Smart home system

Abstract

The invention discloses an intelligent home system, which comprises a main control module, a communication module, a camera module, a fall identification module, a data acquisition module and a sensor module, wherein the camera module collects a picture and transmits the picture to the fall identification module, and the fall identification module performs fall identification and then transmits an identification result to the main control module; the output end of the sensor module is in signal connection with the input end of the data acquisition module, and the output end of the data acquisition module is in signal connection with the input end of the main control module through the communication module. In addition, the invention also comprises a remote control terminal which is used for receiving the abnormal state information of the smart home system. The indoor air quality is monitored through various sensors, meanwhile, whether the elderly fall down at home or not is monitored through an improved fall-down recognition detection algorithm, and abnormal state information is sent to the remote control terminal in time.

Description

Technical Field

[0001] The invention belongs to the field of intelligent detection, and in particular relates to an intelligent home system. Background Art

[0002] With the acceleration of urbanization, indoor air quality issues have become increasingly prominent. Common pollutants include volatile organic compounds, carbon dioxide, and fine particulate matter, which may cause health problems. Traditional air quality monitoring methods rely on regular sampling and experimental analysis, lack of real-time performance, and are difficult to reflect the dynamic changes of air quality in a timely manner. There are problems of delayed response and insufficient early warning. Therefore, some people choose to install smart home systems in their houses to achieve air quality monitoring and other functions such as lighting control and electrical control. Nowadays, some smart home systems also cover fall monitoring functions. This is because the elderly have a higher incidence of falls due to weakened physical functions, which in turn creates a need to monitor the behavior of the elderly. However, the fall detection technology in existing smart home systems mostly relies on a single sensor, with poor detection accuracy and adaptability, prone to false alarms or missed alarms, and delayed alarm mechanisms. There is a lack of linkage with other home functions, and it is impossible to form a comprehensive safety guarantee.

[0003] In addition, although some smart home systems support both air quality monitoring and fall detection functions, there are still some difficulties in integrating the two. On the one hand, air quality and fall detection belong to different monitoring fields, and the data sources are highly heterogeneous, making it difficult to achieve efficient fusion analysis; on the other hand, existing systems mostly adopt independent modular design, and air quality sensors and fall detection modules operate separately, which not only increases hardware costs, but also increases system complexity. In addition, the performance of existing detection algorithms significantly decreases in dynamic environments, further restricting the overall performance of the system.

[0004] Therefore, there is an urgent need for a smart home system with more comprehensive and complete functions. Summary of the invention

[0005] The purpose of the present invention is to provide a smart home system in view of the deficiencies of the current technology.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions: A smart home system, comprising: a main control module, a communication module, a camera module, a fall recognition module, a data acquisition module and a sensor module; wherein the camera module collects images and transmits them to the fall recognition module, and the fall recognition module transmits the recognition result to the main control module after performing fall recognition; the output end of the sensor module is signal-connected to the input end of the data acquisition module, and the output end of the data acquisition module is signal-connected to the input end of the main control module through the communication module; The sensor module includes a temperature and humidity sensor, a carbon dioxide sensor, a formaldehyde sensor, a light intensity sensor, a dust particle sensor and a TVOC (Total Volatile Organic Compounds) gas sensor.

[0007] Preferably, the temperature and humidity sensor, the light intensity sensor and the TVOC sensor transmit and receive data with the data acquisition module through a synchronous serial communication protocol, the carbon dioxide sensor and the formaldehyde sensor transmit and receive data with the data acquisition module through an asynchronous serial communication protocol, and the dust particle sensor transmits and receives data with the data acquisition module through a data line.

[0008] Preferably, the camera module and the main control module are connected via a CSI interface.

[0009] Preferably, the smart home system further comprises a remote control terminal, and the main control module is connected to the remote control terminal by signal. The fall recognition module receives the image captured by the camera module, classifies it through the open source posture prediction model MoveNet posture recognition, and then determines whether a break occurs based on whether the linear velocity of the center of gravity of the trunk, the linear velocity of the center of gravity of the face, and the angular velocity of the ankle reach the expected threshold, and the change in the horizontal angle within the time of reaching the threshold, and finally sends the detection result to the remote control terminal in time, thereby improving the accuracy of fall recognition.

[0010] Preferably, the fall recognition module proposes a fall recognition detection algorithm based on Google's open source posture prediction model MoveNet, collects image data through the camera module, and feeds back the fall behavior detection results to the main control module and the remote control terminal to realize automatic alarm for fall behavior. The specific process is as follows: S1: After the image data collected by the camera is inferred by the MoveNet model, the key point information of the human body is obtained, and the key points are preliminarily classified using a fully convolutional neural network into two categories: normal posture key points and possible fall posture key points; S2: When judging whether a fall has occurred, the coordinate information of the key points of the possible fall posture is combined. , Represents the x-axis and y-axis coordinate values ​​of the key point of the i-th human body posture, when the trunk center of gravity linear velocity is satisfied >1.5m / s and linear speed of facial key points >1.2m / s, and the average angular velocity of key points 10 and 13 >200° / s, the instantaneous fall characteristics are met; among them, S3: The angle between the center of gravity of the torso and face and the horizontal line Hold within 1 second , then the characteristics of falling state are met; among them, S4: When both the instantaneous fall state and the fall state characteristics are met at the same time, it is judged as a fall behavior, and the fall behavior signal is transmitted to the main control module. The main control module transmits the signal to the remote control terminal. The remote control terminal pushes a message of the fall and displays the image captured by the camera module.

[0011] Preferably, the remote control terminal is a computer or a smart phone.

[0012] The beneficial effects of the present invention are: The present invention uses the slave ESP32 to complete the collection of indoor environmental parameters such as temperature and humidity, CO2, formaldehyde, PM2.5, TVOC, etc., and transmits them to the master Raspberry Pi through the Wi-Fi communication protocol. The Raspberry Pi is used as the main control platform, and the air quality data preprocessing task is assigned to ESP32, while the fall detection and linkage tasks are handled by the Raspberry Pi to achieve parallel optimization.

[0013] The fall recognition algorithm described in the present invention first performs posture behavior recognition based on a convolutional neural network. On this basis, a fall threshold parameter that is adaptively adjusted based on the user's daily activity data is proposed. The threshold parameter can be dynamically adjusted according to the elderly's exercise habits to reduce the false alarm rate. Compared with a single sensor detecting fall behavior, the detection accuracy of the algorithm is increased from 70% to 85% (verified by simulation experiments on a real fall data set).

[0014] The present invention combines air quality detection and fall recognition technology to monitor the concentration of harmful substances in the air and fall events in real time. When a fall event or abnormal air quality is detected, the user is immediately reminded in the form of a text message, which facilitates the elderly and their families to respond quickly, provides users with more comprehensive coverage, and improves the safety and comfort of the living environment.

[0015] In summary, the present invention achieves efficient coordination of air quality monitoring and fall detection through multi-sensor fusion, intelligent linkage control and algorithm optimization based on artificial intelligence, makes up for the shortcomings of the existing technology, and provides a safer, healthier and more convenient living environment for the elderly. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a basic structural diagram of a smart home system described in the present invention.

[0017] Figure 2It is a flow chart of the fall recognition algorithm described in the present invention. DETAILED DESCRIPTION

[0018] In order to make the design scheme, purpose and result of this embodiment clearer, the embodiment of this disclosure is further analyzed below in conjunction with the drawings of the embodiment of this disclosure.

[0019] like Figure 1 As shown, the present invention provides a smart home system, including a main control module, a communication module, a camera module, a fall recognition module, a data acquisition module, a sensor module and a remote control terminal; wherein the camera module collects images and transmits them to the fall recognition module, and the fall recognition module transmits the recognition result to the main control module after performing fall recognition; the output end of the sensor module is signal-connected to the input end of the data acquisition module, and the output end of the data acquisition module is signal-connected to the input end of the main control module through the communication module; In a preferred embodiment of the present invention, the main control module adopts Raspberry Pi, the communication module adopts Wi-Fi communication, and the data acquisition module adopts slave ESP32.

[0020] In a preferred embodiment of the present invention, the sensor module integrates multiple environmental sensors such as temperature and humidity, formaldehyde, carbon dioxide, light intensity, dust particles, TVOC, etc. The temperature and humidity sensor, light intensity sensor and TVOC sensor communicate data with ESP32 through the I2C bus, the formaldehyde sensor and carbon dioxide sensor communicate data with ESP32 through the UART bus, and the dust particle sensor completes data communication through a data line.

[0021] The indoor environmental parameters collected by the sensor module are uploaded to the main control Raspberry Pi in real time through the Wi-Fi communication module. The ESP32 module is used. The module has a dual-core structure and 4MB external SPI memory, which provides sufficient processing power for the present invention. At the same time, it integrates the Wi-Fi communication module itself, which is low in cost and complexity and is suitable for indoor communication.

[0022] The main control module is connected to the remote control terminal by signal. The remote control terminal is a computer or a smart phone. It receives the collected environmental information for analysis and processing, and then displays the relevant parameters in real time on the remote control terminal (Web terminal and mobile phone APP terminal).

[0023] like Figure 2 As shown, the fall recognition module in the present invention proposes a fall recognition detection algorithm based on Google's open source posture prediction model MoveNet, collects image data through the camera module, and feeds back the fall behavior detection results to the main control module to realize automatic alarm for fall behavior. The specific process is as follows: S1: After the image data collected by the camera is inferred by the MoveNet model, the key point information of the human body is obtained, and the key points are preliminarily classified using a fully convolutional neural network into two categories: normal posture key points and possible fall posture key points; S2: When judging whether a fall has occurred, the coordinate information of the key points where the fall may occur is combined. , Represents the x-axis and y-axis coordinate values ​​of the key point of the i-th human body posture, when the trunk center of gravity linear velocity is satisfied >1.5m / s and linear speed of facial key points >1.2m / s, and the average angular velocity of key points 10 and 13 >200° / s, the instantaneous fall characteristics are met; among them, S3: The angle between the center of gravity of the torso and face and the horizontal line Hold within 1 second , then the characteristics of falling state are met; among them, S4: When both the instantaneous fall state and the fall state characteristics are met at the same time, it is judged as a fall behavior, and the fall behavior signal is transmitted to the main control module. The main control module transmits the signal to the remote control terminal. The remote control terminal pushes a message of the fall and displays the image captured by the camera module.

[0024] Specifically, the main control module obtains the recognition results of the fall recognition algorithm through the shared folder, and then uses the automation settings and code to call the API function to realize the message notification of the remote control terminal. The automation setting process includes trigger conditions and trigger actions. The trigger condition is the Raspberry Pi user. When a fall occurs, the action is triggered and the picture information in the shared folder is transmitted to realize the fall action alarm function.

[0025] The above specific implementations further illustrate the purpose and technical solutions of the present invention. It should be understood that these are only specific implementations of the present invention and do not limit its protection scope. Any modification or equivalent replacement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A smart home system, characterized in that: include: A main control module, a communication module, a camera module, a fall recognition module, a data acquisition module and a sensor module; wherein the camera module collects images and transmits them to the fall recognition module, and the fall recognition module transmits the recognition result to the main control module after performing fall recognition; the output end of the sensor module is connected to the input end of the data acquisition module by signal, and the output end of the data acquisition module is connected to the input end of the main control module by signal through the communication module; The sensor module includes a temperature and humidity sensor, a carbon dioxide sensor, a formaldehyde sensor, a light intensity sensor, a dust particle sensor and a TVOC gas sensor.

2. The smart home system according to claim 1, characterized in that: The smart home system also includes a remote control terminal, and the main control module is connected to the remote control terminal by signal.

3. The smart home system according to claim 2, characterized in that: The fall recognition module uses a fall recognition detection algorithm to perform fall recognition on the images collected by the camera module. The specific process is as follows: S1: After the image data collected by the camera module is inferred by the MoveNet model, the key point information of the human body is obtained, and the key points are preliminarily classified using a fully convolutional neural network into two categories: normal posture key points and possible fall posture key points. S2: When judging whether a fall has occurred, the coordinate information of the key points of the possible fall posture is combined. , Represents the x-axis and y-axis coordinate values ​​of the key point of the i-th human body posture, when the trunk center of gravity linear velocity is satisfied >1.5m / s and linear speed of facial key points >1.2m / s, and the average angular velocity of key points 10 and 13 >200° / s, the instantaneous fall characteristics are met; among them, S3: The angle between the center of gravity of the torso and face and the horizontal line Hold within 1 second Then the characteristics of falling state are met; among them, S4: When both the instantaneous fall state and the fall state characteristics are met at the same time, it is judged as a fall behavior, and the fall behavior signal is transmitted to the main control module. The main control module transmits the signal to the remote control terminal. The remote control terminal pushes a message of the fall and displays the image captured by the camera module.

4. The smart home system according to claim 2 or 3, characterized in that: The remote control terminal is a computer or a smart phone.

5. The smart home system according to claim 1, characterized in that: The temperature and humidity sensor, the light intensity sensor and the TVOC sensor transmit and receive data with the data acquisition module through a synchronous serial communication protocol.

6. The smart home system according to claim 1, characterized in that: The carbon dioxide sensor and the formaldehyde sensor transmit and receive data with the data acquisition module through an asynchronous serial communication protocol.

7. The smart home system according to claim 1, characterized in that: The dust particle sensor transmits and receives data with the data acquisition module via a data line.

8. The smart home system according to claim 1, characterized in that: The camera module is connected to the main control module via a CSI interface.

Images