Millimeter wave radar system

Abstract

A millimeter-wave radar system includes a millimeter-wave radar sensor and an integrated artificial intelligence (AI) model. The millimeter-wave radar sensor detects environmental data, and the integrated AI model includes an AI encoder and an AI decoder. The AI ​​encoder is coupled to the millimeter-wave radar sensor to compress the environmental data to generate an integrated representation, and the AI ​​decoder is coupled to the AI ​​encoder to analyze the integrated representation to generate multiple task results.

Description

Technical Field

[0001] This invention relates to a millimeter-wave radar system, and more particularly to a millimeter-wave radar system employing an integrated artificial intelligence model. Background Technology

[0002] Environmental sensing refers to the technology of collecting data from the surrounding environment, which can be used for various purposes, including understanding environmental conditions, providing relevant information to users, and controlling devices.

[0003] In environmental sensing, a network of environmental sensors is used to collect and provide information. In theory, almost any type of sensor can be used as an environmental sensor. Common types include temperature sensors, pressure sensors, water sensors, and object sensors.

[0004] Temperature sensors can continuously monitor indoor temperature and provide an average temperature to ensure that the indoor temperature is maintained properly. Pressure sensors, when combined with temperature sensors, can provide a comprehensive understanding of the weather conditions around the house. Water sensors can detect indoor humidity. Object sensors equipped with radio frequency identification (RFID) tags or global positioning system (GPS) trackers can be attached to key objects or users to collect information about the objects or users' conditions and behaviors.

[0005] However, traditional environmental sensors have limited practicality due to their simple output. In contrast, millimeter-wave radar sensors can provide complete radar signals for environmental sensing. These complete radar signals can be analyzed using integrated artificial intelligence models to perform various tasks. Summary of the Invention

[0006] This invention provides a millimeter-wave radar system, including a millimeter-wave radar sensor and an integrated artificial intelligence (AI) model. The millimeter-wave radar sensor detects environmental data, and the integrated AI model includes an AI encoder and an AI decoder. The AI ​​encoder is coupled to the millimeter-wave radar sensor and is used to compress the environmental data to generate an integrated representation. The AI ​​decoder is coupled to the AI ​​encoder and is used to analyze the integrated representation to generate multiple task results.

[0007] This invention provides a millimeter-wave radar system, including a millimeter-wave radar sensor, a digital signal processor (DSP), and an integrated artificial intelligence (AI) model. The millimeter-wave radar sensor detects environmental data. The DSP, coupled to the millimeter-wave radar sensor, performs digital signal preprocessing on the environmental data to generate preprocessed data. The integrated AI model includes an AI encoder and an AI decoder. The AI ​​encoder, coupled to the DSP, compresses the preprocessed data to generate an integrated representation. The AI ​​decoder, coupled to the AI ​​encoder, analyzes the integrated representation to generate multiple task results. Attached Figure Description

[0008] Figure 1 This is a block diagram of a millimeter-wave radar system in an embodiment of the present invention.

[0009] Figures 2A to 2D This is a diagram illustrating the results of a task performed by an integrated AI model.

[0010] Figure 3 This is a block diagram of the integrated AI model in an embodiment of the present invention.

[0011] Figure 4 This is a block diagram of an integrated AI model according to another embodiment of the present invention.

[0012] List of reference numerals

[0013] 10: Millimeter-wave radar system

[0014] 102: Millimeter-wave radar sensor

[0015] 104: Integrating AI Models

[0016] 106: Control Unit

[0017] 206: Result exists

[0018] 208: Tracking Results

[0019] 210: Posture Result

[0020] 212: Gesture Result

[0021] 30, 40: Integrating AI Models

[0022] 302: Complete radar signal

[0023] 304, 408, 108: AI encoders

[0024] 306, 410: Unified representation

[0025] 308, 412, 110: AI Decoder

[0026] 404: Digital Signal Preprocessing

[0027] 406: Data Fusion

[0028] 414: Digital Signal Post-processing Detailed Implementation

[0029] Millimeter-wave (mmWave) sensing is a non-contact technology that uses millimeter-wave radar sensors to measure position, motion, acceleration, and angle with millimeter-level precision. Millimeter-wave radar systems analyze the emitted and received millimeter-wave electromagnetic pulses or continuous waves, detecting targets and their motion from reflected signals. Additional components such as converters, signal processors, and other embedded technologies enhance system performance and open up new applications. Current applications of millimeter-wave radar technology include tracking human and animal movement, sensing human presence, and monitoring vital signs. These applications span multiple industries, including automotive, meteorology, healthcare, and pet health, and are often seen as an alternative to wearable technologies.

[0030] Compared to other radio frequency sensing technologies in the electromagnetic spectrum, such as infrared or ultrawideband, millimeter waves operate in the 10 GHz to 100 GHz range. Typical millimeter wave sensors use the 24 GHz, 60 GHz, and 77 GHz bands, each offering unique advantages for specific applications.

[0031] Figure 1 This is a block diagram of a millimeter-wave radar system 10 according to an embodiment of the present invention. The millimeter-wave radar system 10 may include a millimeter-wave radar sensor 102, an integrated artificial intelligence (AI) model 104, and a control unit 106. The millimeter-wave radar sensor 102, the integrated AI model 104, and the control unit 106 may be housed within an interactive device. The integrated AI model 104 may be connected to the millimeter-wave radar sensor 102 and the control unit 106. The millimeter-wave radar sensor 102 is used to sense environmental data, and the integrated AI model 104 is used to process the environmental data and generate multiple task results. The integrated AI model 104 may include an AI encoder 108 and an AI decoder 110. The AI ​​encoder 108 is trained to compress radar input data into a unified representation of multiple tasks. The AI ​​decoder 110 learns to utilize the unified representation and decodes it into multiple task results. The control unit 106 operates the interactive device based on the multiple task results. In one embodiment, environmental data includes the heart rate, respiratory rate, gestures, posture, location, and / or speed of an organism; in another embodiment, environmental data includes the location and / or speed of a non-living organism or plant.

[0032] Figures 2A to 2DThis is a schematic diagram of the task results performed by the integrated AI model 104. In one embodiment, the millimeter-wave radar sensor 102 senses environmental data to generate millimeter-wave radar signals. The environmental data may include the heart rate, respiratory rate, gestures, postures, position, and / or velocity of an organism. In another embodiment, the environmental data includes the position and / or velocity of inanimate objects or plants. The integrated AI model 104 analyzes the millimeter-wave radar signals to generate various task results, which may include presence result 206, tracking result 208, posture result 210, and gesture result 212. For presence result 206, the integrated AI model 104 determines whether a person is present; for tracking result 208, the integrated AI model 104 tracks the user's position in three-dimensional space (x, y, z); for posture result 210, the integrated AI model 104 identifies the user's posture, such as standing, sitting, or lying down; and for gesture result 212, the integrated AI model 104 identifies various dynamic gestures of the user.

[0033] As described above, the integrated AI model 104 can generate various task results, such as presence result 206, tracking result 208, pose result 210, and gesture result 212. In one embodiment, the integrated AI model 104 can perform a single task or multiple tasks simultaneously. The AI ​​encoder 110 is trained to compress radar signals into an integrated representation of multiple tasks. The AI ​​decoder 110 learns to utilize the integrated representation and decodes it into multiple task results. The control unit 106 can operate interactive devices based on these task results. Interactive devices may include smart fans, smart TVs, spatial audio systems, air conditioners, smart lighting, security monitoring systems, and electric doors. In one embodiment, the millimeter-wave radar sensor 102 transmits multiple frequency-modulated continuous wave (FMCW) signals and receives multiple reflected FMCW signals. Then, the integrated AI model 104 analyzes the multiple reflected FMCW signals to generate multiple task results.

[0034] Figure 3 This is a block diagram of the integrated AI model 30 in an embodiment of the present invention. The integrated AI model 30 includes an AI encoder 304 and an AI decoder 308. In one embodiment, the millimeter-wave radar sensor 102 can transmit FMCW signals and receive multiple reflected FMCW signals 302. The AI ​​encoder 304 compresses the multiple reflected FMCW signals 302 to generate an integrated representation 306, and the AI ​​decoder 308 analyzes the integrated representation 306 to generate multiple output results (such as...). Figure 3The AI ​​decoder performs presence detection, personnel tracking, posture detection, and gesture recognition, and converts multiple output results into multiple task results 206, 208, 210, and 212. The task results include presence result 206, tracking result 208, posture result 210, and gesture result 212. For presence detection and presence result 206, the AI ​​decoder 308 analyzes and integrates the representation 306 to determine whether someone is present. For personnel tracking and tracking result 208, the AI ​​decoder 308 analyzes and integrates the representation 306 to track the user's position (x, y, z). For posture detection and posture result 210, the AI ​​decoder 308 analyzes and integrates the representation 306 to determine the user's posture, such as sitting, standing, or lying down. For gesture recognition and gesture result 212, the AI ​​decoder 308 analyzes and integrates the representation 306 to recognize the user's dynamic gestures and react to the gestures.

[0035] In one embodiment, the AI ​​encoder 304 and the AI ​​decoder 308 include multiple learnable parameters that can be adjusted through a training process optimized by multiple task results. The AI ​​encoder 304 and the AI ​​decoder 308 use a multi-layer learnable parameter architecture, including (but not limited to) 1-D or 2-D convolution, matrix multiplication, and non-parametric operations, including shape operations (transpose, reshaping, etc.) and cell-level activation functions, including ReLU function, Sigmoid function, hyperbolic tangent function, etc.

[0036] Figure 4 This is a block diagram of an integrated AI model 40 according to another embodiment of the present invention. The integrated AI model 40 includes an AI encoder 408 and an AI decoder 412. In one embodiment, a millimeter-wave radar sensor 102 can transmit an FMCW signal and receive multiple reflected FMCW signals 302. A digital signal processor performs digital signal preprocessing 404 on the multiple reflected FMCW signals 302 to generate preprocessed data. The digital signal preprocessing 404 may include Fast Fourier Transform (FFT), Digital Fourier Transform (DFT) decimation, wavelet transform, and / or point cloud analysis. The AI ​​encoder 408 compresses the preprocessed data to generate an integrated representation 410. Then, the AI ​​decoder 412 analyzes the integrated representation 410 to generate multiple output results (such as...). Figure 4 The system performs presence detection, personnel tracking, posture detection, and gesture recognition, and performs digital signal post-processing 414 on multiple output results to generate various task results 206, 208, 210, and 212, including presence result 206, tracking result 208, posture result 210, and gesture result 212.

[0037] In one embodiment, the AI ​​encoder 408 and the AI ​​decoder 412 include multiple learnable parameters that can be adjusted through a training process optimized by multiple task results. The AI ​​encoder 408 and the AI ​​decoder 412 use a multi-layer learnable parameter architecture, including (but not limited to) 1-D or 2-D convolution, matrix multiplication, and non-parametric operations, including shape operations (transpose, reshaping, etc.) and cell-level activation functions, including ReLU function, Sigmoid function, hyperbolic tangent function, etc.

[0038] In one embodiment, the AI ​​decoder 412 generates multiple output results (such as...) Figure 4 The presence detection, personnel tracking, posture detection, and gesture recognition tasks are performed by a digital signal processor (DSP) through digital signal post-processing (414) to generate various task results 206, 208, 210, and 212. The digital signal post-processing (414) may include Kalman filtering, particle filtering, and / or clustering. The task results may include presence result 206, tracking result 208, posture result 210, and gesture result 212. In one embodiment, the task results generated by the digital signal post-processing (414) can be used in conjunction with the pre-processed data of the next round and compressed by the AI ​​encoder (408) after data fusion (406) to generate a unified representation (410). In other words, the previously post-processed task results can be used with the subsequent pre-processed data to enhance the analytical capabilities of the unified representation (410).

[0039] In one embodiment, the orientation of the interactive device is adjusted according to the task results 206, 208, 210, and 212 generated by the AI ​​decoder 412. For example, if the interactive device is a smart fan, the millimeter-wave radar system 10 senses the user's position as the tracking result 208 and adjusts the fan direction to face the user. Similarly, if the interactive device is a spatial speaker, the millimeter-wave radar system 10 senses the user's position as the tracking result 208 and adjusts the speaker to ensure that the sound surrounds the user. In addition, the millimeter-wave radar system 10 can sense the position of inanimate objects or plants and adjust the interactive device accordingly.

[0040] In one embodiment, the control unit 106 opens or closes the interactive device based on the task results 206, 208, 210, and 212 generated by the integrated AI model 40 and the digital signal post-processing 414. For example, the interactive device is a smart TV. The integrated AI model 40 senses a person's posture and / or gesture as posture result 210 and / or gesture result 212. The control unit 106 can adjust or open / close the smart TV according to the posture result 210 and / or gesture result 212. In another embodiment, the interactive device is an electric door. The integrated AI model 40 and the digital signal post-processing 414 generate posture result 210 and / or gesture result 212. The control unit 106 can open or close the electric door according to the posture result 210 and / or gesture result 212. In another embodiment, the interactive device is an air conditioner. The integrated AI model 40 and the digital signal post-processing 414 generate posture result 210 and / or gesture result 212. The control unit 106 can adjust or turn the air conditioner on / off based on the posture result 210 and / or gesture result 212. In another embodiment, the integrated AI model 40 senses the position and / or movement of non-living objects or plants as tracking result 208, and the control unit 106 can turn the interactive device on / off based on the position and / or movement of non-living objects or plants.

[0041] In summary, this invention provides a millimeter-wave radar system 10 that uses integrated AI models 30 and 40 for presence detection, personnel tracking, posture detection, and / or gesture recognition. The integrated AI models 30 and 40, using integrated representations 306 and 410, exhibit higher accuracy than existing technologies and can be applied to presence detection, personnel tracking, posture detection, and / or gesture recognition.

[0042] The above description is only a preferred embodiment of the present invention. All equivalent changes and modifications made in accordance with the claims of the present invention should be included within the scope of the present invention.

Claims

1. A millimeter-wave radar system, comprising: A millimeter-wave radar sensor is used to detect environmental data; and A unified artificial intelligence model, including: An artificial intelligence encoder, coupled to the millimeter-wave radar sensor, is used to compress the environmental data to generate a unified representation; and An AI decoder, coupled to the AI ​​encoder, is used to analyze the integrated representation to generate multiple task results.

2. The millimeter-wave radar system as claimed in claim 1, wherein the plurality of task results include an presence detection result, an object and user tracking result, a posture analysis result, and / or a gesture recognition result.

3. The millimeter-wave radar system of claim 1, wherein the environmental data includes the heart rate, respiratory rate, gestures, posture, position, and / or speed of a living object.

4. The millimeter-wave radar system of claim 1, wherein the environmental data includes the position and / or velocity of a non-living object.

5. The millimeter-wave radar system as described in claim 1, further comprising: An interactive device, coupled to the AI ​​decoder, is used to react based on the results of the multiple tasks.

6. The millimeter-wave radar system as described in claim 5, wherein the interactive device is a smart fan, a smart TV, a spatial audio system, an air conditioner, a smart lighting system, a security monitoring system, or an electric door.

7. The millimeter-wave radar system of claim 1, wherein the millimeter-wave radar sensor transmits a plurality of frequency-modulated continuous wave signals and receives a plurality of reflected frequency-modulated continuous wave signals.

8. The millimeter-wave radar system of claim 7, wherein the artificial intelligence encoder compresses the plurality of reflected frequency-modulated continuous wave signals to generate the integrated representation.

9. A millimeter-wave radar system, comprising: A millimeter-wave radar sensor is used to detect environmental data; A digital signal processor, coupled to the millimeter-wave radar sensor, is used to perform digital signal preprocessing on the environmental data to generate preprocessed data; and A unified artificial intelligence model, including: An artificial intelligence encoder, coupled to the digital signal processor, is used to compress the preprocessed data to generate a unified representation; and An AI decoder, coupled to the AI ​​encoder, is used to analyze the integrated representation to generate multiple task results.

10. The millimeter-wave radar system of claim 9, wherein the plurality of task results include an presence detection result, an object and user tracking result, a posture analysis result, and / or a gesture recognition result.

11. The millimeter-wave radar system of claim 9, wherein the environmental data includes the heart rate, respiratory rate, gestures, posture, position, and / or speed of a living object.

12. The millimeter-wave radar system of claim 9, wherein the environmental data includes the position and / or velocity of a non-living object.

13. The millimeter-wave radar system of claim 9, further comprising: An interactive device, coupled to the AI ​​decoder, is used to react based on the results of the multiple tasks.

14. The millimeter-wave radar system of claim 13, wherein the interactive device is a smart fan, a smart TV, a spatial audio system, an air conditioner, a smart lighting system, a security monitoring system, or an electric door.

15. The millimeter-wave radar system of claim 9, wherein the millimeter-wave radar sensor transmits a plurality of frequency-modulated continuous wave signals and receives a plurality of reflected frequency-modulated continuous wave signals.

16. The millimeter-wave radar system of claim 15, wherein the digital signal processor performs digital signal preprocessing on the plurality of reflected frequency-modulated continuous wave signals to generate the preprocessed data.

17. The millimeter-wave radar system of claim 9, wherein the digital signal processor performs digital signal post-processing on the plurality of task results to generate a plurality of post-processed results.

18. The millimeter-wave radar system of claim 17, wherein digital signal post-processing includes Kalman filtering, particle filtering, and / or clustering.

19. The millimeter-wave radar system of claim 17, wherein the artificial intelligence encoder compresses the preprocessed data and multiple previous postprocessing results to generate the integrated representation.

20. The millimeter-wave radar system of claim 9, wherein digital signal preprocessing includes fast Fourier transform, digital Fourier transform extraction, wavelet transform, and / or point cloud analysis.

Images