Multi-radar cooperative detection method, device and equipment in parking sentry mode

Abstract

The invention discloses a multi-radar cooperative detection method, device and equipment in a parking sentry mode, and the method comprises the steps: controlling an IR-UWB radar to transmit a UWB signal, obtaining a channel pulse response, and generating a first fast and slow time matrix; removing a direct current component of a slow time dimension in the first fast and slow time matrix through fast Fourier transform to obtain a second fast and slow time matrix; performing fast Fourier transform processing on the second fast and slow time matrix to obtain a distance Doppler matrix; performing distance-amplitude plane projection on the distance Doppler matrix to obtain a one-dimensional signal along the distance dimension, and performing one-dimensional GO-CFAR detection on the one-dimensional signal along the distance dimension to obtain a target detection result of whether an intrusion target exists; and performing cancellation on the target detection results at different moments by using XOR operation to obtain an external intrusion matrix. By using a multi-radar cooperative detection algorithm, the false alarm rate is significantly reduced, the algorithm complexity is simplified, the energy consumption is reduced, the privacy protection is enhanced, the detection precision and reliability are improved, and the method is suitable for intrusion detection in a parking sentry mode.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of communication, in particular to a multi-radar cooperative detection method, device and equipment for a parking sentry mode. BACKGROUND

[0002] The parking sentry mode is an intelligent vehicle safety mode, which is used to detect and record possible threats around the vehicle in the parking state, and issue an alarm if necessary. The system can reduce the risk of potential damage and theft of the vehicle, and record evidence. The parking sentry mode can alert intruders or vehicles through horn sounding or other means when necessary, thereby preventing problems from occurring; on the other hand, it can record the process through external cameras when the vehicle is damaged or stolen, providing evidence for subsequent accountability. With the development of intelligent vehicles, the parking sentry mode has become an important means to provide more security for parking.

[0003] In related technologies, two types of sentry modes are mainly used in the field of vehicle parking safety protection, including a monitoring scheme based on a vibration sensor and a pure visual perception scheme based on a camera.

[0004] The monitoring scheme based on the vibration sensor triggers an alarm by detecting physical vibration or impact of the vehicle body, but its technical defects are as follows: (1) contradiction between environmental interference sensitivity and detection sensitivity: the vibration sensor is easily disturbed by non-threatening vibrations (such as strong wind and passing vehicles) leading to false alarms, and reducing the sensitivity of the vibration sensor may miss minor scratches and other low-intensity threats; (2) limitations of local perception and threat type detection: only mechanical vibrations of the vehicle body (such as door collision) can be responded to, and areas such as the roof and chassis and non-vibration damage behaviors (such as static scratches and license plate theft) cannot be effectively monitored; (3) long-term energy consumption burden: the continuous standby state of the vibration sensor combined with the high-power mode of the linked camera accelerates the battery power consumption during vehicle parking, significantly shortening the sustainable working time of the sentry mode.

[0005] The pure visual perception scheme based on the camera relies on computer vision algorithms to analyze environmental information in real time to achieve active early warning, and its technical defects are as follows: (1) environmental dependence of algorithm performance: deep learning models are sensitive to lighting conditions (such as low light at night and backlight) and adverse weather (rain, snow and fog), and need to rely on large-scale multi-scene data sets to improve generalization ability; (2) system power consumption and endurance constraints: continuous image data collection by multiple cameras and real-time computation at the edge end result in high energy consumption, with daily power consumption reaching 5%-10%, which severely limits the endurance of the sentry mode; (3) privacy compliance risks: sensitive information such as pedestrian biometric features (such as faces) and vehicle identifiers (such as license plates) are collected indiscriminately, which poses a risk of data leakage and may violate privacy protection regulations.

[0006] It can be seen that the vibration sensor scheme is difficult to achieve accurate threat recognition due to the limitation of physical sensing mechanism, and the pure visual scheme has the potential of active early warning, but is restricted by algorithm robustness, energy consumption and privacy problems, both of which are difficult to meet the needs of low false alarm rate, full scene adaptability and privacy compliance, therefore, how to break through these technical bottlenecks to realize the efficient, accurate and compliant operation of the parked sentinel mode has become a technical problem to be solved. SUMMARY

[0007] The application provides a multi-radar cooperative detection method, device and equipment for a parked sentinel mode, which can reduce the false alarm rate of the parked sentinel mode, prolong the endurance time of the parked mode, simplify the complexity of the dynamic target detection algorithm, and improve the accuracy of the intrusion detection of the parked sentinel mode.

[0008] In a first aspect, an embodiment of the application provides a multi-radar cooperative detection method for a parked sentinel mode, which comprises: multiplexing a UWB positioning chip into an IR-UWB radar, controlling the IR-UWB radar to emit a UWB signal, and acquiring a channel impulse response to generate a first fast-slow time matrix; removing a direct current component of the slow time dimension in the first fast-slow time matrix by fast Fourier transform to obtain a second fast-slow time matrix; performing fast Fourier transform processing on the second fast-slow time matrix to obtain a range-Doppler matrix; performing projection on a range-amplitude plane on the range-Doppler matrix to obtain a one-dimensional signal along the range dimension, and performing one-dimensional GO-CFAR detection on the one-dimensional signal along the range dimension to obtain a target detection result of whether an intrusion target exists; using exclusive OR operation to cancel out the target detection results at different times to obtain an external intrusion matrix.

[0009] In combination with the first aspect, in an implementation, multiplexing a UWB positioning chip into an IR-UWB radar, controlling the IR-UWB radar to emit a UWB signal, and acquiring a channel impulse response to generate a first fast-slow time matrix comprises: recording each frame of channel impulse response acquired by the IR-UWB radar as a row vector:

[0010] wherein, is the i-th frame of channel impulse response data, and M is the number of sampling points of each frame of channel impulse response; generating a first fast-slow time matrix according to N frames of channel impulse response:

[0011] wherein, N is the total frame number of the first time-slow matrix.

[0012] In combination with the first aspect, in an implementation, the direct current component of the slow time dimension in the first time-slow matrix is removed by fast Fourier transform to obtain a second time-slow matrix, including: The first time-slow matrix is expressed as a sequence along the slow time dimension:

[0013] wherein, , is the signal intensity of the jth column in the first time-slow matrix at different time intervals n, is the element of the jth column in the first time-slow matrix at the nth row; Fast Fourier transform is performed on each column signal in the first time-slow matrix to obtain a first frequency domain signal:

[0014] wherein, is the first frequency domain signal, and k is a frequency domain index; The direct current component in the first frequency domain signal is removed, and inverse fast Fourier transform is performed on the first frequency domain signal after removing the direct current component to obtain a slow time dimension signal after removing the direct current component:

[0015]

[0016] wherein, is the first frequency domain signal after removing the direct current component, is the slow time dimension signal after removing the direct current component; The slow time dimension signal after removing the direct current component is recombined to obtain the second time-slow matrix:

[0017] wherein, , is the second time-slow matrix.

[0018] In combination with the first aspect, in an implementation, fast Fourier transform is performed on the second time-slow matrix to obtain a range-Doppler matrix, including: Fast Fourier transform is performed on each column of the second time-slow matrix to obtain a corresponding second frequency domain signal, and each second frequency domain signal is expressed as a column vector:

[0019] Combining all column vectors yields the range-Doppler matrix:

[0020] in, is a column vector, is the range-Doppler matrix.

[0021] In combination with the first aspect, in one embodiment, projecting the range-Doppler matrix onto a range-amplitude plane to obtain a one-dimensional signal along the range dimension includes: The maximum value of each column of the range-Doppler matrix is ​​taken to obtain the one-dimensional signal along the range dimension:

[0022] Wherein, V is the one-dimensional signal along the distance dimension, Represents the maximum value of each column from the 1st to the Mth column in the range-Doppler matrix.

[0023] In conjunction with the first aspect, in one embodiment, performing one-dimensional GO-CFAR detection on the one-dimensional signal along the range dimension to obtain a target detection result of whether an intruder target exists includes: The algorithm parameters for one-dimensional GO-CFAR detection are set as follows: the test unit window size is γ, the protection unit window size is α, and the threshold factor is λ; Calculate the forward reference window mean of the i-th distance unit as:

[0024] Calculate the backward reference window mean of the i-th distance unit:

[0025] Estimate the size of background clutter for the i-th detection unit, and select the larger of the forward reference window mean and the backward reference window mean as the detection estimation result:

[0026] According to the preset judgment criteria, Whether there is an intrusion target in each detection unit:

[0027] in, Indicates that there is no intrusion target in the i-th detection unit, Indicates that there is an intrusion target in the i-th detection unit; The judgment result is recorded as:

[0028] Combining the decision results of all detection units to obtain the target detection result:

[0029] wherein, the target detection result, the target detection result of the first detection unit to the Mth detection unit.

[0030] In combination with the first aspect, in an implementation, the target detection results at different times are cancelled using XOR operation to obtain an external intrusion matrix, including: XOR operation is performed on the target detection results of each distance unit to obtain an external intrusion matrix:

[0031] wherein, the external intrusion matrix, the target detection result at t0, the target detection result at t0, the target detection result at t0, the target detection result at t0.

[0032] In combination with the first aspect, in an implementation, the IR-UWB radar is arranged on the vehicle in an anchor point arrangement mode of four-anchor-point TDoA positioning.

[0033] In the second aspect, the embodiments of the present application provide a multi-radar cooperative detection device in a parked sentry mode, including: A generating module is configured to multiplex a UWB positioning chip into an IR-UWB radar, control the IR-UWB radar to emit a UWB signal, and acquire a channel impulse response to generate a first fast-slow-time matrix; A first transforming module is configured to remove a direct current component in a slow-time dimension of the first fast-slow-time matrix through fast Fourier transform to obtain a second fast-slow-time matrix; A second transforming module is configured to perform fast Fourier transform processing on the second fast-slow-time matrix to obtain a range-Doppler matrix; A detection module is configured to perform projection on a range-amplitude plane of the range-Doppler matrix to obtain a one-dimensional signal along a range dimension, and perform one-dimensional GO-CFAR detection on the one-dimensional signal along the range dimension to obtain a target detection result of whether an intrusion target exists; A cancelling module is configured to cancel the target detection results at different times using XOR operation to obtain an external intrusion matrix.

[0034] In a third aspect, an embodiment of the present application provides a parked sentry mode multi-radar cooperative detection device, the parked sentry mode multi-radar cooperative detection device comprising a processor, a memory, and a parked sentry mode multi-radar cooperative detection program stored in the memory and executable by the processor, wherein the parked sentry mode multi-radar cooperative detection program, when executed by the processor, implements the steps of the parked sentry mode multi-radar cooperative detection method according to any one of the preceding aspects.

[0035] The technical scheme provided by the embodiments of the present application has the following beneficial effects: By multiplexing the UWB positioning chip as an IR-UWB radar, controlling the IR-UWB radar to emit a UWB signal, and acquiring a channel impulse response to generate a first fast-slow time matrix, removing the direct current component of the slow time dimension in the first fast-slow time matrix through fast Fourier transform to obtain a second fast-slow time matrix, performing fast Fourier transform processing on the second fast-slow time matrix to obtain a range-doppler matrix, performing projection on the range-amplitude plane of the range-doppler matrix to obtain a one-dimensional signal along the range dimension, performing one-dimensional GO-CFAR detection on the one-dimensional signal along the range dimension to obtain a target detection result of whether an intrusion target exists, and using XOR operation to cancel out the target detection results at different times to obtain an external intrusion matrix, the potential threat in parking can be perceived in advance, the algorithm complexity is simplified, a low false alarm rate can be maintained in a complex parking environment, the detection accuracy is effectively improved, the IR-UWB radar is used as a main sensing unit to perceive potential threats in advance, the detection accuracy is improved, the one-dimensional maximum selection constant false alarm (GO-CFAR) algorithm is selected for intrusion target detection, the algorithm complexity is simplified, a low false alarm rate can be maintained in a complex parking environment, through multi-time target detection cancellation, quasi-stationary objects are filtered out, and the accuracy of intrusion detection is improved. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 A flowchart of an embodiment of the parked sentry mode multi-radar cooperative detection method of the present application; Figure 2 A radar arrangement diagram of the present application; Figure 3 An intrusion detection algorithm flowchart of the present application; Figure 4 A one-dimensional GO-CFAR algorithm flowchart of the present application; Figure 5 A function module diagram of an embodiment of the parked sentry mode multi-radar cooperative detection device of the present application; Figure 6 A hardware structure diagram of the parked sentry mode multi-radar cooperative detection device involved in the embodiment of the present application. DETAILED DESCRIPTION

[0037] In order to better understand the technical scheme of the present application, the technical scheme of the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0038] In order to make the purpose, technical scheme and advantages of the present application clearer, the embodiments of the present application will be further described in detail below in conjunction with the drawings.

[0039] In a first aspect, the embodiments of the present application provide a multi-radar cooperative detection method in a parking sentry mode.

[0040] In an embodiment, with reference to Figure 1 , Figure 1 The flowchart of the first embodiment of the multi-radar cooperative detection method in a parking sentry mode of the present application is shown in FIG. 1. As shown in FIG. 1, the multi-radar cooperative detection method in a parking sentry mode includes the following steps. Figure 1 Step S101, multiplexing a UWB positioning chip into an IR-UWB radar, controlling the IR-UWB radar to emit a UWB signal, and acquiring a channel impulse response to generate a first fast-slow time matrix.

[0041] In an embodiment, the IR-UWB radar is arranged on the vehicle in an anchor point arrangement mode of four-anchor-point TDoA positioning.

[0042] It is worth noting that UWB (Ultra-Wideband) is a wireless communication technology that can transmit and receive signals over a very wide frequency band. UWB positioning chips are usually used for high-precision positioning and ranging, with high resolution and low power consumption. In the present application, the UWB positioning chip is multiplexed into an IR-UWB radar (Impulse Radio Ultra-Wideband Radar), i.e., radar detection using UWB signals.

[0043] TDoA (Time Difference of Arrival) is a positioning technology that determines the position of a target by measuring the time difference of signals arriving at multiple known-position anchor points. As shown in FIG. 2, the TDoA positioning principle is shown in FIG. 2. Figure 2 ​As shown, in the anchor arrangement mode of the four-anchor TDoA positioning in the embodiment, the four anchors are arranged in a rectangle in space. The four anchors are arranged on the vehicle, and the specific positions are optimized according to the structure of the vehicle and the detection requirements. This arrangement not only meets the requirements of the positioning function, but also indirectly calculates the position of the object relative to the vehicle through the directional multi-radar ranging mode.

[0044] The arrangement position of the radar needs to consider the requirements of the positioning function, and also consider the hardware limitations of the UWB positioning chip. Such a chip usually has only one receiving channel and does not have the hardware basis to realize radar angle measurement, so it cannot determine the direction of the object relative to the vehicle through a single radar. In order to realize the selective opening of the camera at different positions and play a role in saving power, the application adopts a directional multi-radar ranging mode, which indirectly calculates the position of the object relative to the vehicle through the cooperative work of multiple radars. Such a radar arrangement not only is compatible with the UWB four-anchor TDoA positioning, but also can effectively improve the positioning accuracy and detection capability of the system.

[0045] Further, the IR-UWB radar is controlled to periodically emit UWB signals at a certain frequency, wherein the emission frequency and pulse width of the signals can be adjusted according to specific application requirements. These UWB signals will encounter various objects (such as obstacles around the vehicle, pedestrians, etc.) during propagation and produce echoes, and the IR-UWB radar performs channel impulse response estimation CIRE during signal reception to obtain the channel impulse response CIR.

[0046] Each frame of channel impulse response obtained by the IR-UWB radar is recorded as a row vector form:

[0047] wherein, represents the current received frame of CIR data, is the i-th frame of channel impulse response data, and M is the number of sampling points of each frame of channel impulse response.

[0048] After receiving N frames of channel impulse response, a first fast-slow time matrix is generated according to the N frames of channel impulse response:

[0049] wherein, is the first fast-slow time matrix, and N is the total number of frames of the first fast-slow time matrix. Therefore, the first fast-slow time matrix is an N x M matrix.

[0050] In step S102, the direct current component of the slow time dimension in the first fast-slow time matrix is removed by fast Fourier transform to obtain a second fast-slow time matrix.

[0051] After obtaining the first fast-slow time matrix, intrusion detection can be performed. The basis of the intrusion detection algorithm in this embodiment is the speed measurement and distance measurement result of each IR-UWB radar. On this basis, a dynamic target detection algorithm is implemented. After each radar filters out quasi-static objects without displacement according to multiple target detection results, target positioning and relative speed are obtained according to the target detection results of each radar, and then early warning decisions are made according to certain criteria.

[0052] As shown in Figure 2 , in order to improve the accuracy of intrusion detection, it is necessary to filter noise from the original data before detection to improve the signal-to-noise ratio of the data. Static objects do not displace, and the reflection intensity of the echo signal caused by the static objects at different times in a short time will not change basically, which mainly manifests as a direct current component along the slow time dimension. Therefore, by removing the direct current component along the slow time dimension, static object noise can be effectively filtered out, false positives can be reduced, and the accuracy of detection can be improved.

[0053] In an embodiment, the direct current component along the slow time dimension in the first fast-slow time matrix is removed by fast Fourier transform to obtain a second fast-slow time matrix, including: The sequence representation of the first fast-slow time matrix along the slow time dimension is:

[0054] wherein, , is the signal intensity of the jth column in the first time fast-slow matrix at different time intervals n, is the element of the jth column in the first time fast-slow matrix at the nth row; Fast Fourier transform is performed on each column signal in the first fast-slow time matrix to obtain a first frequency domain signal:

[0055] wherein, is the first frequency domain signal, and k is the frequency domain index; The direct current component in the first frequency domain signal is removed, and inverse fast Fourier transform is performed on the first frequency domain signal after removing the direct current component to obtain a slow time dimension signal after removing the direct current component:

[0056]

[0057] wherein, is the first frequency domain signal after removing the direct current component, is the slow time dimension signal after removing the direct current component; The slow time dimension signal after removing the direct current component is recombined to obtain the second fast-slow time matrix:

[0058] wherein, , is the second slow-time matrix.

[0059] The embodiment effectively filters out static object noise by removing the direct current component of the slow-time dimension, improves the signal-to-noise ratio of the signal, which makes the signal of the dynamic target more prominent, and facilitates subsequent detection and processing. The echo signal of the static object is mainly reflected as a direct current component. Removing these components can reduce false positives caused by static objects and improve the accuracy of intrusion detection. Through FFT and IFFT processing, the signal is analyzed and processed in the frequency domain, optimizing the data processing process and improving the real-time performance and efficiency of the system.

[0060] Step S103, performing fast Fourier transform processing on the second fast-slow time matrix to obtain a range-Doppler matrix.

[0061] In the embodiment, the fast Fourier transform processing on the second fast-slow time matrix is Doppler dimension FFT processing. The Doppler dimension is the velocity dimension. The Doppler dimension FFT is to perform FFT on the range-time matrix along the slow-time dimension, and the FFT result reflects the phase difference between the two frames of signals.

[0062] Since the time interval of two frames of CIR data of the IR-UWB radar is in the order of milliseconds, even for a car traveling at a speed of 60km / h, the time travel distance within the two frames of CIR data is only a few centimeters, which is much smaller than its range resolution. Therefore, in the parked sentry mode, the target to be detected is always in the same range cell in the fast-slow time matrix, and its movement is reflected as a phase difference between the two frames of CIR signals rather than a movement of the range cell. Through Doppler dimension FFT, the phase difference can be converted into a Doppler shift in the frequency domain, thereby improving the resolution.

[0063] In an embodiment, the fast Fourier transform processing on the second fast-slow time matrix to obtain a range-Doppler matrix comprises: performing fast Fourier transform on each column of the second fast-slow time matrix to obtain a corresponding second frequency domain signal, and representing each second frequency domain signal as a column vector, or the Doppler dimension FFT can directly take the column vector representation. The column vector is:

[0064] Combining all column vectors to obtain the range-Doppler matrix:

[0065] wherein,​​ is a column vector, is the range-doppler matrix.

[0066] It is worth noting that the embodiment converts the phase difference between two frames of CIR signals into Doppler frequency shift in the frequency domain through Doppler FFT, which significantly improves the velocity resolution. This enables the system to more accurately detect the velocity of the target, thereby improving the accuracy of intrusion detection.

[0067] In step S104, the range-doppler matrix is projected on the range-amplitude plane to obtain a one-dimensional signal along the range dimension, and one-dimensional GO-CFAR detection is performed on the one-dimensional signal along the range dimension to obtain a target detection result of whether an intrusion target exists.

[0068] In an embodiment, projecting the range-doppler matrix on the range-amplitude plane to obtain a one-dimensional signal along the range dimension comprises: taking a maximum value of each column of the range-doppler matrix DR to obtain the one-dimensional signal along the range dimension:

[0069] wherein V is the one-dimensional signal along the range dimension, represents the maximum value of each column from the first 1 to the Mth column in the range-doppler matrix.

[0070] Further, as shown in Figure 3 performing one-dimensional GO-CFAR detection on the one-dimensional signal along the range dimension to obtain a target detection result of whether an intrusion target exists comprises: setting the algorithm parameters of one-dimensional GO-CFAR detection, including a cell window size of γ, a guard cell window size of α, and a threshold factor of λ; calculating a forward reference window mean value of the i-th range cell as:

[0071] calculating a backward reference window mean value of the i-th range cell as:

[0072] performing background clutter size estimation on the i-th detection cell, and selecting the larger one of the forward reference window mean value and the backward reference window mean value as a detection estimation result:

[0073] It is worth noting that the criterion for GO-CFAR algorithm to perform background clutter size estimation on the i-th detection cell is to select and ​The larger one is taken as the detection estimation result, and when the number of the forward or backward detection units of the detection unit does not meet the window size requirement, the backward or forward window mean value meeting the window size requirement is directly selected as the detection estimation result.

[0074] whether the intrusion target exists in the i-th detection unit according to a preset decision criterion:

[0075] Among them, represents that the intrusion target does not exist in the i-th detection unit, represents that the intrusion target exists in the i-th detection unit. The decision result is recorded as:

[0076] The decision results of all detection units are combined to obtain the target detection result:

[0077] Among them, is the target detection result, is the target detection result of the first to M-th detection units.

[0078] It is worth noting that the idea of the GO-CFAR algorithm is to set the decision threshold of each detection unit according to the statistics of the noise near each detection unit, so as to keep the target detection false alarm probability unchanged under different clutter backgrounds. This method is particularly suitable for environments with large background noise changes, and can effectively improve the robustness of target detection.

[0079] The one-dimensional GO-CFAR algorithm can be applied to the fast-slow time matrix and can detect whether the target exists at different distances. By statistically analyzing the noise near each detection unit, the decision threshold of each detection unit is set, so that the target detection false alarm probability remains unchanged under different clutter backgrounds. The two-dimensional GO-CFAR algorithm can be applied to the range-doppler matrix and can detect whether the target exists under different range and speed conditions. The detection object of the two-dimensional GO-CFAR algorithm increases by one dimension (speed), but its time complexity is O(n2), which is higher than the time complexity O(n) of the one-dimensional GO-CFAR algorithm.

[0080] In order to ensure the real-time performance and deployment cost of the algorithm, and at the same time detect the motion amplitude of the target, the present application selects the range-amplitude plane projection of the range-doppler matrix for one-dimensional FFT processing, instead of directly applying the two-dimensional GO-CFAR algorithm.

[0081] ​This embodiment uses a one-dimensional GO-CFAR algorithm to significantly reduce computational complexity, improve the real-time performance of the system, and maintain a constant false alarm probability for target detection under different clutter backgrounds, thereby improving detection accuracy and reliability. Range-amplitude plane projection simplifies algorithm complexity, improves the algorithm's real-time performance, and improves deployment cost-effectiveness. By statistically analyzing the noise near each unit to be detected and dynamically adjusting the detection threshold, the system can adapt to complex background noise environments and reduce false alarms.

[0082] Step S105: Use an XOR operation to cancel the target detection results at different times to obtain an external intrusion matrix.

[0083] It's worth noting that targets that remain essentially unchanged over short periods of time, such as windswept branches and shrubs, are considered quasi-static. By comparing target detection results at different times, if the same distance unit shows no movement between two consecutive detections, the target is considered quasi-static and can be excluded from the list of intrusion targets. By filtering out quasi-static targets, the system can more accurately identify true intrusion targets, improving the precision of intrusion detection and ultimately enhancing the accuracy and reliability of the system.

[0084] In one embodiment, an XOR operation is performed on the target detection result of each range cell to obtain an external intrusion matrix:

[0085] in, is the external intrusion matrix, is the target detection result at time t0, is the interval from time t0 After the duration Target detection results at the moment.

[0086] As a preferred embodiment, when no external intrusion is detected, the camera controlling the parking sentry mode linkage is in an off state; when an external intrusion is detected, the camera controlling the parking sentry mode linkage is activated, thereby significantly reducing system power consumption.

[0087] The multi-radar collaborative detection method for the parking sentry mode provided in the embodiment of the present application uses IR-UWB radar as the main sensing unit. Compared with pure vision solutions, it overcomes environmental interference such as light, rain, snow, and fog, and significantly improves the accuracy and reliability of detection. Compared with vibration sensor solutions, IR-UWB radar can sense potential threats (such as human approach) in advance, further improving detection accuracy.

[0088] Through the multi-radar collaborative architecture, the camera is activated only when the radar determines that there is an intrusion risk, which significantly reduces the system power consumption. Compared with the pure vision scheme, this dynamic energy consumption optimization mechanism significantly prolongs the endurance time of the parking mode, improves the practicability and economy of the system. At the same time, the camera adopts a non-continuous operation mode, which avoids indiscriminate collection of sensitive data such as pedestrian biological information (such as face) and vehicle identification (such as license plate) from the root, significantly reduces the risk of data leakage, and enhances privacy compliance.

[0089] The IR-UWB radar and the vehicle-mounted UWB positioning chip share a hardware platform, reducing the cost of additional devices. The radar signal processing algorithm has a lower complexity than the vision AI model, further reducing the overall cost of the system. Through distance dimension projection, a one-dimensional maximum GO-CFAR algorithm is selected for intrusion target detection, which significantly simplifies the algorithm complexity and maintains a low false alarm rate in complex parking environments. Through multi-time target detection cancellation, quasi-stationary objects (such as tree branches blown by the wind) are filtered out, further improving the accuracy of intrusion detection.

[0090] The present application overcomes the false alarm rate caused by harsh environments, prolongs the endurance time of the parking mode, and simplifies the complexity of the dynamic target detection algorithm. The scheme can be widely applied to intrusion detection in the parking sentinel mode, providing a more reliable and efficient solution for vehicle parking safety.

[0091] In a second aspect, the embodiments of the present application also provide a multi-radar collaborative detection device for a parking sentinel mode.

[0092] In an embodiment, with reference to Figure 5 , Figure 5 is a functional module schematic diagram of an embodiment of the multi-radar collaborative detection device for the parking sentinel mode of the present application. As Figure 5 shown, the multi-radar collaborative detection device for the parking sentinel mode comprises: A generation module for multiplexing a UWB positioning chip into an IR-UWB radar, controlling the IR-UWB radar to emit a UWB signal, and acquiring a channel impulse response to generate a first fast-slow time matrix; A first transformation module for removing the direct current component of the slow time dimension in the first fast-slow time matrix through fast Fourier transform to obtain a second fast-slow time matrix; A second transformation module for performing fast Fourier transform processing on the second fast-slow time matrix to obtain a range-Doppler matrix; A detection module for projecting the range-Doppler matrix on a range-amplitude plane to obtain a one-dimensional signal along the range dimension, and performing one-dimensional GO-CFAR detection on the one-dimensional signal along the range dimension to obtain a target detection result of whether there is an intrusion target; The canceling module is configured to cancel the target detection results at different times using an exclusive OR operation to obtain an external intrusion matrix.

[0093] Further, in an embodiment, the generating module is further configured to: multiplex the UWB positioning chip into an IR-UWB radar, control the IR-UWB radar to emit a UWB signal, and acquire a channel impulse response to generate a first fast-slow time matrix, including: record each frame of the channel impulse response acquired by the IR-UWB radar as a row vector:

[0094] wherein, is the i-th frame of channel impulse response data, and M is the number of sampling points of each frame of channel impulse response; generate the first fast-slow time matrix according to N frames of channel impulse response:

[0095] wherein, is the first fast-slow time matrix, and N is the total number of frames of the first fast-slow time matrix.

[0096] Further, in an embodiment, the first transforming module is further configured to: represent the first fast-slow time matrix along the sequence of the slow time dimension as:

[0097] wherein, , is the signal intensity of the j-th column in the first fast-slow time matrix at different time intervals n, is the element of the j-th column of the n+1-th row in the first fast-slow time matrix; perform a fast Fourier transform on each column signal in the first fast-slow time matrix to obtain a first frequency domain signal:

[0098] wherein, is the first frequency domain signal, and k is a frequency domain index; remove a direct current component in the first frequency domain signal, and perform an inverse fast Fourier transform on the first frequency domain signal after removing the direct current component to obtain a slow time dimension signal after removing the direct current component:

[0099]

[0100] wherein, a first frequency domain signal after removing a direct current component, a slow time dimension signal after removing a direct current component; recombine the slow time dimension signal after removing a direct current component to obtain the second fast-slow time matrix:

[0101] wherein, , the second fast-slow time matrix.

[0102] Further, in an embodiment, the second transformation module is further configured to: perform a fast Fourier transform on each column of the second fast-slow time matrix to obtain a corresponding second frequency domain signal, and express each second frequency domain signal as a column vector:

[0103] combine all column vectors to obtain the range-doppler matrix:

[0104] wherein, is a column vector, the range-doppler matrix.

[0105] Further, in an embodiment, the detection module is further configured to: take a maximum value of each column of the range-doppler matrix to obtain the one-dimensional signal along the range dimension:

[0106] wherein, V is the one-dimensional signal along the range dimension, represents a maximum value of each of the first 1 to M columns in the range-doppler matrix.

[0107] Further, in an embodiment, the detection module is further configured to: set algorithm parameters of one-dimensional GO-CFAR detection, including a cell window size as γ, a guard cell window size as α, and a threshold factor as λ; calculate a forward reference window mean value of the i-th range cell as:

[0108] calculate a backward reference window mean value of the i-th range cell as:

[0109] perform background clutter size estimation on the i-th detection cell, and select a larger one of the forward reference window mean value and the backward reference window mean value as a detection estimation result:

[0110] determine whether there is an intrusion target in the i-th detection unit according to a preset decision criterion:

[0111] wherein, represents that there is no intrusion target in the i-th detection unit, represents that there is an intrusion target in the i-th detection unit; record the decision result as:

[0112] combine the decision results of all detection units to obtain the target detection result:

[0113] wherein, is the target detection result, is the target detection result of the 1st to Mth detection units.

[0114] Further, in an embodiment, the cancellation module is further configured to: perform exclusive OR operation on the target detection results of each distance unit to obtain an external intrusion matrix:

[0115] wherein, is the external intrusion matrix, is the target detection result at t0, is the target detection result at t0+T interval.

[0116] Further, in an embodiment, the IR-UWB radar is arranged on the vehicle in an anchor point arrangement mode of four-anchor-point TDoA positioning.

[0117] The functions of each module in the multi-radar cooperative detection device in the parked sentry mode correspond to the steps in the multi-radar cooperative detection method in the parked sentry mode, and the functions and implementation processes will not be repeated here.

[0118] In a third aspect, the embodiments of the present application provide a multi-radar cooperative detection device in a parked sentry mode. The multi-radar cooperative detection device in the parked sentry mode can be a vehicle controller, a vehicle-mounted computer, or other devices with data processing functions.

[0119] Referring to Figure 6 , Figure 6 ​​​A hardware structure diagram of the multi-radar cooperative detection device in the parking sentry mode involved in the embodiment of the present application. In the embodiment of the present application, the multi-radar cooperative detection device in the parking sentry mode can include a processor, a memory, a communication interface, and a communication bus.

[0120] The communication bus can be any type for realizing the interconnection of the processor, the memory, and the communication interface.

[0121] The communication interface includes an input / output (I / O) interface, a physical interface, and a logical interface, and the like, which are interfaces for realizing the interconnection of devices inside the multi-radar cooperative detection device in the parking sentry mode, and interfaces for realizing the interconnection of the multi-radar cooperative detection device in the parking sentry mode and other devices (for example, other computing devices or user devices). The physical interface can be an Ethernet interface, a fiber interface, an ATM interface, and the like; the user device can be a display (Display), a keyboard (Keyboard), and the like.

[0122] The memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), and the like.

[0123] The processor can be a general-purpose processor, which can invoke the multi-radar cooperative detection program in the parking sentry mode stored in the memory and execute the multi-radar cooperative detection method in the parking sentry mode provided by the embodiment of the present application. For example, the general-purpose processor can be a central processing unit (CPU). The method executed when the multi-radar cooperative detection program in the parking sentry mode is invoked can refer to each embodiment of the multi-radar cooperative detection method in the parking sentry mode of the present application, which will not be described here.

[0124] Those skilled in the art can understand that the hardware structure shown in the figure does not constitute a limitation on the present application, and can include more or fewer components than the figure, or combine certain components, or different component arrangements.

[0125] It should be noted that the above sequence numbers of the embodiments of the present application are only for description, and do not represent the advantages and disadvantages of the embodiments.

[0126] The terms “include,” “comprise,” “have,” and any variations thereof, in the Specification and in the Claims of the present application, and the above-mentioned drawings, are intended to cover a non-exclusive inclusion. For example, a process, method, system, product, or device that includes a list of steps or units is not limited to the listed steps or units, but can optionally further include steps or units not listed, or can optionally further include other steps or units inherent to such processes, methods, products, or devices. The terms “first”, “second”, and “third” and the like descriptions are used to distinguish different objects, and do not represent the order or limit the types of “first”, “second”, and “third”.

[0127] In the description of the embodiments of the present application, “exemplary”, “for example”, or “for instance” is used to represent an example, illustration, or description. Any embodiment or design scheme described as “exemplary”, “for example”, or “for instance” in the embodiments of the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the words “exemplary”, “for example”, or “for instance” are intended to present the relevant concept in a specific manner.

[0128] In the description of the embodiments of the present application, unless otherwise specified, “ / ” represents the meaning of or, for example, A / B can represent A or B; “and / or” in the text only represents a description of the relationship between the associated objects, which means that there can be three relationships, for example, A and / or B can represent the three cases of A alone, A and B together, and B alone. In addition, in the description of the embodiments of the present application, “multiple” means two or more than two.

[0129] In some of the processes described in the embodiments of the present application, a plurality of operations or steps are included in a specific order, but it should be understood that these operations or steps can be executed or performed in parallel or in an order different from that in which they appear in the embodiments of the present application. The serial number of the operation is only used to distinguish different operations, and the serial number itself does not represent any execution order. In addition, these processes can include more or fewer operations, and these operations or steps can be executed in sequence or in parallel, and these operations or steps can be combined.

[0130] From the above description of the embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment method can be realized by means of software and a general hardware platform as required, of course, it can also be realized by hardware, but in many cases the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium (such as a ROM / RAM, a magnetic disk, an optical disk) as described above, and includes a plurality of instructions for causing a terminal device to execute the methods described in the embodiments of the present application.

[0131] The preferred embodiments of the present application have been described above with the illustrated embodiments, and are not intended to limit the scope of patent protection for the present application. Any equivalent structure or equivalent process variations, which directly or indirectly incorporate the contents of the specification and drawings of the present application, are also intended to be included within the scope of patent protection for the present application.

Claims

1. A multi-radar cooperative detection method for parking sentry mode, characterized in that: The multi-radar collaborative detection method of the parking sentry mode includes: Multiplexing the UWB positioning chip into an IR-UWB radar, controlling the IR-UWB radar to transmit a UWB signal, and obtaining a channel impulse response to generate a first fast-slow time matrix; Removing the DC component of the slow time dimension in the first fast-slow time matrix by fast Fourier transform to obtain a second fast-slow time matrix; Performing fast Fourier transform processing on the second fast and slow time matrix to obtain a range Doppler matrix; Projecting the range-Doppler matrix onto a range-amplitude plane to obtain a one-dimensional signal along the range dimension, performing one-dimensional GO-CFAR detection on the one-dimensional signal along the range dimension to obtain a target detection result of whether an intruder target exists; The target detection results at different times are canceled using XOR operation to obtain the external intrusion matrix.

2. The multi-radar cooperative detection method of the parking sentry mode according to claim 1, characterized in that: The UWB positioning chip is multiplexed into an IR-UWB radar, the IR-UWB radar is controlled to transmit a UWB signal, and a channel impulse response is obtained to generate a first fast-slow time matrix, including: The channel impulse response of each frame obtained by the IR-UWB radar is recorded in the form of a row vector: in, is the channel impulse response data of the i-th frame, M is the number of sampling points of the channel impulse response per frame; Generate the first fast and slow time matrix according to the N-frame channel impulse response: in, is the first fast-slow time matrix, and N is the total number of frames of the first fast-slow time matrix.

3. The multi-radar cooperative detection method of the parking sentry mode according to claim 2, characterized in that: Removing the DC component of the slow time dimension in the first fast-slow time matrix by fast Fourier transform to obtain a second fast-slow time matrix, including: The sequence of the first fast-slow time matrix along the slow time dimension is expressed as: in, , is the signal strength of the jth column in the first time fast-slow matrix at different time intervals n, is the element in the n+1th row and jth column of the first time fast-slow matrix; Perform a fast Fourier transform on each column signal in the first fast-slow time matrix to obtain a first frequency domain signal: in, is the first frequency domain signal, k is the frequency domain index; The DC component in the first frequency domain signal is removed, and an inverse fast Fourier transform is performed on the first frequency domain signal after the DC component is removed to obtain a slow time dimension signal after the DC component is removed: in, is the first frequency domain signal after removing the DC component, is the slow time dimension signal after removing DC component; The slow-time dimension signal after removing the DC component is recombined to obtain the second fast-slow time matrix: in, , is the second fast and slow time matrix.

4. The multi-radar cooperative detection method of the parking sentry mode according to claim 3, characterized in that: Performing fast Fourier transform processing on the second fast and slow time matrix to obtain a range Doppler matrix, including: Perform a fast Fourier transform on each column of the second fast-slow time matrix to obtain a corresponding second frequency domain signal, and represent each second frequency domain signal as a column vector: Combining all column vectors yields the range-Doppler matrix: in, is a column vector, is the range-Doppler matrix.

5. The multi-radar cooperative detection method of the parking sentry mode according to claim 4, characterized in that: Projecting the range-Doppler matrix onto a range-amplitude plane to obtain a one-dimensional signal along the range dimension includes: The maximum value of each column of the range-Doppler matrix is ​​taken to obtain the one-dimensional signal along the range dimension: Wherein, V is the one-dimensional signal along the distance dimension, Represents the maximum value of each column from the 1st to the Mth column in the range-Doppler matrix.

6. The multi-radar cooperative detection method of the parking sentry mode according to claim 5, characterized in that: Performing one-dimensional GO-CFAR detection on the one-dimensional signal along the distance dimension to obtain a target detection result of whether an intruder target exists includes: The algorithm parameters for one-dimensional GO-CFAR detection are set as follows: the test unit window size is γ, the protection unit window size is α, and the threshold factor is λ; Calculate the forward reference window mean of the i-th distance unit as: Calculate the backward reference window mean of the i-th distance unit: Estimate the size of background clutter for the i-th detection unit, and select the larger of the forward reference window mean and the backward reference window mean as the detection estimation result: According to the preset judgment criteria, Whether there is an intrusion target in each detection unit: in, Indicates that there is no intrusion target in the i-th detection unit, Indicates that there is an intrusion target in the i-th detection unit; The judgment result is recorded as: The target detection result is obtained by combining the judgment results of all detection units: in, is the target detection result, are the target detection results of the 1st to Mth detection units.

7. The multi-radar cooperative detection method of the parking sentry mode according to claim 6, characterized in that: Use XOR operation to cancel the target detection results at different times to obtain the external intrusion matrix, including: Perform an XOR operation on the target detection results of each distance unit to obtain the external intrusion matrix: in, is the external intrusion matrix, is the target detection result at time t0, is the interval from time t0 After the duration Target detection results at the moment.

8. The multi-radar cooperative detection method in registered sentinel mode according to claim 1, characterized in that: The IR-UWB radar is arranged on the vehicle in an anchor point arrangement manner of four-anchor point TDoA positioning.

9. A multi-radar cooperative detection device in parking sentry mode, characterized in that: The multi-radar cooperative detection device in the parking sentry mode includes: A generation module, which is used to multiplex the UWB positioning chip into an IR-UWB radar, control the IR-UWB radar to transmit a UWB signal, obtain a channel impulse response, and generate a first fast-slow time matrix; A first transformation module, configured to remove a DC component of a slow time dimension in the first fast-slow time matrix by fast Fourier transform to obtain a second fast-slow time matrix; a second transform module, configured to perform fast Fourier transform processing on the second fast-slow time matrix to obtain a range-Doppler matrix; a detection module configured to project the range-Doppler matrix onto a range-amplitude plane to obtain a one-dimensional signal along the range dimension, and perform one-dimensional GO-CFAR detection on the one-dimensional signal along the range dimension to obtain a target detection result of whether an intruder target is present; The cancellation module is used to cancel the target detection results at different times using an XOR operation to obtain an external intrusion matrix.

10. A multi-radar cooperative detection device in parking sentry mode, characterized in that: The multi-radar collaborative detection device of the parking sentry mode includes a processor, a memory, and a multi-radar collaborative detection program of the parking sentry mode stored on the memory and executable by the processor, wherein when the multi-radar collaborative detection program of the parking sentry mode is executed by the processor, the steps of the multi-radar collaborative detection method of the parking sentry mode as described in any one of claims 1 to 8 are implemented.

Images