A two-wheeled vehicle seat detection method based on millimeter wave radar point cloud occupancy grid analysis

Abstract

The present invention discloses a two-wheeled vehicle seat occupancy detection method based on millimeter-wave radar point cloud occupancy grid analysis, which belongs to the field of millimeter-wave radar technology and includes the following steps: initial seat in-seat signal; S3, data distribution calculation of real-time echo signal of millimeter-wave radar; S4, occupancy probability calculation; S5, update occupancy grid; S6, micro-motion detection; S7, finally judge that when someone is currently seated, the two-wheeled vehicle is powered on; when it is judged that the seat is left, the power supply of the two-wheeled vehicle is disconnected. Through the above-mentioned method, the present invention makes full use of the millimeter-wave radar's sensitivity to micro-motions to solve the pain point problem of the pressure sensor's inability to distinguish between people and objects; the present invention adopts a low-cost millimeter-wave radar to detect the point cloud caused by the deformation of the seat, and introduces an occupancy grid to make the binary detection of seat entry and seat vacancy more stable.

Description

Technical Field

[0001] The present invention relates to the technical field of millimeter wave radars, and in particular to a two-wheeled vehicle seat occupancy detection method based on millimeter wave radar point cloud occupancy grid analysis. Background Art

[0002] In recent years, with the acceleration of urbanization and the increasingly serious traffic congestion problem, two-wheeled electric vehicles have received widespread attention as a green and portable means of travel.

[0003] Traditional two-wheeled electric vehicles require users to use keys or buttons to manually control the power on and off. This method has problems such as inconvenient operation and forgetting to turn off the power, resulting in power waste.

[0004] In recent years, there has also been a rise in methods based on pressure sensors to detect seat entry and exit and start or disconnect the power of electric vehicles. However, this type of solution also has the following disadvantages: a) false alarms caused by the placement of other debris, and b) unstable measured pressure due to bumpy roads.

[0005] As a common type of vehicle-mounted sensor, millimeter-wave radar has the advantages of small size, sensitive motion detection, and good directionality. How to use millimeter-wave radar to realize seat detection of two-wheeled vehicles is a technical problem that needs to be solved.

[0006] Based on this, the present invention designs a two-wheeled vehicle seat occupancy detection method based on millimeter-wave radar point cloud occupancy grid analysis to solve the above problems. Summary of the invention

[0007] In view of the above-mentioned shortcomings of the prior art, the present invention provides a two-wheeled vehicle seat occupancy detection method based on millimeter wave radar point cloud occupancy grid analysis.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0009] A two-wheeled vehicle seat detection method based on millimeter wave radar point cloud occupancy grid analysis includes the following steps:

[0010] S1. First, install the millimeter wave radar inside the seat under the cushion of the two-wheeled vehicle to detect the echo signal and the point cloud changes of sitting and leaving the seat;

[0011] S2, initial in-seat signal, within the initial N-frame time period after the millimeter-wave radar is powered on, the echo signal is counted as the initial distance dimension 1D-FFT sample spectrum; then the occupancy grid is initialized;

[0012] S3, performing data distribution calculation on the real-time echo signal of the millimeter wave radar;

[0013] S4, occupation probability calculation: based on the distance measurement value, angle measurement value and data distribution calculation result of each point, calculate the likelihood occupation probability attached to the measurement point;

[0014] S5, update the occupied grid;

[0015] S6, micro-motion detection; calculating a single-frame sum value matrix, and then calculating a change value of the sum value matrix between frames;

[0016] S7. Determine whether the following two conditions are met at the same time:

[0017] a) There are cells in the occupancy grid whose occupancy confidence is higher than a certain threshold;

[0018] b) The change value of the sum matrix between frames in micro-motion detection is higher than the set threshold A;

[0019] If so, it is determined that someone is currently sitting in the car, and the two-wheeled vehicle is powered on;

[0020] If not, it is determined that the seat is left and the power supply of the two-wheeled vehicle is disconnected.

[0021] Furthermore, the specific steps of step S2 are:

[0022] 2.1. Perform one-dimensional fast Fourier transform (1D-FFT) processing on the echo signal ADC to obtain the distance-dimensional 1D-FFT spectrum corresponding to the scene inside the seat;

[0023] 2.2. The above spectrum data is accumulated, and when the accumulation reaches the Nth frame, the average value of the spectrum data of the initial N frames is calculated as the initial distance dimension 1D-FFT sample spectrum;

[0024] 2.3. Occupancy grid initialization: Taking the installation position of the millimeter-wave radar as the zero point, discretize the plane range corresponding to the seat cushion area into a two-dimensional grid map of (width / grid_size, length / grid_size), and initialize the occupancy probability of each cell in the occupancy grid to 0.5.

[0025] Furthermore, the specific steps of step S3 are:

[0026] 3.1. Perform one-dimensional fast Fourier transform processing on the real-time echo signal ADC of the millimeter-wave radar to obtain the real-time distance dimension 1D-FFT spectrum, and perform two-dimensional fast Fourier transform processing based on the result to obtain the distance-speed dimension spectrum;

[0027] 3.2. Subtract the real-time distance dimension 1D-FFT spectrum from the initial distance dimension 1D-FFT sample spectrum to obtain a difference 1D-FFT spectrum, thereby removing the influence of the initial scene signal data;

[0028] 3.3. Perform two-dimensional fast Fourier transform processing on the difference 1D-FFT spectrum to obtain the distance-velocity spectrum;

[0029] 3.4. Perform constant false alarm processing (CFAR) on the distance-velocity spectrum to filter out noise, obtain the current frame point cloud and the corresponding distance measurement value, Doppler measurement value and angle measurement value, and calculate the corresponding signal-to-noise ratio;

[0030] 3.5. The true existence probability of each measurement point is derived from the signal-to-noise ratio. The formula is:

[0031] det_prob=false_alarm_rate^(1.0 / (1.0+SNR))

[0032] Among them, false_alarm_rate refers to the false alarm rate of the radar, which is obtained by the previous test evaluation, det_prob refers to the probability that the measurement point actually exists, and SNR refers to the signal-to-noise ratio.

[0033] Furthermore, the specific steps of step S4 are:

[0034] 4.1 Fuzzy Range Approximation

[0035] First, the ambiguity of distance measurement and angle measurement is simulated respectively. For the measurement value of each point, the range Delta is selected as the fuzzy area, and the range is mapped to a two-dimensional grid map. The grid vertices closest to the measurement point together form a triangle of the fuzzy range, and the maximum and minimum values ​​of the distance and angle of each range triangle vertex are recorded.

[0036] 4.2. For each measurement point, the distance or angle measurement value is taken as the mean, and the radar distance or angle prior measurement accuracy is taken as the variance to construct a normal distribution, and the distance or angle cumulative probability distribution is calculated based on this. According to the relative position of the distance or angle measurement value within the range, the corresponding probability value on the cumulative probability distribution is obtained, and the occupancy probability value of the distance or angle dimension is obtained after correction by the actual existence probability det_prob of the measurement point;

[0037] 4.3. Multiply the distance occupancy probability value and the angle occupancy probability value obtained to obtain the likelihood occupancy probability associated with the measurement point.

[0038] Furthermore, the specific steps of step S5 are:

[0039] 5.1. Loop through each cell in the grid. If there is a measurement point in the cell, perform the calculation in 5.2.

[0040] 5.2. Formula 1 is used to convert the binary Bayesian filter from the binary problem of seat entry and seat departure into a confidence calculation in time series to avoid the probability value being truncated at 0 and 1.

[0041] (Formula 1)

[0042] in, represents the likelihood occupancy probability of the measurement point in the cell, Represents the confidence that the cell is occupied in the current frame, Represents the confidence of cell occupancy calculated in the previous frame.

[0043] Furthermore, the specific steps of step S6 are:

[0044] 6.1. Perform fast Fourier transform processing on the echo signal ADC to obtain complex FFT data in the distance dimension, and incoherently accumulate the complex distance dimension FFT matrices of multiple chirps in a single frame in a channel to obtain a single frame sum value matrix;

[0045] 6.2. Calculate the change value of the sum value matrix between frames, and find the maximum value of the change value within the distance range of the cushion area. If the maximum value is higher than the set threshold A, it indicates that there is movement within the cushion range.

[0046] Furthermore, in step S5, according to the principle of binary Bayesian filtering, the occupancy probability of the cell is updated according to the likelihood occupancy probability of each measurement point.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] The present invention makes full use of the characteristic of millimeter wave radar being sensitive to micro-movements, and solves the pain point problem that the pressure sensor cannot distinguish between people and objects.

[0049] The present invention adopts a low-cost millimeter-wave radar to detect the point cloud caused by the deformation of the seat when sitting down, and introduces an occupancy grid to make the binary detection of sitting down and leaving the seat more stable. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0051] Figure 1 The present invention is a flow chart of a two-wheeled vehicle seat occupancy detection method based on millimeter wave radar point cloud occupancy grid analysis. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Example

[0053] In some embodiments, please refer to the attached instructions. Figure 1 A two-wheeled vehicle seat detection method based on millimeter wave radar point cloud occupancy grid analysis includes the following steps:

[0054] S1. First, install the millimeter wave radar inside the seat under the cushion of the two-wheeled vehicle to detect the echo signal and the point cloud changes of sitting and leaving the seat;

[0055] S2, initial in-seat signal;

[0056] In order to prevent other debris in the car from affecting subsequent judgments, the echo signal needs to be counted as the initial distance dimension 1D-FFT sample spectrum within the initial N frame time period after the millimeter-wave radar is powered on.

[0057] The specific steps of step S2 are:

[0058] 2.1. Perform one-dimensional fast Fourier transform (1D-FFT) processing on the echo signal ADC to obtain the distance-dimensional 1D-FFT spectrum corresponding to the scene inside the seat;

[0059] 2.2. The above spectrum data is accumulated, and when the accumulation reaches the Nth frame, the average value of the spectrum data of the initial N frames is calculated as the initial distance dimension 1D-FFT sample spectrum.

[0060] 2.3. Occupancy grid initialization: Taking the installation position of the millimeter-wave radar as the zero point, discretize the plane range corresponding to the cushion area into a two-dimensional grid map of (width / grid_size, length / grid_size), and initialize the occupancy probability of each cell in the occupancy grid to 0.5;

[0061] S3, performing data distribution calculation on the real-time echo signal of the millimeter wave radar;

[0062] The specific steps of step S3 are:

[0063] 3.1. Perform one-dimensional fast Fourier transform (1D-FFT) processing on the real-time echo signal ADC of the millimeter-wave radar to obtain the real-time distance dimension 1D-FFT spectrum, and perform two-dimensional fast Fourier transform (2D-FFT) processing based on the result to obtain the distance-speed dimension spectrum.

[0064] 3.2. Subtract the real-time distance dimension 1D-FFT spectrum from the initial distance dimension 1D-FFT sample spectrum to obtain a difference 1D-FFT spectrum, thereby removing the influence of the initial scene signal data.

[0065] 3.3. Perform two-dimensional fast Fourier transform (2D-FFT) processing on the difference 1D-FFT spectrum to obtain the distance-velocity spectrum.

[0066] 3.4. Perform constant false alarm processing (CFAR) on the range-velocity spectrum to filter out noise, obtain the current frame point cloud and the corresponding distance measurement value, Doppler measurement value and angle measurement value, and calculate the corresponding signal-to-noise ratio (SNR).

[0067] 3.5. The true existence probability of each measurement point is derived from the signal-to-noise ratio. The formula is:

[0068] det_prob=false_alarm_rate^(1.0 / (1.0+SNR))

[0069] Among them, false_alarm_rate refers to the false alarm rate of the radar, which is obtained by the previous test evaluation, det_prob refers to the probability that the measurement point actually exists, and SNR refers to the signal-to-noise ratio.

[0070] S4, occupation probability calculation;

[0071] Through steps S1-S3, the distance measurement value and angle measurement value of the point cloud measured in the current frame can be obtained, and the likelihood occupancy probability attached to the measurement point can be obtained according to the measurement value of each point;

[0072] The specific steps of step S4 are:

[0073] 4.1, fuzzy range approximation;

[0074] First, the ambiguity of distance measurement and angle measurement is simulated respectively. For the measurement value of each point, the range Delta is selected as the fuzzy area, for example, ((range (distance) - Delta, angle (angle) - Delta), (range + Delta, angle + Delta)), and the range is mapped to the two-dimensional grid map. The grid vertices closest to the measurement point together form a triangle of the fuzzy range, and the maximum and minimum values ​​of the distance and angle of each range triangle vertex are recorded;

[0075] 4.2. For each measurement point, the distance or angle measurement value is taken as the mean, and the radar distance or angle prior measurement accuracy is taken as the variance to construct a normal distribution, and the distance or angle cumulative probability distribution (CDF) is calculated based on this. According to the relative position of the distance or angle measurement value within the range, the corresponding probability value on the cumulative probability distribution is obtained, and the occupancy probability value of the distance or angle dimension is obtained after correction by the actual existence probability det_prob of the measurement point;

[0076] 4.3. Multiply the distance occupation probability value and the angle occupation probability value obtained to obtain the likelihood occupation probability associated with the measurement point;

[0077] S5, update the occupied grid;

[0078] According to the principle of binary Bayesian filtering, the occupancy probability of the cell is updated according to the likelihood occupancy probability of each measurement point;

[0079] The specific steps of step S5 are:

[0080] 5.1. Loop through each cell in the grid. If there is a measurement point in the cell, perform the calculation in 5.2.

[0081] 5.2. Formula 1 is used to convert the binary Bayesian filter from the binary problem of seat entry and seat departure into a confidence calculation in time series to avoid the probability value being truncated at 0 and 1.

[0082] (Formula 1)

[0083] in, represents the likelihood occupancy probability of the measurement point in the cell, Represents the confidence that the cell is occupied in the current frame, Represents the confidence of cell occupancy calculated in the previous frame.

[0084] S6, micro motion detection;

[0085] In addition to detecting occupancy probability, it is also necessary to detect micro-movements caused by people taking seats to avoid false alarms caused by stationary objects placed on seats.

[0086] The specific steps of step S6 are:

[0087] 6.1. Perform fast Fourier transform (FFT) processing on the echo signal ADC to obtain complex FFT data in the distance dimension, and incoherently accumulate the complex distance dimension FFT matrices of multiple chirps in a single frame in a channel to obtain a single frame sum value matrix;

[0088] 6.2. Calculate the change value of the sum value matrix between frames, and find the maximum value of the change value within the distance range of the cushion area. If the maximum value is higher than a certain threshold, it indicates that there is a certain movement within the cushion range, such as human breathing, rather than a static object;

[0089] S7. Final judgment

[0090] Determine whether both of the following conditions are met:

[0091] a) There are cells in the occupancy grid whose occupancy confidence is higher than a certain threshold;

[0092] b) The change value of the sum matrix between frames in micro-motion detection is higher than a certain threshold;

[0093] If so, it is determined that someone is currently sitting in the car, and the two-wheeled vehicle is powered on;

[0094] If not, it is determined that the seat is left and the power supply of the two-wheeled vehicle is disconnected.

[0095] The present invention makes full use of the characteristic of millimeter wave radar being sensitive to micro-movements, and solves the pain point problem that the pressure sensor cannot distinguish between people and objects.

[0096] The present invention adopts a low-cost millimeter-wave radar to detect the point cloud caused by the deformation of the seat when sitting down, and introduces an occupancy grid to make the binary detection of sitting down and leaving the seat more stable.

[0097] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A two-wheeled vehicle seat detection method based on millimeter wave radar point cloud occupancy grid analysis, characterized in that: The following steps are involved: S1. First, install the millimeter wave radar inside the seat under the cushion of the two-wheeled vehicle to detect the echo signal and the point cloud changes of sitting and leaving the seat; S2, initial in-seat signal, within the initial N-frame time period after the millimeter-wave radar is powered on, the echo signal is counted as the initial distance dimension 1D-FFT sample spectrum; then the occupancy grid is initialized; S3, performing data distribution calculation on the real-time echo signal of the millimeter wave radar; S4, occupation probability calculation: based on the distance measurement value, angle measurement value and data distribution calculation result of each point, calculate the likelihood occupation probability attached to the measurement point; S5, update the occupied grid; S6, micro-motion detection; calculating a single-frame sum value matrix, and then calculating a change value of the sum value matrix between frames; S7. Determine whether the following two conditions are met at the same time: a) There are cells in the occupancy grid whose occupancy confidence is higher than a certain threshold; b) The change value of the sum matrix between frames in micro-motion detection is higher than the set threshold A; If so, it is determined that someone is currently sitting in the car, and the two-wheeled vehicle is powered on; If not, it is determined that the seat is left and the power supply of the two-wheeled vehicle is disconnected.

2. The two-wheeled vehicle seat detection method based on millimeter wave radar point cloud occupancy grid analysis according to claim 1 is characterized in that: The specific steps of step S2 are: 2.

1. Perform one-dimensional fast Fourier transform processing on the echo signal ADC to obtain the distance dimension 1D-FFT spectrum corresponding to the scene inside the car seat; 2.

2. The above spectrum data is accumulated, and when the accumulation reaches the Nth frame, the average value of the spectrum data of the initial N frames is calculated as the initial distance dimension 1D-FFT sample spectrum; 2.

3. Occupancy grid initialization: Taking the installation position of the millimeter-wave radar as the zero point, discretize the plane range corresponding to the seat cushion area into a two-dimensional grid map of (width / grid_size, length / grid_size), and initialize the occupancy probability of each cell in the occupancy grid to 0.

5.

3. The two-wheeled vehicle seat detection method based on millimeter wave radar point cloud occupancy grid analysis according to claim 2 is characterized in that: The specific steps of step S3 are: 3.

1. Perform one-dimensional fast Fourier transform processing on the real-time echo signal ADC of the millimeter-wave radar to obtain the real-time distance dimension 1D-FFT spectrum, and perform two-dimensional fast Fourier transform processing based on the result to obtain the distance-speed dimension spectrum; 3.

2. Subtract the real-time distance dimension 1D-FFT spectrum from the initial distance dimension 1D-FFT sample spectrum to obtain a difference 1D-FFT spectrum, thereby removing the influence of the initial scene signal data; 3.

3. Perform two-dimensional fast Fourier transform processing on the difference 1D-FFT spectrum to obtain the distance-velocity spectrum; 3.

4. Perform constant false alarm processing (CFAR) on the distance-velocity spectrum to filter out noise, obtain the current frame point cloud and the corresponding distance measurement value, Doppler measurement value and angle measurement value, and calculate the corresponding signal-to-noise ratio; 3.

5. The true existence probability of each measurement point is derived from the signal-to-noise ratio. The formula is: det_prob=false_alarm_rate^(1.0 / (1.0+SNR)) Among them, false_alarm_rate refers to the false alarm rate of the radar, which is obtained by the previous test evaluation, det_prob refers to the probability that the measurement point actually exists, and SNR refers to the signal-to-noise ratio.

4. The two-wheeled vehicle seat detection method based on millimeter wave radar point cloud occupancy grid analysis according to claim 3 is characterized in that: The specific steps of step S4 are: 4.1 Fuzzy Range Approximation First, the ambiguity of distance measurement and angle measurement is simulated respectively. For the measurement value of each point, the range Delta is selected as the fuzzy area, and the range is mapped to a two-dimensional grid map. The grid vertices closest to the measurement point together form a triangle of the fuzzy range, and the maximum and minimum values ​​of the distance and angle of each range triangle vertex are recorded. 4.

2. For each measurement point, the distance or angle measurement value is taken as the mean, and the radar distance or angle prior measurement accuracy is taken as the variance to construct a normal distribution, and the distance or angle cumulative probability distribution is calculated based on this. According to the relative position of the distance or angle measurement value within the range, the corresponding probability value on the cumulative probability distribution is obtained, and the occupancy probability value of the distance or angle dimension is obtained after correction by the actual existence probability det_prob of the measurement point; 4.

3. Multiply the distance occupancy probability value and the angle occupancy probability value obtained to obtain the likelihood occupancy probability associated with the measurement point.

5. The two-wheeled vehicle seat occupancy detection method based on millimeter wave radar point cloud occupancy grid analysis according to claim 4 is characterized in that: The specific steps of step S5 are: 5.

1. Loop through each cell in the grid. If there is a measurement point in the cell, perform the calculation in 5.

2. 5.

2. Formula 1 is used to convert the binary problem of seat entry and seat departure into a confidence calculation in time series to avoid truncation of probability values ​​at 0 and 1. (Formula 1) in, represents the likelihood occupancy probability of the measurement point in the cell, Represents the confidence that the cell is occupied in the current frame, Represents the confidence of cell occupancy calculated in the previous frame.

6. The two-wheeled vehicle seat detection method based on millimeter wave radar point cloud occupancy grid analysis according to claim 5 is characterized in that: The specific steps of step S6 are: 6.

1. Perform fast Fourier transform processing on the echo signal ADC to obtain complex FFT data in the distance dimension, and incoherently accumulate the complex distance dimension FFT matrices of multiple chirps in a single frame in a channel to obtain a single frame sum value matrix; 6.

2. Calculate the change value of the sum value matrix between frames, and find the maximum value of the change value within the distance range of the cushion area. If the maximum value is higher than the set threshold A, it indicates that there is movement within the cushion range.

7. The two-wheeled vehicle seat occupancy detection method based on millimeter wave radar point cloud occupancy grid analysis according to claim 6 is characterized in that: In step S5, according to the principle of binary Bayesian filtering, the occupancy probability of the cell is updated according to the likelihood occupancy probability of each measurement point.

Images