Intelligent toilet human body detection system based on millimeter wave radar

Abstract

The present application relates to the technical field of millimeter wave radar system, and especially relates to a human body detection system of intelligent closestool based on millimeter wave radar.The present application provides a human body detection system of intelligent closestool based on millimeter wave radar, which comprises a millimeter wave radio frequency transceiver unit, a micro control unit, a transmitting antenna, a receiving antenna and an intelligent closestool, wherein the millimeter wave radio frequency transceiver unit is connected with the micro control unit, the transmitting antenna and the receiving antenna respectively.The human body detection algorithm used in the system is mainly based on the clustering formed by the sampling points of the body of the tester by the millimeter wave radar, and the human body motion track is predicted in combination with the relevant knowledge of physics, so as to judge whether the person is close to, passes by or is away from the intelligent closestool, and to determine whether the closestool is opened or not.The recognition accuracy of the system for the above-mentioned basic state of the human body can reach more than 95%, and the related modules of the system are convenient to integrate and install, have small power consumption and good privacy protection.

Description

Technical Field

[0001] This invention relates to the field of millimeter-wave radar system technology, and in particular to a smart toilet human body detection system based on millimeter-wave radar. Background Technology

[0002] The rapid growth in automation has facilitated more efficient and convenient human-computer interaction. Radar sensing of human movement further enhances this interactive experience. Most studies on human detection using millimeter-wave frequency-modulated continuous-wave radar employ methods that estimate distance and velocity from raw data, such as time-spectrum maps, micro-Doppler spectra, and range-Doppler imaging. Furthermore, angle estimation using multiple antennas also contains rich information about human movement, particularly distinguishing between lateral and longitudinal movements.

[0003] In the field of human-computer interaction, sensor-based dynamic and static target recognition is a crucial component. Moving away from physical control interfaces (such as button and touchscreen gesture recognition) is a hot research area. Numerous studies have explored gesture recognition based on various sensors, including image-based solutions using OpenCV and wearable device-based trajectory tracking solutions. Wireless human detection, as a non-contact technology, allows wireless signals to penetrate materials such as plastic, walls, and clothing. It can operate at night and its performance is unaffected by visible light. Compared to visible light-based solutions, wireless solutions offer significant advantages in privacy protection. Millimeter-wave radar is a promising wireless technology. It possesses low-power system-on-chips and sensors, enabling simultaneous signal feature extraction and classification. Currently, research on millimeter-wave radar for human position and velocity detection and determination in smart home applications (such as smart toilets) is limited. Therefore, designing a millimeter-wave radar human detection system with high recognition rates, strong anti-interference capabilities, scalability, and comprehensive functionality is crucial for the commercial application of millimeter-wave radar. Summary of the Invention

[0004] The purpose of this invention is to address the limited application of existing millimeter-wave radar human body detection in the smart home field and the lack of more mature and comprehensive systems for more accurate human body detection. Taking the practical application of smart toilets as an example, this invention proposes a smart toilet human body detection system based on millimeter-wave radar to identify various states such as frontal entry, side entry, passing by, and exit.

[0005] A human body detection system for a smart toilet based on millimeter-wave radar includes: a millimeter-wave radar processing module and a smart toilet;

[0006] The millimeter-wave radar processing module is installed behind the smart toilet seat. The module includes a millimeter-wave radio frequency transceiver unit, a microcontroller unit, a transmitting antenna, and a receiving antenna. The millimeter-wave radio frequency transceiver unit is connected to the microcontroller unit, the transmitting antenna, and the receiving antenna, respectively.

[0007] The millimeter-wave radio frequency transceiver unit generates a corresponding millimeter-wave radar signal based on the radar waveform parameters, receives the echo signal reflected back by the test personnel from the millimeter-wave radar signal, and then transmits the echo signal to the microcontroller unit.

[0008] The microcontroller unit is a single-chip microcomputer that can be programmed and run human detection programs. It is used to process echo signals to obtain human detection judgment results and transmit them to the smart toilet.

[0009] The transmitting antenna is used to transmit millimeter-wave radar signals generated by the millimeter-wave radio frequency transceiver unit.

[0010] The receiving antenna is used to receive millimeter-wave radar signals reflected back in space and transmit them to the millimeter-wave radio frequency transceiver unit.

[0011] The smart toilet controls the opening or closing of the lid based on the human body detection results returned by the millimeter-wave radar processing module and related communication protocols.

[0012] Preferably, the human body detection algorithm of the system is completed during the processing of echo signals by the microcontroller unit. The operation of the human body detection system and the smart toilet includes the following steps:

[0013] S1. The microcontroller unit obtains data information from multiple points in each frame from the tester based on the echo signal;

[0014] S2. Run clustering algorithm A to obtain the location and energy information of the test personnel;

[0015] S3. Run the tracking algorithm to obtain the predicted position of the tester and the movement speed in the X and Y axes;

[0016] S4. Combine the data information calculated in steps S2 and S3 to run the human detection algorithm and obtain the human detection judgment result.

[0017] S5: The smart toilet decides whether to open the lid based on the processing results and communication protocol returned by the microcontroller unit.

[0018] Preferably, the data information in step S1 includes:

[0019] The frame number of the current frame, frameNow;

[0020] The distance r from the point cloud acquired in the current frame to the radar center;

[0021] The three-dimensional Cartesian coordinates (x, y, z) of the point cloud acquired in the current frame;

[0022] The velocity of the point cloud captured in the current frame;

[0023] The point cloud power pow acquired in the current frame;

[0024] The number of valid point clouds acquired in the current frame after processing by millimeter-wave radar is TargetNum.

[0025] Preferably, the clustering algorithm A in step S2 specifically includes the following steps:

[0026] S2.1 Sort the point cloud data of each frame obtained by the millimeter-wave radar and the underlying signal processing in ascending order according to the distance r to the radar center. The sorting algorithm adopts the bubble sort method, which only sorts the sequence number of the point cloud data without changing the point cloud space.

[0027] S2.2, Traverse all point cloud data in each frame after sorting, starting from ascending order of distance to the radar center r:

[0028] S2.2.1 If the current point is not assigned, a new class is assigned to the current point, and the current point is marked as assigned. Then, the points in the first preset radius neighborhood are grouped into the class where the current point is located, and the weighted average of the three-dimensional spatial coordinates and the weighted energy value of all point cloud data in the class are updated. At the same time, all points in the class are marked as assigned.

[0029] If the current point has already been assigned, skip the current loop and continue to the next point;

[0030] S2.2.2 Repeat process S2.2.1 until all points in the current frame have been allocated;

[0031] S2.3 After traversing all point cloud data in the current frame, if the number of classes obtained in step S2.2 is greater than 1, classes with fewer than 2 points are first filtered out, and then the class with the larger weighted average energy is selected. The weighted average coordinates and weighted energy value of the class are used as the clustering result returned by the algorithm.

[0032] Preferably, the tracking algorithm described in step S3 is as follows:

[0033] S3.1. Run clustering algorithm B on each frame of point cloud data obtained from millimeter-wave radar acquisition and low-level signal processing. The algorithm steps are as follows:

[0034] S3.1.1. Find the maximum power point for each frame of point cloud data obtained from the millimeter-wave radar by traversing and searching, and record its power value MaxPow and its index MaxPowIdx.

[0035] S3.1.2 Select point cloud data with power greater than 0.4*MaxPow within the second preset radius neighborhood of the maximum power point. Then, classify these points together with the maximum power point into one class and update the weighted average coordinates and weighted energy values ​​of all point cloud data in this class as the clustering result returned by the algorithm.

[0036] S3.2. Based on the point cloud clustering results obtained from clustering algorithm B for each frame, the trajectory of the test personnel in the plane parallel to the ground is tracked and predicted. The corresponding iterative calculation process is as follows:

[0037] S3.2.1 Initialize the tester's tracking position (t_x, t_y) and tracking speed (t_vx, t_vy):

[0038] t_x=0m, t_y=0m, t_vx=0m / s, t_vy=0m / s

[0039] S3.2.2. Based on the tester's tracking position (t_x, t_y) and tracking speed (t_vx, t_vy) of the previous frame, the frame interval Frame_T, and the clustering position (xCenter, yCenter) of the current frame obtained by clustering algorithm B, calculate the tester's predicted position (x_predict, y_predict) and prediction bias (delta_x, delta_y) for the current frame:

[0040] x_predict=t_x+t_vx*Frame_T;

[0041] y_predict=t_y+t_vx*Frame_T;

[0042] delta_x=xCenter-x_predict;

[0043] delta_y=yCenter-y_predict;

[0044] S3.2.3 Update the tracking position (t_x, t_y) and tracking speed (t_vx, t_vy) of the current frame:

[0045] t_x = x_predict + λ * delta_x;

[0046] t_y = y_predict + λ * delta_y;

[0047] t_vx=t_vx+ω / Frame_T*delta_x;

[0048] t_vy=t_vx+ω / Frame_T*delta_y;

[0049] Where λ and ω take values ​​from 0 to 1 and can be adjusted according to actual needs, the clustering algorithm B is better than the clustering algorithm A at close ranges within 1m and is used to assist the speed estimation of the tracking algorithm.

[0050] Preferably, the human body detection algorithm in step S4 divides the human body state as follows:

[0051] NOTHING: The millimeter-wave radar did not detect any target;

[0052] NOTARGET: The millimeter-wave radar received a signal, but no target was detected approaching the smart toilet;

[0053] BODY_IN: Millimeter-wave radar detected a target approaching the smart toilet, but the target has not yet entered the smart toilet's sensing area and the conditions for opening the lid are not met.

[0054] BODY_PASS: Millimeter-wave radar detected a target approaching the smart toilet, but the target left the sensing area before the conditions for opening the lid were met after entering the smart toilet's sensing area.

[0055] BODY_STAY: The target enters the smart toilet's sensing area and meets the conditions for opening the lid, and stays in the current sensing area;

[0056] BODY_AWAY: The target leaves the smart toilet's sensor area and meets the condition of closing the lid;

[0057] The relationships between the various human body states are as follows:

[0058] NOTHING: Maintain or transition to one of the three states: NOTARGET, BODY_IN, or BODY_STAY;

[0059] NOTARGET: Maintain or transition to one of the three states: NOTHING, BODY_IN, or BODY_STAY.

[0060] BODY_IN: Maintain or transition to one of the four states: NOTHING, NOTARGET, BODY_PASS, BODY_STAY;

[0061] BODY_PASS: Cannot be maintained, will be forcibly switched to BODY_AWAY state after one frame;

[0062] BODY_STAY: Maintain or transition to one of the three states: BODY_AWAY, NOTARGET, or NOTHING;

[0063] BODY_AWAY: Maintain or transition to one of the three states: NOTHING, NOTARGET, or BODY_STAY.

[0064] The NOTHING, NOTARGET, and BODY_IN states can all enter the BODY_STAY state without going through other states.

[0065] Preferably, the conditions for entering the BODY_STAY state fall into two categories:

[0066] Scenario 1: When a person enters the smart toilet's sensing area from the front, the lid opens quickly. Scenario 2: When a person enters the smart toilet's sensing area from the side, the lid opens slowly when the person's movement speed in the Y-axis direction is less than the speed threshold and the number of frames spent in the sensing area reaches the frame number threshold.

[0067] The criteria for determining whether someone enters the smart toilet's sensor area from the front or the side are as follows:

[0068] a. The sensing area is a rectangular area of ​​a specified size directly in front of the smart toilet. The width of this rectangular area is the same as that of the smart toilet, and the length ranges from 0.4m to 0.5m.

[0069] b. When testers enter the sensing area from the left or right sides of the sensing area or from behind the smart toilet, it is considered as entering from the side; otherwise, it is considered as entering from the front.

[0070] Preferably, step S5 specifically includes:

[0071] S5.1 The smart toilet motherboard receives the processing result returned by the microcontroller. When the processing result is BODY_STAY, it controls the smart toilet to open the lid; otherwise, it does not open the lid.

[0072] S5.2 If the current person is in the BODY_STAY state and the toilet seat is open, the smart toilet will close the lid when the person leaves the sensing area and is in one of the three states: BODY_AWAY, NOTARGET, or NOTHING; otherwise, the lid will remain open.

[0073] The present invention has the following advantages:

[0074] 1. The present invention provides an intelligent toilet human body detection system based on millimeter-wave radar, which can identify various states such as front entry, side entry, passing by and exiting.

[0075] 2. The present invention has an accuracy rate of over 95% in recognizing various basic states of the human body, and the related devices of the system are easy to integrate and install, with low overall power consumption.

[0076] 3. As a wireless human detection system based on millimeter-wave radar, this invention can work at night and its performance is not affected by visible light. Compared with solutions based on visible light, the technical solution of this invention has a significant advantage in terms of privacy protection. Attached Figure Description

[0077] Figure 1 Basic block diagram of a smart toilet for installing millimeter-wave radar;

[0078] Figure 2 This is a basic block diagram of a human detection algorithm;

[0079] Figure 3 A diagram showing the correspondence between human body states, numerical codes, and their meanings. Detailed Implementation

[0080] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the various embodiments of the present invention will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details have been presented in the various embodiments of the present invention to enable the reader to better understand the present invention. However, the technical solutions claimed in the present invention can be implemented even without these technical details and various changes and modifications based on the following embodiments.

[0081] A human body detection system for a smart toilet based on millimeter-wave radar includes: a millimeter-wave radar processing module and a smart toilet;

[0082] The millimeter-wave radar processing module is installed behind the smart toilet seat. The module includes a millimeter-wave radio frequency transceiver unit, a microcontroller unit, a transmitting antenna, and a receiving antenna. The millimeter-wave radio frequency transceiver unit is connected to the microcontroller unit, the transmitting antenna, and the receiving antenna, respectively.

[0083] The millimeter-wave radio frequency transceiver unit generates the corresponding millimeter-wave radar signal according to the radar waveform parameters, receives the echo signal reflected back by the test personnel from the millimeter-wave radar signal, and then transmits the echo signal to the microcontroller unit.

[0084] The microcontroller unit is a single-chip microcomputer that can be programmed and run human detection programs. It is used to process echo signals to obtain human detection judgment results and transmit them to the smart toilet.

[0085] The transmitting antenna is used to transmit millimeter-wave radar signals generated by the millimeter-wave radio frequency transceiver unit;

[0086] The receiving antenna is used to receive millimeter-wave radar signals reflected back from space and transmit them to the millimeter-wave radio frequency transceiver unit.

[0087] The smart toilet opens or closes its lid according to the human detection results returned by the millimeter-wave radar processing module and in accordance with the relevant software protocol.

[0088] The installation location and angle of millimeter-wave radar in smart toilets are as follows: Figure 1 As shown. The radar panel is in the YZ plane. The X-axis is perpendicular to the radar, and the front of the panel is the positive direction of the X-axis. The Y-axis is parallel to the ground; when facing the radar, the left hand is the negative direction of the Y-axis, and the right hand is the positive direction of the Y-axis. The Z-axis is perpendicular to the ground; upward is the positive direction of the Z-axis, and downward is the negative direction of the Z-axis. β is the angle between the target point and the horizontal plane, and α is the angle between the short side of the radar panel and the horizontal plane. Note that the RF chip of the radar module must be on top. Since the radar's beam angle range is approximately (-60°, 60°), the effective target points above the horizontal plane satisfy β < 60 - α, meaning that within this range, the radar collects the most target points, resulting in the best effect.

[0089] This invention provides an intelligent toilet human body detection system based on millimeter-wave radar, which can identify human body states such as... Figure 3 As shown.

[0090] Combination Figure 2 The human detection algorithm for the smart toilet described in this invention specifically includes the following steps:

[0091] Step 1: Point cloud processing and clustering algorithm A;

[0092] (1) Sort the point cloud data of each frame obtained by the millimeter-wave radar and the underlying signal processing according to the distance r to the radar center from small to large. The specific sorting algorithm is the bubble sort algorithm. Only the sequence number of the point cloud data needs to be sorted, and the point cloud space does not need to be changed.

[0093] (2) Traverse all point cloud data in each frame after sorting, starting from ascending order of distance to radar center r from smallest to largest:

[0094] (a) If the current point is not assigned, a new class is assigned to the current point, and the current point is marked as assigned. Then, the points in the neighborhood of the current point within the first preset radius (ranging from 0.1 to 0.2 m, and in this embodiment the first preset radius is 0.2 m) are grouped into the class of the current point, and the weighted average of the three-dimensional spatial coordinates and the weighted energy value of all point cloud data in the class are updated, and all points in the class are marked as assigned. If the current point is already assigned, the current loop is skipped and the next point is traversed.

[0095] (b) Repeat process (a) until all points in the current frame have been assigned;

[0096] (3) After traversing all point cloud data in the current frame, if the number of classes obtained in step (2) is greater than 1, then first filter out classes with fewer than 2 points, and then select the class with the larger weighted average energy, and use the weighted average of the coordinates and the weighted energy value of the class as the clustering result returned by the algorithm.

[0097] Step 2, tracking algorithm;

[0098] (1) Run clustering algorithm B on each frame of point cloud data acquired by millimeter-wave radar and processed by low-level signal processing. The specific steps include the following:

[0099] (a) The maximum power point is found by traversing and searching each frame of point cloud data acquired by millimeter-wave radar, and its power value MaxPow and index MaxPowIdx are recorded.

[0100] (b) Select point cloud data with power greater than 0.4*MaxPow in the neighborhood of the maximum power point with a second preset radius (range 0.3-0.5m, the second preset radius is 0.5m in this embodiment), then classify these points together with the maximum power point into one class, and update the weighted average coordinates and weighted energy values ​​of all point cloud data in the class as the clustering result returned by the algorithm;

[0101] (2) Based on the point cloud clustering results of each frame obtained by clustering algorithm B, the trajectory of the test personnel in the plane parallel to the ground is tracked and predicted. The corresponding iterative calculation process is as follows:

[0102] a. Initialize the tester's tracking position (t_x, t_y) and tracking speed (t_vx, t_vy):

[0103] t_x=0m, t_y=0m, t_vx=0m / s, t_vy=0m / s

[0104] b. Based on the tester's tracking position (t_x, t_y) and tracking speed (t_vx, t_vy) in the previous frame, the frame interval Frame_T, and the clustering position (xCenter, yCenter) obtained by clustering algorithm B in the current frame, calculate the tester's predicted position (x_predict, y_predict) and prediction bias (delta_x, delta_y) for the current frame:

[0105] x_predict=t_x+t_vx*Frame_T;

[0106] y_predict=t_y+t_vx*Frame_T;

[0107] delta_x=xCenter-x_predict;

[0108] delta_y=yCenter-y_predict;

[0109] c. Update the tracking position (t_x, t_y) and tracking speed (t_vx, t_vy) of the current frame:

[0110] t_x = x_predict + λ * delta_x;

[0111] t_y = y_predict + λ * delta_y;

[0112] t_vx=t_vx+ω / Frame_T*delta_x;

[0113] t_vy=t_vx+ω / Frame_T*delta_y;

[0114] The values ​​of λ and ω are between 0 and 1 and can be adjusted according to actual needs. Here, clustering algorithm B performs better than clustering algorithm A at close ranges within 1m, which is convenient for speed estimation in conjunction with tracking algorithms.

[0115] Step 3: Human detection algorithm;

[0116] The relationships between the various human body states can be described using pseudocode as follows:

[0117] BodyState represents the human body's state, and its value range is as follows: Figure 3 As shown;

[0118] switch(BodyState);

[0119] {

[0120] Case NOTHING:

[0121] A signal is present but the distance is far, exceeding 2.0m but within 2.5m, BodyState = NOTARGET;

[0122] A signal is detected and is relatively close, within 2.0m, but not within the sensing area;

[0123] Then BodyState = BODY_IN;

[0124] If a target signal appears and enters the sensing area, the conditions for opening the lid are met, then BodyState = BODY_STAY;

[0125] In other cases, remain NOTHING.

[0126] Case NOTARGET:

[0127] If there is no signal for 10 consecutive frames, then BodyState = NOTHING;

[0128] A signal is detected and is relatively close, within 2.0m, but not within the sensing area;

[0129] Then BodyState = BODY_IN;

[0130] If a target signal appears and enters the sensing area, the conditions for opening the lid are met, then BodyState = BODY_STAY;

[0131] In other cases, NOTARGET remains unchanged.

[0132] Case BODY_IN:

[0133] If there is no signal for 10 consecutive frames, then BodyState = NOTHING;

[0134] A signal is present but at a distance of more than 2.0m but within 2.5m:

[0135] Then BodyState = NOTARGET;

[0136] If a target signal appears and enters the sensing area, the conditions for opening the lid are met, then BodyState = BODY_STAY;

[0137] If a target signal appears and enters the sensing area but leaves before the conditions for opening the cover are met, then BodyState = BODY_PASS;

[0138] In other cases, BODY_IN is maintained;

[0139] Case BODY_PASS:

[0140] Force BodyState = BODY_AWAY after 1 frame;

[0141] Case BODY_STAY:

[0142] If there is no signal for 10 consecutive frames, then BodyState = NOTHING;

[0143] A signal is present but at a distance of more than 2.0m but within 2.5m:

[0144] Then BodyState = NOTARGET;

[0145] If the target signal leaves the sensing area and the cover-off condition is met, then BodyState = BODY_AWAY;

[0146] Case BODY_AWAY:

[0147] A signal is present but at a distance of more than 2.0m but within 2.5m:

[0148] Then BodyState = NOTARGET;

[0149] If a target signal appears and enters the sensing area, the conditions for opening the lid are met, then BodyState = BODY_STAY;

[0150] If there is no signal for 10 consecutive frames, then BodyState = NOTHING;

[0151] }

[0152] The sensing area is defined as follows:

[0153] (1) The sensing area is a rectangular area of ​​a specified size directly in front of the smart toilet. The width of the rectangular area is the same as that of the smart toilet, and the length ranges from 0.4m to 0.5m.

[0154] (2) When testers enter the sensing area from the left or right sides of the sensing area or from behind the smart toilet, it is considered as entering from the side; otherwise, it is considered as entering from the front.

[0155] Specifically, the conditions for opening the lid, i.e., the conditions for entering the BODY_STAY state, can be divided into two main categories:

[0156] Human body enters the toilet from the front: The lid opens immediately 1 frame after the human body enters the sensing area;

[0157] Human body enters from the side of the toilet: When the human body enters the sensing area and its movement speed along the Y-axis drops below the speed threshold (range 0.30m / s-0.50m / s, in this embodiment the speed threshold is 0.50m / s) and continues for a certain number of frames to reach the frame threshold (range 5-10 frames, in this embodiment 10 frames), the lid will open immediately.

[0158] Step 4: Millimeter-wave radar processing results and smart toilet communication protocol, details are as follows:

[0159] (1) The mainboard of the smart toilet receives the processing result returned by the microcontroller. According to the relevant communication protocol, when the processing result is BODY_STAY, the smart toilet lid is opened; otherwise, the lid is not opened.

[0160] (2) If the current person is in the BODY_STAY state and the toilet seat is open, the smart toilet will close the lid when the person leaves the sensing area and is in one of the three states: BODY_AWAY, NOTARGET, or NOTHING. Otherwise, the lid will not close.

[0161] The present invention provides an intelligent toilet human body detection system based on millimeter-wave radar. Through simulation verification of a large number of human body approach states in all weather conditions and from multiple directions, its recognition accuracy can reach more than 95%.

[0162] The embodiments described above are merely preferred embodiments of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications, improvements, and substitutions without departing from the concept of the present invention, and these all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A smart toilet human body detection system based on millimeter-wave radar, characterized in that, include: Millimeter-wave radar processing module and smart toilet; The millimeter-wave radar processing module is installed behind the smart toilet seat. The module includes a millimeter-wave radio frequency transceiver unit, a microcontroller unit, a transmitting antenna, and a receiving antenna. The millimeter-wave radio frequency transceiver unit is connected to the microcontroller unit, the transmitting antenna, and the receiving antenna, respectively. The millimeter-wave radio frequency transceiver unit generates a corresponding millimeter-wave radar signal based on the radar waveform parameters, receives the echo signal reflected back by the test personnel from the millimeter-wave radar signal, and then transmits the echo signal to the microcontroller unit. The microcontroller unit is a single-chip microcomputer that can be programmed and run human detection programs. It is used to process echo signals to obtain human detection judgment results and transmit them to the smart toilet. The transmitting antenna is used to transmit millimeter-wave radar signals generated by the millimeter-wave radio frequency transceiver unit. The receiving antenna is used to receive millimeter-wave radar signals reflected back in space and transmit them to the millimeter-wave radio frequency transceiver unit. The smart toilet controls the opening or closing of the lid based on the human body detection results returned by the millimeter-wave radar processing module through relevant communication protocols. The human detection algorithm of the system is completed during the processing of echo signals by the microcontroller unit. The operation of the human detection system and the smart toilet includes the following steps: S1. The microcontroller unit obtains data information from multiple points in each frame from the tester based on the echo signal; S2. Run clustering algorithm A to obtain the location and energy information of the test personnel; S3. Run the tracking algorithm to obtain the predicted position of the tester and the movement speed in the X and Y axes; S4. Combine the data information calculated in steps S2 and S3 to run the human detection algorithm and obtain the human detection judgment result. S5. The smart toilet decides whether to open the lid based on the processing results and communication protocol returned by the microcontroller unit. The clustering algorithm A in step S2 includes the following steps: S2.1 Sort the point cloud data of each frame obtained by the millimeter-wave radar and the underlying signal processing in ascending order according to the distance r to the radar center. The sorting algorithm adopts the bubble sort method, which only sorts the sequence number of the point cloud data without changing the point cloud space. S2.2, Traverse all point cloud data in each frame after sorting, starting from ascending order of distance to the radar center r: S2.2.1 If the current point is not assigned, a new class is assigned to the current point, and the current point is marked as assigned. Then, the points in the first preset radius neighborhood are grouped into the class where the current point is located, and the weighted average of the three-dimensional spatial coordinates and the weighted energy value of all point cloud data in the class are updated. At the same time, all points in the class are marked as assigned. If the current point has already been assigned, skip the current loop and continue to the next point; S2.2.2 Repeat process S2.2.1 until all points in the current frame have been allocated; S2.3 After traversing all point cloud data in the current frame, if the number of classes obtained in step S2.2 is greater than 1, classes with fewer than 2 points are first filtered out, and then the class with the larger weighted average energy is selected. The weighted average of the coordinates and the weighted energy value of the class are used as the clustering result returned by the algorithm. The human detection algorithm in step S4 divides the human body state as follows: NOTHING: The millimeter-wave radar did not detect any target; NOTARGET: The millimeter-wave radar received a signal, but no target was detected approaching the smart toilet; BODY_IN: Millimeter-wave radar detected a target approaching the smart toilet, but the target has not yet entered the smart toilet's sensing area and the conditions for opening the lid are not met. BODY_PASS: Millimeter-wave radar detected a target approaching the smart toilet, but the target left the sensing area before the conditions for opening the lid were met after entering the smart toilet's sensing area. BODY_STAY: The target enters the smart toilet's sensing area and meets the conditions for opening the lid, and stays in the current sensing area; BODY_AWAY: The target leaves the smart toilet's sensor area and meets the condition of closing the lid; The relationships between the various human body states are as follows: NOTHING: Maintain or transition to one of the three states: NOTARGET, BODY_IN, or BODY_STAY; NOTARGET: Maintain or transition to one of the three states: NOTHING, BODY_IN, or BODY_STAY. BODY_IN: Maintain or transition to one of the four states: NOTHING, NOTARGET, BODY_PASS, BODY_STAY; BODY_PASS: Cannot be maintained, will be forcibly switched to BODY_AWAY state after one frame; BODY_STAY: Maintain or transition to one of the three states: BODY_AWAY, NOTARGET, or NOTHING; BODY_AWAY: Maintain or transition to one of the three states: NOTHING, NOTARGET, or BODY_STAY. The NOTHING, NOTARGET, and BODY_IN states can all enter the BODY_STAY state without going through other states; The conditions for entering the BODY_STAY state are divided into two categories: Scenario 1: When a person enters the smart toilet's sensing area from the front, the lid opens quickly. Scenario 2: When a person enters the smart toilet's sensing area from the side, the lid opens slowly when the person's movement speed in the Y-axis direction is less than the speed threshold and the number of frames spent in the sensing area reaches the frame number threshold. The criteria for determining whether someone enters the smart toilet's sensor area from the front or the side are as follows: a. The sensing area is a rectangular area of ​​a specified size directly in front of the smart toilet. The width of this rectangular area is the same as that of the smart toilet, and the length ranges from 0.4m to 0.5m. b. When testers enter the sensing area from the left or right sides of the sensing area or from behind the smart toilet, it is considered as entering from the side; otherwise, it is considered as entering from the front. Step S5 specifically includes: S5.1 The smart toilet motherboard receives the processing result returned by the microcontroller. When the processing result is BODY_STAY, it controls the smart toilet to open the lid; otherwise, it does not open the lid. S5.2 If the current person is in the BODY_STAY state and the toilet seat is open, the smart toilet will close the lid when the person leaves the sensing area and is in one of the three states: BODY_AWAY, NOTARGET, or NOTHING; otherwise, the lid will remain open.

2. The intelligent toilet human body detection system based on millimeter-wave radar according to claim 1, characterized in that, The data information in step S1 includes: The frame number of the current frame, frameNow; The distance r from the point cloud acquired in the current frame to the radar center; The 3D Cartesian coordinates (x, y, z) of the point cloud acquired in the current frame; The velocity of the point cloud captured in the current frame; The point cloud power pow acquired in the current frame; The number of valid point clouds acquired in the current frame after processing by millimeter-wave radar is TargetNum.

3. The intelligent toilet human body detection system based on millimeter-wave radar according to claim 1, characterized in that, The tracking algorithm described in step S3 is as follows: S3.

1. Run clustering algorithm B on each frame of point cloud data obtained from millimeter-wave radar acquisition and low-level signal processing. The algorithm steps are as follows: S3.1.

1. Find the maximum power point for each frame of point cloud data obtained from the millimeter-wave radar by traversing and searching, and record its power value MaxPow and its index MaxPowIdx. S3.1.2 Select point cloud data with power greater than 0.4*MaxPow within the second preset radius neighborhood of the maximum power point. Then, classify these points together with the maximum power point into one class and update the weighted average coordinates and weighted energy values ​​of all point cloud data in this class as the clustering result returned by the algorithm. S3.

2. Based on the point cloud clustering results obtained from clustering algorithm B for each frame, the trajectory of the test personnel in the plane parallel to the ground is tracked and predicted. The corresponding iterative calculation process is as follows: S3.2.1 Initialize the test personnel tracking position and tracking speed : ； S3.2.2, Track the position of the tester based on the previous frame. and tracking speed Frame interval time The clustering position of the current frame obtained by clustering algorithm B Calculate the predicted position of the tester in the current frame. and prediction bias : ； ； ； ； S3.2.3 Update the tracking position of the current frame. and tracking speed : ； ； ； ； in The value is between 0 and 1 and can be adjusted according to actual needs. The clustering algorithm B is better than the clustering algorithm A at close ranges within 1m and is used to assist the speed estimation of the tracking algorithm.

Images