module 10. The main control module 20 establishes a local coordinate system for the tool and calculates the transformation matrix of this coordinate system relative to the camera coordinate system. Specifically, the system extracts the geometric center position of the styling tool and the principal axis vector representing the tool's operation direction (such as the central axis direction of a round brush).

[0104] Step S520: The system has a pre-set relative pose constraint model between the air outlet and the styling tool. This model defines the ideal spatial relationship that the hair dryer's air outlet should maintain relative to the styling tool (e.g., the air outlet is always located 10 cm directly above the comb, and the airflow direction is perpendicular to the comb's plane). The main control module 20 calculates the target pose matrix of the robotic arm's end effector (i.e., the hair dryer's air outlet) at the next moment based on the real-time acquired tool pose and the relative pose constraint model. The calculation logic satisfies the following transformation relationship:

[0105] ;

[0106] In the formula, This is the homogeneous transformation matrix of the target at the end of the robotic arm in the camera coordinate system, which includes the target position vector and the target rotation matrix; This is the current pose matrix of the modeling tool in the camera coordinate system, detected in real time. The preset relative pose offset matrix is ​​set by the system according to styling requirements (such as straightening or curling hair), and represents the fixed spatial constraint of the blower relative to the tool.

[0107] The matrix operations described above can be decomposed into position control formulas, which can be expressed as:

[0108] ;

[0109] In the formula, The target's three-dimensional coordinate vector at the end of the robotic arm; The three-dimensional coordinate vector of the center of the modeling tool; This is the current rotation matrix of the modeling tool, used to describe the tool's orientation. A constant distance vector defined in the tool's local coordinate system, such as a vector This indicates that the tool is positioned 100mm away from the local Y-axis.

[0110] Through the above calculations, no matter how the user moves or rotates the styling tool in their hand, the system can calculate in real time the accompanying position that the hair dryer should reach, thus realizing the logical locking of the air vent moving together with the comb tool at the algorithm level.

[0111] Step S530: Inverse kinematics transformation and servo drive. Obtain the target pose. Next, the robotic arm drive module 30 needs to calculate the angles that each joint motor should rotate. The system calls the inverse kinematics solver to construct a system of kinematic equations based on the robotic arm's link parameters (DH parameter table) and solve for the joint angle vectors. :

[0112] ;

[0113] In the formula, It is a vector containing the angles of each joint of the robotic arm; This is the inverse kinematics function mapping for the robotic arm; and These are the target position and target attitude calculated in step S520, respectively.

[0114] To address the potential multiple solutions in inverse kinematics, the system employs the principle of minimum movement cost, selecting the solution with the smallest difference from the current joint angle as the final control command. This ensures smooth and rapid response of the robotic arm's movement. The robotic arm drive module 30 then calculates the... The data is sent to the underlying servo driver, which drives the motors of each axis to operate synchronously, enabling the blower body to quickly and accurately reach the calculated accompanying position.

[0115] This technology allows users to style their hair without having to hold and adjust the hair dryer while using a comb. The robotic arm automatically moves the hair dryer to maintain the optimal blowing angle and distance, following the movement of the comb, greatly reducing the difficulty of operation when creating complex styles.

[0116] Please see the appendix Figure 2 This invention provides a smart hair dryer control method based on gesture recognition, comprising the following steps:

[0117] S1. The visual acquisition module 10 continuously acquires environmental image data and transmits it to the main control module 20;

[0118] S2. The main control module 20 calculates the motion index in the image data. When the motion index exceeds the motion threshold, it sends an airflow suppression command to the wind control module 40.

[0119] S3, the wind control module 40 responds to the airflow suppression command and controls the fan motor speed to decrease to the maintenance speed;

[0120] S4. During the period of speed reduction, the main control module 20 identifies hand features and tool features in the image;

[0121] S5. If it is determined that the direct gesture guidance mode has been entered, the main control module 20 controls the nozzle adjustment module 50 to change the cross-sectional area of ​​the air outlet according to the gesture expansion degree of the hand features, and controls the robotic arm drive module 30 to adjust the position of the machine body according to the spatial displacement of the hand features.

[0122] S6. If it is determined that the tool collaborative following mode has been entered, the main control module 20 calculates the current pose matrix of the modeling tool and solves the target pose matrix of the end of the robotic arm according to the preset offset matrix, and controls the robotic arm drive module 30 to follow the movement of the modeling tool.

[0123] S7. After the interaction ends, the main control module 20 sends a recovery command to the wind control module 40, and the wind control module 40 controls the wind turbine motor speed to recover.

[0124] The above method and solution are based on the aforementioned system solution, and the technical details are the same, so they will not be repeated here.

Claims

1. A smart hair dryer control system based on gesture recognition, characterized in that, include: The visual acquisition module (10) is used to acquire an image data stream containing depth and color information covering the air outlet area; The main control module (20) is used to detect motion indicators based on the image data stream, send airflow suppression instructions when the motion indicators exceed the limit, and generate spatial pose instructions and shape control instructions based on the extracted hand or tool features under airflow suppression. The robotic arm drive module (30) is used to convert the spatial pose command into motor electrical signals to change the position and orientation of the robotic body; The wind control module (40) is used to adjust the rotation speed and power, and to execute a sudden drop in wind speed in response to the airflow suppression command; The nozzle adjustment module (50) is used to convert the shape control command into a driving action to change the effective cross-sectional diameter of the air outlet.

2. The intelligent hair dryer control system based on gesture recognition according to claim 1, characterized in that, The adjustment of rotational speed and power, and the execution of a sudden wind speed reduction in response to the airflow suppression command, specifically includes the following steps: When the airflow suppression command is received, the fan motor is controlled to reduce from the operating speed to the maintenance speed, which is the speed value at which the motor maintains the minimum rotational inertia without generating disturbing airflow through the hair strands. After the interaction ends, the speed of the fan motor is controlled to return from the maintenance speed to the operating speed according to the S-shaped speed planning curve.

3. The intelligent hair dryer control system based on gesture recognition according to claim 1, characterized in that, The acquisition of the image data stream containing depth and color information covering the air outlet area specifically includes the following steps: The acquired depth image frames are mapped to the coordinate system of the two-dimensional color image frames using the extrinsic parameter matrix to generate an aligned image; Statistical outlier removal filtering is performed on the point cloud data of the aligned image. The filtering process includes calculating the average distance between each point and a preset number of nearest neighbors, and removing pixels whose average distance is greater than the sum of the global average distance and the standard deviation as noise points.

4. The intelligent hair dryer control system based on gesture recognition according to claim 1, characterized in that, The step of generating spatial pose commands and shape control commands based on extracted hand or tool features specifically includes the following steps: When only hand features are detected in the field of view, or tool features are detected but no locking gesture is detected, the direct gesture guidance mode is entered. When both hand and tool features are detected in the field of view, and the spatial distance between them is less than the association threshold or the hand is making a directional gesture, the tool collaborative following mode is entered. The hand features are the extracted set of key points of the human hand within the field of view, including the coordinates of the wrist, palm, and knuckles of the five fingers. The tool is characterized by using a target detection algorithm to identify whether a modeling tool from a pre-set library exists within the field of view.

5. A smart hair dryer control system based on gesture recognition according to claim 4, characterized in that, The entry tool collaborative follow mode specifically includes the following steps: Perform six-degree-of-freedom pose estimation on the modeling tool corresponding to the tool features to obtain its current pose matrix; By performing matrix multiplication between the current pose matrix and the preset relative pose offset matrix, the target pose matrix at the end of the robotic arm drive module (30) is calculated, thus maintaining the relative position lock between the fan and the modeling tool.

6. The intelligent hair dryer control system based on gesture recognition according to claim 5, characterized in that, The spatial pose command is converted into a motor electrical signal to change the position and orientation of the machine body, specifically including the following steps: Based on the target position and target attitude contained in the target pose matrix, the inverse kinematics solver is called to construct a set of kinematic equations and solve them to obtain the joint angle vectors of each joint motor. If multiple solutions exist during the solution process, the solution with the smallest difference from the current joint angle is selected as the final control command based on the principle of minimum movement cost.

7. The intelligent hair dryer control system based on gesture recognition according to claim 4, characterized in that, Entering the direct gesture guidance mode specifically includes: Extract the geometric center of the palm and the fingertips from the hand features, and calculate the gesture expansion. The calculation steps for the gesture expansion degree are as follows: Calculate the Euclidean distance of each fingertip position relative to the geometric center of the palm, and obtain the arithmetic mean of all distances.

8. The intelligent hair dryer control system based on gesture recognition according to claim 7, characterized in that, The shape control command specifically corresponds to a target physical diameter, which is calculated as follows: A linear mapping algorithm is used to solve the problem, and it is assumed that there is a positive correlation between the target physical diameter and the gesture expansion degree, and that the target physical diameter increases proportionally when the gesture expansion degree increases.

9. A smart hair dryer control system based on gesture recognition according to claim 1, characterized in that, The main control module (20) also includes logic for executing wind pressure coordinated control, the specific steps of which are as follows: When the shape control command drives the air outlet to perform a retraction action, a control signal to reduce the speed of the fan motor is generated simultaneously. When the shape control command drives the air outlet to perform an expansion action, a control signal to increase the speed of the fan motor is generated simultaneously.

10. A smart hair dryer control method based on gesture recognition, characterized in that, An intelligent hair dryer control system based on gesture recognition as described in any one of claims 1-9 includes the following steps: The visual acquisition module (10) continuously acquires environmental image data and transmits it to the main control module (20); The main control module (20) calculates the motion index in the image data. When the motion index exceeds the action threshold, it sends an airflow suppression command to the wind control module (40). In response to the airflow suppression command, the wind control module (40) controls the fan motor speed to decrease to the maintenance speed; During the period of speed reduction, the main control module (20) identifies hand features and tool features in the image; If the system is determined to enter the direct gesture guidance mode, the main control module (20) controls the nozzle adjustment module (50) to change the cross-sectional area of ​​the air outlet according to the gesture expansion degree of the hand features, and controls the robotic arm drive module (30) to adjust the position of the machine body according to the spatial displacement of the hand features. If the tool collaborative following mode is entered, the main control module (20) calculates the current pose matrix of the modeling tool and solves the target pose matrix of the end of the robotic arm according to the preset offset matrix, and controls the robotic arm drive module (30) to follow the modeling tool. After the interaction ends, the main control module (20) sends a recovery command to the wind control module (40), and the wind control module (40) controls the wind turbine motor speed to recover.

Images