A method and system for preventing cross-body static trigger detection based on millimeter wave radar

Abstract

The application provides a millimeter wave radar-based anti-crossing human body static trigger detection method and system, comprising the following steps: S1, transmitting and receiving radar signals by a radar, and preprocessing the received radar echo signals to generate a range-Doppler matrix reflecting target distance and Doppler information; S2, extracting region of interest data within a preset distance range from the range-Doppler matrix; S3, respectively performing energy accumulation on positive Doppler units and negative Doppler units in the region of interest data to obtain positive Doppler total energy and negative Doppler total energy, and calculating an energy proportion coefficient for representing a target motion state; S4, performing stationarity statistics on the energy proportion coefficients obtained by continuous multiple frames by using a sliding window to obtain a stationarity statistical result; and S5, judging the existence state of a target by a three-level state machine based on the stationarity statistical result. The application realizes accurate distinction between "crossing through" and "static staying".

Description

Technical Field

[0001] This invention relates to the field of radar signal processing technology, and in particular to a method and system for detecting static human movement across a body using millimeter-wave radar. Background Technology

[0002] In the smart toilet scenario, human presence detection is the core element for triggering functions such as opening and closing the lid and lighting. The core requirement is that the function should only be triggered when a person is standing or sitting still in the predetermined area in front of the toilet. Normal crossing, slow passing, and back-and-forth walking should not trigger the function to avoid a poor user experience caused by accidental operation.

[0003] The existing technology has the following core defects: 1. Traditional infrared sensors cannot adapt to the requirements: Passive infrared sensors can only detect moving people, not stationary people, and are easily affected by ambient temperature and light, either frequently triggering false triggers or missing detections when the person is stationary; 2. Conventional radar has extremely poor detection and false trigger prevention capabilities: Mainstream FMCW radar human presence detection schemes only determine whether there is a person by motion / micro-motion characteristics. As long as a moving target enters the detection area, it will be triggered, and it cannot distinguish between "crossing over" and "stationary standing still". If a user walks past the toilet, it will be falsely triggered to open the lid, which is completely unsuitable for the scenario requirements; 3. Poor versatility: Existing solutions are highly dependent on fixed amplitude thresholds and installation calibration. After changing the toilet model, installation height / angle, a large number of parameters need to be recalibrated, which is not universal.

[0004] In summary, there is currently a lack of a human presence detection solution that is compatible with low-cost radar, does not rely on fixed amplitude thresholds, can accurately distinguish between "crossing motion" and "stationary stop", and has excellent anti-accidental touch effects. Summary of the Invention

[0005] To address the aforementioned problems, the purpose of this invention is to provide a method and system for detecting static human movement across a body using millimeter-wave radar.

[0006] This invention employs the following method: a millimeter-wave radar-based method for detecting stationary human movement across obstacles, comprising the following steps: S1. By transmitting and receiving radar signals through radar, and preprocessing the received radar echo signals, a range-Doppler matrix reflecting the target distance and Doppler information is generated. S2. Extract region of interest data within a preset distance range from the distance-Doppler matrix; S3. Accumulate the energy of the positive Doppler cells and negative Doppler cells in the region of interest data to obtain the total positive Doppler energy and the total negative Doppler energy, and calculate the energy ratio coefficient used to characterize the motion state of the target based on the total positive Doppler energy and the total negative Doppler energy. S4. Use a sliding window to perform stationarity statistics on the energy proportion coefficients calculated for multiple consecutive frames, and obtain the stationarity statistical result; S5. Based on the stationarity statistical result, use a three-level state machine to judge the presence state of the target. The three-level state machine at least includes an idle state, a moving state, and a stationary trigger state; only when the target completes the movement and enters the complete state transition of stationary stability, output a human presence trigger signal, otherwise determine it as a non-stationary target or a crossing target and prohibit triggering.

[0007] Further, the preprocessing in step S1 specifically includes: performing mixing, ADC sampling, range-direction fast Fourier transform (FFT), and Doppler-direction fast Fourier transform (FFT) on the echo signal in sequence to generate a range-Doppler matrix with the range unit as the horizontal axis and the Doppler unit as the vertical axis, and the matrix element is the signal amplitude corresponding to the range-velocity dimension. Doppler matrix, and the matrix element is the signal amplitude corresponding to the range velocity dimension.

[0008] Further, the calculation method of the Doppler energy proportion coefficient in step S3 is: Kratio = max(Epos, Eneg) / min(Epos, Eneg), where Epos is the total energy of positive Doppler units and Eneg is the total energy of negative Doppler units; when the target is in the crossing / moving state, the value of the Doppler energy proportion coefficient is large and fluctuates violently; when the target is in the stationary micro-motion state, the value of the Doppler energy proportion coefficient is small and stable.

[0009] Further, the sliding window stationarity statistics in step S4 includes: statistically constructing a circular sliding window with a fixed length, and circularly writing the Doppler energy proportion coefficients of consecutive frames into the cache; statistically counting the number of frames Nstable within the range of 0 < Kratio < Kth in the window, where Kth is the upper limit threshold of the stationary range.

[0010] Further, the decision logic of the three-level state machine in step S5 includes: a. In the idle state, determine whether there is a moving target entering the detection area according to the stationarity statistical result. If so, switch to the moving state; b. In the moving state, determine whether the moving target has changed to the stationary state according to the stationarity statistical result. If so, switch to the trigger state and record the current timestamp; if the target is not detected to change to the stationary state within the preset first timeout time, automatically switch back to the idle state; c. In the trigger state, output a human presence trigger signal and continuously monitor the stationarity statistical result; if the target is detected to move again, switch back to the idle state.

[0011] Furthermore, the condition for determining that the target changes to a stationary state under motion is: within the sliding window, the number of frames in which the energy ratio coefficient falls into a preset stable range is greater than or equal to a first preset threshold. The condition for determining the presence of a moving target in the idle state is: within the sliding window, the number of frames in which the energy ratio coefficient exceeds the preset stable range is greater than or equal to the second preset threshold.

[0012] The condition for the static state to transition to the idle state is: within the sliding window, the number of frames in which the energy ratio coefficient exceeds the preset stable range is less than the first threshold.

[0013] Furthermore, it also includes step S6: distance information output step: in the triggered state, the distance cell where the target is located is extracted according to the distance-Doppler matrix, and the target distance information is output.

[0014] Furthermore, the presence of a human body trigger signal is used to control the automatic opening of the smart toilet lid, the switching on and off of smart lights, or the triggering of the access control system.

[0015] Another system is provided: a millimeter-wave radar-based system for detecting stationary human movement across obstacles, including: The radar data acquisition and preprocessing module is used to control the radar's signal transmission and reception, and to perform frequency mixing, sampling, and two-dimensional FFT processing on the echo signals to generate a range-Doppler matrix. The region of interest extraction module has its input end connected to the output end of the radar data acquisition and preprocessing module, and is used to extract data within a preset range from the range-Doppler matrix; An energy ratio feature extraction module, whose input is connected to the output of the region of interest extraction module, is used to accumulate positive and negative Doppler energies and calculate the energy ratio coefficient. The stationarity statistics module, whose input is connected to the output of the energy ratio feature extraction module, is used to perform stationarity statistics on the energy ratio coefficients of multiple consecutive frames through a sliding window. The three-level state machine judgment module has its input end connected to the output end of the stationarity statistics module. It is used to execute state switching logic based on the stationarity statistics results and generate a human presence trigger signal when in the trigger state.

[0016] Furthermore, the three-level state machine judgment module specifically includes: The idle state processing unit is used to monitor the stationarity statistics when there is no target, and triggers a switch to the motion state when a moving target is detected; The motion processing unit is used to monitor the stability statistics when the target is moving, and to trigger the switch to the trigger state when the target is detected to become stationary, while implementing the timeout reset logic; The trigger state processing unit is used to output a trigger signal and trigger a switch back to the idle state when the target moves again.

[0017] The beneficial effects of this invention are as follows: This invention provides a method and system for detecting stationary human bodies crossing the road based on millimeter-wave radar. Compared with the prior art, this invention has at least the following technical effects: 1. The “three-level state machine” and the “output of the trigger signal only when the target completes the motion and enters the complete state flow to a stationary and stable state” enable precise differentiation between “crossing” and “stationary stop”, effectively filtering non-stationary targets such as normal crossing, slow passing, and back and forth walking, so as to adapt to application scenarios such as smart toilets with extremely high requirements for the false touch rate.

[0018] 2. The "Doppler energy ratio coefficient" and "sliding window stationarity statistics" can get rid of the dependence on fixed amplitude thresholds in traditional schemes. This ratio characteristic reflects the relative distribution of human motion energy and can adapt to changes in environmental clutter, radar installation height and angle, and radar gain attenuation. This greatly improves the robustness and versatility of the detection system in different environments and avoids the problems of poor versatility and frequent calibration required by existing technologies. Attached Figure Description

[0019] Figure 1 This is a flowchart of a method for detecting stationary human bodies crossing the road based on millimeter-wave radar according to the present invention.

[0020] Figure 2 This is a system principle block diagram of a millimeter-wave radar-based static trigger detection system for preventing human movement across the road, according to the present invention.

[0021] Figure 3 This is a waveform diagram of human movement detected by the millimeter-wave radar of the present invention. The blue box part is the "tangential forward movement -> tangential backward movement" in front of the radar. The radial speed of this movement is the smallest, and it involves a positive and negative speed conversion in the middle (the real-time speed is briefly 0), which can still be filtered out normally.

[0022] The diagram shows the following modules: 1. Radar data acquisition and preprocessing module; 2. Region of interest extraction module; 3. Energy ratio feature extraction module; 4. Stationarity statistics module; 5. Three-level state machine judgment module. Detailed Implementation

[0023] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0024] Please see Figures 1 to 3 A method for detecting stationary human bodies crossing the road based on millimeter-wave radar includes the following steps: S1. The radar transmits and receives radar signals, and preprocesses the received radar echo signals to generate a range-Doppler matrix that reflects the target distance and Doppler information; the radar can be an FMCW radar of different frequency bands such as 24GHz or 60GHz, and the specific algorithm for preprocessing can be adjusted according to the radar parameters.

[0025] S2. Extract region of interest data within a preset distance range from the distance-Doppler matrix; by limiting the detection range, clutter interference from irrelevant areas is effectively eliminated, reducing the computational load of subsequent processing.

[0026] In practical applications, the preset distance range can be flexibly set according to the application scenario. For example, it can be set to 0.2m-1.5m in the smart toilet scenario and 0.5m-3m in the access control scenario.

[0027] S3. The positive and negative Doppler cells in the region of interest data are accumulated to obtain the total positive and negative Doppler energy. An energy ratio coefficient, used to characterize the target's motion state, is then calculated based on this total energy. By calculating the ratio of positive to negative Doppler energy, a relative value feature related to the target's motion direction and intensity is constructed. This feature is insensitive to absolute signal strength, thus possessing strong environmental adaptability. The energy ratio coefficient can be calculated as a ratio, difference, or other mathematical combination.

[0028] S4. The energy ratio coefficients calculated from multiple consecutive frames are statistically analyzed using a sliding window to obtain the stationarity statistical results. Through multi-frame statistics, the noise and random fluctuations that may exist in single-frame data are smoothed out, making the judgment of the target motion state more stable and reliable. The length of the sliding window (e.g., 8 frames, 16 frames) and the statistical indicators (e.g., the number of frames falling into the interval, variance, etc.) can be configured according to the required response speed and stability requirements.

[0029] S5. Based on the stationarity statistics, the existence state of the target is determined using a three-level state machine, which includes at least an idle state, a moving state, and a stationary trigger state. A human presence trigger signal is output only when the target completes its movement and enters a stable, stationary state; otherwise, it is determined to be a non-stationary target or a crossing target, and triggering is prohibited. By completing the target state transition from "idle" to "moving" and then to "stationary," the possibility of directly identifying a moving target as a valid target is completely eliminated, perfectly solving the problem of false triggering in moving scenarios such as crossing.

[0030] Further, the preprocessing described in step S1 specifically includes: performing mixing, ADC sampling, range-direction fast Fourier transform (FFT), and Doppler-direction fast Fourier transform (FFT) on the echo signal in sequence to generate a range-Doppler matrix with range cells as the horizontal axis and Doppler cells as the vertical axis, and the matrix elements are the signal amplitudes corresponding to the range-Doppler dimensions. It clarifies the standard low-computation implementation path from the original echo to the range-Doppler matrix, without the need for complex time-frequency analysis or super-resolution algorithms, ensuring that the solution can run in real time on a low-cost MCU, which is beneficial for implementation on low-cost hardware such as 1T1R single-antenna radars. Doppler matrix, and the matrix elements are the signal amplitudes corresponding to the range velocity dimension. It clarifies the standard low-computation implementation path from the original echo to the range-Doppler matrix, without the need for complex time-frequency analysis or super-resolution algorithms, ensuring that the solution can run in real time on a low-cost MCU, which is beneficial for implementation on low-cost hardware such as 1T1R single-antenna radars.

[0031] Further, the calculation method of the Doppler energy ratio coefficient in step S3 is: Kratio = max(Epos, Eneg) / min(Epos, Eneg), where Epos is the total energy of positive Doppler cells and Eneg is the total energy of negative Doppler cells; when the target is in a crossing or moving state, the value of the Doppler energy ratio coefficient is large and fluctuates violently; when the target is in a stationary micro-motion state, the value of the Doppler energy ratio coefficient is small and stable. Using the ratio of positive and negative energies (to ensure statistical simplicity, the high value is divided by the low value) can clearly distinguish "targets with a clear direction of motion" and "micro-motion targets tending to be stationary", with clear physical meaning and simple and efficient calculation.

[0032] Further, the sliding window stability statistics in step S4 include: statistically constructing a circular sliding window with a fixed length, and circularly writing the Doppler energy ratio coefficients of consecutive frames into the cache; statistically counting the number of frames Nstable within the range of 0 < Kratio < Kth in the window, where Kth is the upper limit threshold of the stable range. The multi-frame statistics smooth the noise and accidental fluctuations that may exist in the single-frame features, making the judgment of the motion / static state more stable and reliable; at the same time, the fixed length of the circular window ensures the certainty of the calculation amount and memory occupancy, facilitating implementation on an embedded platform. Kth, as a configurable upper limit threshold, defines the decision boundary of the "stationary" state. The value of Kth can be calibrated through experiments, for example, set to 1.4. In addition, variance or standard deviation can also be used as a measure of stability.

[0033] Further, the three-level state machine decision logic in step S5 includes: a. In the idle state, determine whether there is a moving target entering the detection area according to the stability statistics result. If so, switch to the moving state; b. In motion state, determine whether the moving target has changed to a stationary state based on the stability statistics. If so, switch to the triggered state and record the current timestamp. If the target is not detected to have changed to a stationary state within the preset first timeout period, automatically switch back to the idle state. c. In the triggered state, a human presence trigger signal is output, and the stability statistics are continuously monitored; if the target is detected to move again, the system switches back to the idle state. The first timeout can be set according to the typical traversal time of the scenario, for example, 4000ms in a smart toilet scenario. This forms a complete "movement entry → stationary stability" closed-loop decision mechanism. It ensures that even if a target that briefly passes through (traverses) creates a partial stationary illusion, it will be automatically cleared because the stationary transition has not been completed within the timeout period, thus completely eliminating false triggering. At the same time, continuous monitoring in the triggered state also ensures that triggering stops in time after the human body leaves.

[0034] Furthermore, the condition for determining that the target changes to a stationary state under motion is: within the sliding window, the number of frames in which the energy ratio coefficient falls into a preset stable range is greater than or equal to a first preset threshold. The condition for determining the presence of a moving target in the idle state is: within the sliding window, the number of frames in which the energy ratio coefficient exceeds the preset stable range is greater than or equal to the second preset threshold.

[0035] The condition for transitioning from a static state to an idle state is: within the sliding window, the number of frames in which the energy ratio coefficient exceeds a preset stable range is less than a first threshold. This transforms the state switching logic into explicit numerical criteria, ensuring the decision result is deterministic and repeatable, thus avoiding jitter caused by single-frame jumps. The first threshold is typically high (e.g., 7 frames) to ensure the reliability of static confirmation; the second threshold is relatively low (e.g., 6 frames) to ensure the sensitivity of the response. By independently setting the first (higher) and second (lower) thresholds, a good balance can be achieved between "reliability of static confirmation" and "sensitivity of motion response." In practical applications, these thresholds can be optimized and adjusted according to the frame period and the desired response time.

[0036] Furthermore, it also includes step S6: distance information output step: In the triggered state, the distance cell where the target is located is extracted according to the distance-Doppler matrix, and the target distance information is output. While determining that the human body is stationary, the distance information of the target is provided, offering richer input for subsequent intelligent control (such as adjusting the toilet seat temperature or automatic lid opening angle based on distance). In practical applications, a stable distance value can be calculated from multiple candidate distance cells using methods such as energy-weighted averaging.

[0037] Furthermore, the presence of a human body trigger signal is used to control the automatic opening of the smart toilet lid, the switching on and off of smart lights, or the triggering of an access control system. This trigger signal can be any form, such as a high or low level signal, a serial port command, or a wireless signal, to adapt to the control interfaces of different downstream devices.

[0038] Please see Figure 3 Additionally, a millimeter-wave radar-based system for detecting stationary human movement across obstacles is provided, comprising: Radar data acquisition and preprocessing module 1 is used to control the radar to transmit and receive signals, and to perform frequency mixing, sampling and two-dimensional FFT processing on the echo signals to generate a range-Doppler matrix; The region of interest extraction module 2 has its input end connected to the output end of the radar data acquisition and preprocessing module 1, and is used to extract data within a preset range from the range-Doppler matrix; The energy ratio feature extraction module 3 has its input end connected to the output end of the region of interest extraction module 2, and is used to accumulate positive and negative Doppler energies and calculate the energy ratio coefficient. The stationarity statistics module 4, whose input is connected to the output of the energy ratio feature extraction module 3, is used to perform stationarity statistics on the energy ratio coefficients of multiple consecutive frames through a sliding window. The third-level state machine judgment module 5 has its input end connected to the output end of the stationarity statistics module 4. It is used to execute state switching logic based on the stationarity statistics results and generate a human presence trigger signal when in the trigger state.

[0039] Furthermore, the three-level state machine judgment module 5 specifically includes: The idle state processing unit is used to monitor the stationarity statistics when there is no target, and triggers a switch to the motion state when a moving target is detected; The motion processing unit is used to monitor the stability statistics when the target is moving, and to trigger the switch to the trigger state when the target is detected to become stationary, while implementing the timeout reset logic; The trigger state processing unit is used to output a trigger signal and trigger a switch back to the idle state when the target moves again.

[0040] The present invention has the following working principle: First, the system controls the radar front-end to transmit FMCW wave signals and receive echoes via radar data acquisition and preprocessing module 1. The echo signals are mixed to obtain intermediate frequency signals, which are then sampled by an ADC and processed by a two-dimensional FFT (range-dimensional FFT + Doppler-dimensional FFT) to finally generate a two-dimensional range-Doppler matrix (RDM). Each element of this matrix represents the signal energy at a specific distance and velocity.

[0041] Subsequently, the region of interest extraction module 2 crops the data of the corresponding distance unit from the complete RDM according to a preset detection distance range (such as 0.5m to 2.5m in front of the smart toilet), forming the ROI-RDM. This operation effectively excludes the interference of stationary objects such as walls and furniture outside the detection range.

[0042] Next, the energy ratio feature extraction module 3 processes the ROI-RDM. It accumulates the energy values of all positive Doppler units (representing motion away from the radar) and negative Doppler units (representing motion towards the radar) respectively, obtaining Epos and Eneg. Then, the energy ratio coefficient Kratio of the current frame is calculated according to the formula Kratio = max(Epos, Eneg) / min(Epos, Eneg). When a person walks horizontally in front of the radar, a strong single-direction radial velocity will be generated, resulting in a huge difference between Epos and Eneg, and the value of Kratio will be very large; while when a person stands or sits still, there is only slight movement such as chest breathing, and its movement direction changes rapidly and slightly between positive and negative, making Epos and Eneg tend to be equal, and the value of Kratio will be very small (close to 1).

[0043] Then, the stationarity statistics module 4 sends the Kratio value calculated for each frame into a circular sliding window with a fixed length. Each time the window is updated, the number of frames Nstable in which the Kratio value within the window falls into a preset stationary interval (such as 0 < Kratio < 1.4) is counted. This statistic Nstable effectively reflects the motion stationarity of the target in the recent period of time.

[0044] Finally, the three-level state machine judgment module 5 performs the core logical judgment based on Nstable. The state machine is initially in the "idle state". When Nstable is continuously low for multiple frames (that is, the Kratio values of most frames within the window are large), it indicates that a moving target has entered, and the state machine switches to the "motion state". In the "motion state", the system continuously monitors Nstable. When Nstable reaches a higher threshold (such as 7 frames), it indicates that the target has changed from motion to static, and the state machine immediately switches to the "static trigger state". If the transition to static is not detected within the preset timeout period (such as 4000ms) in the "motion state", it is determined as a cross-through, and the state machine automatically resets to the "idle state". Once entering the "static trigger state", the system immediately outputs a human presence trigger signal (such as an opening lid command). In this state, the system still continuously monitors Nstable. Once it is found that Nstable is lower than the threshold, it indicates that the target is moving again or has left, and the state machine then switches back to the "idle state" and stops the trigger signal.

[0045] Through the above-described workflow of "feature extraction - stationarity statistics - state machine closed-loop decision", this invention achieves accurate, reliable, and low-cost differentiation and detection of "crossing motion" and "stationary stop".

[0046] The above description is only a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions that fall within the scope of the present invention are within the scope of protection of the present invention.

[0047] It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of this invention should also be considered within the scope of protection of this invention.

Claims

1. A method for detecting stationary human bodies crossing the road based on millimeter-wave radar, characterized in that, It includes the following steps: S1. Transmit and receive radar signals through a radar, and preprocess the received radar echo signals to generate a range-Doppler matrix reflecting target range and Doppler information; S2. Extract the data of the region of interest within a preset range from the range-Doppler matrix; S3. Accumulate the energies of the positive Doppler cells and negative Doppler cells in the data of the region of interest respectively to obtain the total positive Doppler energy and the total negative Doppler energy, and calculate an energy ratio coefficient for characterizing the target motion state according to the total positive Doppler energy and the total negative Doppler energy; S4. Use a sliding window to perform stationarity statistics on the energy ratio coefficients calculated for multiple consecutive frames to obtain a stationarity statistical result; S5. Based on the stationarity statistical result, judge the presence state of the target through a three-level state machine, and the three-level state machine at least includes an idle state, a motion state and a stationary trigger state; only when the target completes the motion and enters the complete state transition to stationary stability, output a human presence trigger signal, otherwise judge it as a non-stationary target or a crossing target and prohibit triggering.

2. The method for detecting stationary human bodies crossing the road based on millimeter-wave radar according to claim 1, characterized in that, The preprocessing described in step S1 specifically includes: sequentially performing frequency mixing, ADC sampling, range-to-fast Fourier transform (FFT), and Doppler-to-fast Fourier transform (FFT) on the echo signal to generate a range signal with the range unit as the horizontal axis and the Doppler unit as the vertical axis. Doppler matrix, where the matrix elements are the corresponding distances. Signal amplitude in the velocity dimension.

3. The method for detecting stationary human bodies crossing the road based on millimeter-wave radar according to claim 1, characterized in that, The calculation method of the Doppler energy ratio coefficient in step S3 is: Kratio = max(Epos, Eneg) / min(Epos, Eneg), where Epos is the total energy of the positive Doppler cells and Eneg is the total energy of the negative Doppler cells; when the target is in a crossing / motion state, the value of the Doppler energy ratio coefficient is large and fluctuates violently; when the target is in a stationary micro-motion state, the value of the Doppler energy ratio coefficient is small and stable.

4. The method for detecting stationary human bodies crossing the road based on millimeter-wave radar according to claim 3, characterized in that, The sliding window stationarity statistics in step S4 include: statistically constructing a circular sliding window with a fixed length, and circularly writing the Doppler energy ratio coefficients of consecutive frames into a cache; statistically counting the number of frames Nstable within the range of 0 < Kratio < Kth in the sliding window, where Kth is the upper limit threshold of the stationary interval.

5. The method for detecting stationary human bodies crossing the road based on millimeter-wave radar according to claim 1, characterized in that, The decision logic of the three-level state machine in step S5 includes: a. In the idle state, judge whether there is a moving target entering the detection area according to the stationarity statistical result, and if so, switch to the motion state; b. In the motion state, judge whether the moving target has changed to a stationary state according to the stationarity statistical result, and if so, switch to the trigger state and record the current timestamp; if the target is not detected to change to stationary within a preset first timeout time, automatically switch back to the idle state; c. In the trigger state, output a human presence trigger signal and continuously monitor the stationarity statistical result; if it is detected that the target moves again, switch back to the idle state.

6. The method for detecting stationary human bodies crossing the road based on millimeter-wave radar according to claim 5, characterized in that, The condition for judging that the target changes to a stationary state in the motion state is: within the sliding window, the number of frames in which the energy ratio coefficient falls within a preset stationary interval ≥ the first preset threshold; The condition for judging that there is a moving target entering in the idle state is: within the sliding window, the number of frames in which the energy ratio coefficient exceeds the preset stationary interval ≥ the second preset threshold. The condition for converting the stationary state to the idle state is: within the sliding window, the number of frames in which the energy ratio coefficient exceeds the preset stationary interval < the first threshold.

7. The method for detecting stationary human bodies crossing the road based on millimeter-wave radar according to claim 6, characterized in that, It also includes step S6: distance information output step: in the triggered state, the distance cell where the target is located is extracted according to the distance-Doppler matrix, and the target distance information is output.

8. A method for detecting stationary human bodies crossing the road based on millimeter-wave radar according to any one of claims 1-7, characterized in that, The presence of a human body trigger signal is used to control the automatic opening of the smart toilet lid, the switching on and off of smart lights, or the triggering of the access control system.

9. A millimeter-wave radar-based system for detecting stationary human bodies crossing obstacles, used to implement the method described in any one of claims 1 to 8, characterized in that, include: The radar data acquisition and preprocessing module is used to control the radar's signal transmission and reception, and to perform frequency mixing, sampling, and two-dimensional FFT processing on the echo signals to generate a range-Doppler matrix. The region of interest extraction module has its input end connected to the output end of the radar data acquisition and preprocessing module, and is used to extract data within a preset range from the range-Doppler matrix; An energy ratio feature extraction module, whose input is connected to the output of the region of interest extraction module, is used to accumulate positive and negative Doppler energies and calculate the energy ratio coefficient. The stationarity statistics module, whose input is connected to the output of the energy ratio feature extraction module, is used to perform stationarity statistics on the energy ratio coefficients of multiple consecutive frames through a sliding window. The three-level state machine judgment module has its input end connected to the output end of the stationarity statistics module. It is used to execute state switching logic based on the stationarity statistics results and generate a human presence trigger signal when in the trigger state.

10. The millimeter-wave radar-based static trigger detection system for preventing human movement across obstacles, as described in claim 9, is characterized in that... The three-level state machine judgment module specifically includes: The idle state processing unit is used to monitor the stationarity statistics when there is no target, and triggers a switch to the motion state when a moving target is detected; The motion processing unit is used to monitor the stability statistics when the target is moving, and to trigger the switch to the triggered state when the target is detected to become stationary, while implementing the timeout reset logic; The trigger state processing unit is used to output a trigger signal and trigger a switch back to the idle state when the target moves again.

Images