A vehicle wireless non-contact sensing method and system based on channel state information

Abstract

The application provides a vehicle wireless non-contact sensing method and system based on channel state information, uses decimeter waves as a signal source for wireless sensing, is not affected by ambient light, has high non-contact sensing surrounding environment sensing precision, determines an obstacle through window calculation of a moving energy value, rejudges confidence of the obstacle to confirm a target object, realizes accurate positioning of the target object through construction of a boundary intersection point of mutually perpendicular Fresnel zones, is easy to realize and apply, effectively reduces misrecognition of automatic driving vehicle sensing, can realize wireless non-contact sensing and positioning, has high sensing precision, and has great application prospect.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of unmanned vehicle perception technology, and particularly relates to a vehicle wireless non-contact perception method and system based on channel state information. BACKGROUND

[0002] With the popularization of networked vehicles, the perception ability of unmanned vehicles is required to be higher and higher. In the vehicle perception technology in a short distance, the perception of ultrasonic radar is relatively mature, but has great limitations. It is difficult to reflect sound waves for small measured objects, and it is difficult to capture reflected waves for certain specific shapes. Temperature changes of more than 5-10 degrees will affect the sensing accuracy, and the stability and reliability are unstable. On the contrary, the use of decimeter wave signals to wirelessly perceive the surrounding environment can achieve millimeter-level perception accuracy, and can be used in both day and night. Combined with camera perception technology, it can achieve all-weather and full-area perception coverage at a low cost.

[0003] In the field of mobile computing, research has found that radio signals can not only be used for data transmission, but also can be used for environmental perception. The radio waves generated by the signal transmitter propagate through multiple paths such as direct reflection, scattering, etc., and the multipath superposition signals formed at the signal receiver carry information reflecting the characteristics of the environment. Using radio signals for gesture, action and motion state capture provides a new way and method for human behavior recognition, which has become a relatively mature application.

[0004] The perception method based on decimeter waves is less applied. The early related similar method is a non-contact perception based on WiFi. The precision and application range of this perception are very limited. Until the channel state information (CSI) of the 802.11n physical layer can be obtained from commercial equipment, with the help of its ability to characterize the signal multipath propagation and the fine amplitude and phase information reflected on each subcarrier, the perception application based on WiFi has rapidly developed. After that, a series of applications such as detecting human fall activities with coarse granularity, counting the number of people through walls, perceiving walking direction, and detecting gestures and breathing heartbeat with fine granularity have gradually emerged.

[0005] Manikanta Kotaru of the United States proposed to use the information provided by the 30 subcarriers of the WiFi card to increase the number of virtual antennas, thereby reducing the influence of the number of ordinary WiFi card antennas on signal discrimination and angle of arrival estimation, improving the angle of arrival estimation precision of direct path signals, and positioning the WiFi signal transmitting device. However, the method is used to position the position of the WiFi signal transmitting device, and cannot determine the position of a moving reflective WiFi signal object. Kiran Joshi of the United States proposed to establish a self-transmitting and self-receiving wireless local area network device, the signal emitted by the device is reflected and then received by the device, and the length, angle of arrival and signal strength of each reflection path are analyzed through post-processing. According to the activity of the positioning target, the path reflected by the positioning target is found out, and finally the position of the target is determined. However, this method needs to modify the existing WiFi device and change the traditional WiFi working mode, and cannot be realized on commercial WiFi devices. SUMMARY

[0006] In order to solve the problems that the current WiFi-based non-contact sensing technology cannot position the position of the target object and the sensing precision and application range are limited, the application provides a vehicle wireless non-contact sensing method based on channel state information, uses decimeter waves as a signal source for wireless sensing, is not affected by surrounding light, has high non-contact sensing precision for the surrounding environment, determines the confidence of the obstacle by calculating the moving energy value of each window to confirm the target object, and realizes accurate positioning of the target object by constructing the boundary intersection points of the mutually perpendicular Fresnel zones. The application also relates to a vehicle wireless non-contact sensing system based on channel state information.

[0007] The technical scheme of the application is as follows:

[0008] A vehicle wireless non-contact sensing method based on channel state information, characterized in that it comprises the following steps:

[0009] S1, arranging a decimeter wave signal transmitting end supporting measurement of channel state information and two decimeter wave signal receiving ends providing channel state information corresponding to the decimeter wave signal transmitting end on the top of the vehicle, receiving the environmental sensing data packet emitted by the decimeter wave signal transmitting end by the decimeter wave signal receiving end, and extracting the channel state information as the measured wireless sensing decimeter wave signal from the environmental sensing data packet;

[0010] S2, windowing the channel state information, calculating the moving energy value of each window, detecting the continuously appearing moving energy value, sensing the surrounding environment and predicting the possibility of collision, and judging that there is an obstacle around when the continuously appearing moving energy value exceeds the set obstacle threshold value;

[0011] S3, calculating the confidence of the obstacle existing around, and determining the target object sensed by wireless non-contact when the confidence exceeds a set confidence threshold;

[0012] S4, positioning the target object sensed by wireless non-contact: the measured wireless sensing decimeter wave signal obtained by the two decimeter wave signal receiving ends from the decimeter wave signal transmitting end contains a plurality of pairs of carriers, at least one pair of carriers in the plurality of pairs of carriers constructs M layers of Fresnel zones with the position of the decimeter wave signal transmitting end and the position of the first decimeter wave signal receiving end as two foci of an ellipse, at least another pair of carriers in the plurality of pairs of carriers constructs N layers of Fresnel zones with the position of the decimeter wave signal transmitting end and the position of the second decimeter wave signal receiving end as two foci of an ellipse, and the N layers of Fresnel zones are perpendicular to the M layers of Fresnel zones; according to the wireless sensing decimeter wave signal received by the first decimeter wave signal receiving end within a time period, in the M layers of Fresnel zones, the Fresnel phase difference and the Fresnel phase shift are used to determine the Fresnel zone level where the target object is located; according to the wireless sensing decimeter wave signal received by the second decimeter wave signal receiving end within a corresponding time period, in the N layers of Fresnel zones, the Fresnel phase difference and the Fresnel phase shift are used to determine the Fresnel zone level where the target object is located; the boundary intersection point of the Fresnel zone level where the target object is located in the M layers of Fresnel zones and the Fresnel zone level where the target object is located in the N layers of Fresnel zones is determined, and the position of the target object is determined according to the boundary intersection point.

[0013] Preferably, in the S1 step, after the channel state information is extracted, the Kalman filtering algorithm is used for filtering fusion and error supplement processing, and the processed channel state information is used as the measured wireless sensing signal; in the S2 step, the processed channel state information is windowed.

[0014] Preferably, in the S2 step, the moving energy value of each window is calculated as follows: the mean value of a long window signal is calculated first, and the channel state information is subtracted from the mean value to remove the direct current component in the CSI, then the windowed signal is converted from the time domain to the frequency domain through FFT transformation to obtain each coefficient of the FFT transformation result, and the moving energy is calculated according to each coefficient.

[0015] Preferably, the M layers of Fresnel zones and the N layers of Fresnel zones are a series of concentric ellipses-Fresnel zones formed around the receiving and transmitting ends, and the signal propagation path of the first layer of Fresnel zones is more than half a wavelength away from the corresponding receiving and transmitting end, the signal propagation path of the second layer of Fresnel zones is more than half a wavelength away from the corresponding receiving and transmitting end, and each layer of Fresnel zones is more than half a wavelength away from the previous layer.

[0016] Preferably, the two decimeter wave signal receiving ends are arranged on both sides of the vehicle head, the decimeter wave signal transmitting end is arranged at the vehicle tail, each receiving and transmitting end is provided with a corresponding antenna, a uniform linear antenna array is built with the antennas, and the interval between the antennas is not more than half the wavelength of the used decimeter wave, the environmental perception data packet emitted by the decimeter wave signal transmitting end is received by the decimeter wave signal receiving end with the uniform linear antenna array, and the channel state information is extracted therefrom as the measured wireless perception decimeter wave signal.

[0017] Preferably, the S3 step is to calculate the confidence and determine the specific category by using a neural network, the neural network is trained by a large amount of data with artificial annotation of channel state information in advance, the input of the neural network is the channel state information, and the output is the category of the obstacle and the confidence thereof.

[0018] A vehicle wireless non-contact perception system based on channel state information, characterized in that it comprises, which are connected in sequence, a decimeter wave wireless perception module, an obstacle judgment module, a confidence calculation module, and a Fresnel zone positioning module,

[0019] The decimeter wave wireless perception module comprises a decimeter wave signal transmitting end and two decimeter wave signal receiving ends corresponding to the decimeter wave signal transmitting end, which are arranged on the top of the vehicle and support the measurement of channel state information, the environmental perception data packet emitted by the decimeter wave signal transmitting end is received by the decimeter wave signal receiving end, and the channel state information is extracted therefrom as the measured wireless perception decimeter wave signal;

[0020] The obstacle judgment module divides the channel state information into windows, calculates the moving energy value of each window, detects the continuously appearing moving energy value, perceives the surrounding environment, and predicts the possibility of collision, and judges that there is an obstacle around when the continuously appearing moving energy value exceeds the set obstacle threshold value;

[0021] The confidence calculation module calculates the confidence of the obstacle existing around, and determines the target object perceived by the wireless non-contact perception when the confidence exceeds the set confidence threshold value;

[0022] The Fresnel zone positioning module positions the target object sensed by the wireless non-contact: two decimeter wave signal receiving ends simultaneously obtain the measured wireless sensing decimeter wave signal from the decimeter wave signal transmitting end, the wireless sensing decimeter wave signal contains a plurality of pairs of carriers, at least one pair of carriers in the plurality of pairs of carriers constructs M layers of Fresnel zones with the position of the decimeter wave signal transmitting end and the position of the first decimeter wave signal receiving end as two foci of an ellipse, at least another pair of carriers in the plurality of pairs of carriers constructs N layers of Fresnel zones with the position of the decimeter wave signal transmitting end and the position of the second decimeter wave signal receiving end as two foci of an ellipse, and the N layers of Fresnel zones are perpendicular to the M layers of Fresnel zones, according to the wireless sensing decimeter wave signal received by the first decimeter wave signal receiving end in a time period, in the M layers of Fresnel zones, the Fresnel phase difference and the Fresnel phase offset are used to determine the Fresnel zone level where the target object is located, according to the wireless sensing decimeter wave signal received by the second decimeter wave signal receiving end in a corresponding time period, in the N layers of Fresnel zones, the Fresnel phase difference and the Fresnel phase offset are used to determine the Fresnel zone level where the target object is located, the boundary intersection point of the Fresnel zone level where the target object is located in the M layers of Fresnel zones and the Fresnel zone level where the target object is located in the N layers of Fresnel zones is determined, and the position of the target object is determined according to the boundary intersection point.

[0023] Preferably, the decimeter wave wireless sensing module performs filtering fusion and error supplement processing on the extracted channel state information through a Kalman filtering algorithm, and the processed channel state information is used as the measured wireless sensing signal; and the obstacle judgment module divides the processed channel state information into windows.

[0024] Preferably, the obstacle judgment module calculates the moving energy value of each window as follows: the mean value of a long window signal is calculated first, and the channel state information is subtracted from the mean value to remove the direct current component in the CSI, then the signal after windowing is converted from the time domain to the frequency domain through FFT transformation to obtain each coefficient of the FFT transformation result, and the moving energy is calculated according to each coefficient.

[0025] Preferably, the confidence calculation module calculates the confidence and determines the specific type by using a neural network, the neural network is trained by a large amount of channel state information data with artificial annotation, the input of the neural network is the channel state information, and the output is the type of the obstacle and the confidence thereof.

[0026] The present application has the following advantages:

[0027] The application provides a vehicle wireless non-contact sensing method based on channel state information, uses decimeter waves as a signal source for wireless sensing, has high non-contact sensing surrounding environment sensing precision, can reach millimeter level, greatly improves close-range sensing of unmanned driving, compared with traditional millimeter wave radars, is not suitable for close-range sensing, and has limited sensing range; the decimeter wave signal sensing has no dead angle and can achieve 360-degree space dead angle-free sensing; compared with laser radar sensors, has lower cost; compared with camera sensors, can achieve all-time domain sensing in day and night and is not affected by surrounding light changes. The decimeter wave sensing method can be used in day and night, can achieve all-weather and all-domain sensing coverage with lower cost than the camera; the sensing precision is adjustable, sensing precision adjustment under different sensing requirements can be realized, and the method is suitable for different sensing task requirements. The decimeter wave sensing limit is improved to centimeter level, while the penetration ability of the decimeter wave is maintained, and interference such as leaves and plastic bags is filtered. The filtering ability refers to that electromagnetic waves with a wavelength of 43 cm can penetrate road-side plastic bags, leaves and other obstacles which have little influence on vehicle driving; when the electromagnetic waves penetrate the obstacles, the signal received by the decimeter wave signal receiving end is not affected, so the signal is not regarded as an obstacle, which better solves the problem that laser radars and millimeter wave radars of automatic driving vehicles detect the area of leaves and plastic bags as an untravelable area.The application applies decimeter wave sensing technology to the sensing of unmanned vehicles, including the state information of people, vehicles and roads, which breaks through the traditional video and image monitoring detection technology, does not depend on the environment and the requirement of correctly setting the equipment, provides an efficient and low-cost driving behavior real-time sensing method, extracts channel state information (CSI) in the decimeter wave signal for windowing, calculates the moving energy value of each window, detects the continuously appearing moving energy value, senses the surrounding environment and predicts the possibility of collision, judges that there is an obstacle around when the continuously appearing moving energy value exceeds the set obstacle threshold, the obstacle may be different types such as people, vehicles and objects, then the confidence can be calculated by using the neural network to determine the specific type, and then determine the target object sensed by wireless non-contact sensing, combine with a branch decimeter wave signal transmitting end and two decimeter wave signal receiving ends arranged on the top of the vehicle, construct specific mutually perpendicular and intersecting Fresnel zones, and cover the surrounding space, based on the Fresnel phase difference and the Fresnel phase shift and using the Fresnel zone positioning technology, the precise positioning of the target object is realized through the boundary intersection point of the levels of the two mutually perpendicular Fresnel zones where the target object is located, the problems that the existing technology cannot position the target object position and the sensing accuracy and application range are limited are solved, the sensing method of the application can reflect the motion change of the surrounding objects at any time, and the change of the surrounding environment is obtained in real time, the sensing of the environment provides a basis for the decision planning of the autonomous vehicle, and the sensing method is not afraid of environmental changes and is not affected by the surrounding light, which is an all-weather vehicle sensing method, the method is easy to implement and apply, and effectively reduces the misidentification of autonomous vehicle sensing, can realize wireless non-contact identification and positioning, has high sensing accuracy, and has great application prospect.

[0028] The application also relates to a vehicle wireless non-contact sensing system based on channel state information, which corresponds to the vehicle wireless non-contact sensing method based on channel state information, and can be understood as an implementation system of the vehicle wireless non-contact sensing method based on channel state information, and comprises a decimeter wave wireless sensing module, an obstacle judgment module, a confidence degree calculation module and a Fresnel zone positioning module connected in sequence, each module works cooperatively, decimeter waves are used as a signal source for wireless sensing, the system is not affected by ambient light, the non-contact sensing of the surrounding environment has high sensing precision, then the CSI signal measured at the receiving end is divided into windows, the moving energy value of each window is calculated, the continuously appearing moving energy values are detected, the confidence degree of the obstacle is judged to confirm the target object, and the boundary intersection points of the mutually perpendicular Fresnel zones are constructed to realize accurate positioning of the target object, the application applies the decimeter wave sensing technology to the sensing of the unmanned vehicle, includes the state information of people, vehicles and roads, the overall structure of the system is designed ingeniously, and the system is easy to realize and use, and the wireless non-contact sensing precision of the vehicle is greatly improved. BRIEF DESCRIPTION OF DRAWINGS

[0029] Figure 1 is a flowchart of the vehicle wireless non-contact sensing method based on channel state information.

[0030] Figure 2 is a schematic diagram of the arrangement of the transmitting and receiving ends in the vehicle wireless non-contact sensing method based on channel state information.

[0031] Figure 3 is a schematic diagram of the construction of the mutually perpendicular Fresnel zones in the vehicle wireless non-contact sensing method based on channel state information.

[0032] Figure 4 is a flowchart of the positioning of the target object in the vehicle wireless non-contact sensing method based on channel state information. DETAILED DESCRIPTION

[0033] The application will be described below in combination with the drawings.

[0034] A vehicle wireless non-contact sensing method based on channel state information, also known as a global sensing method for unmanned vehicles using decimeter wave intelligent sensing. Decimeter wave is an electromagnetic wave with a frequency of 700 MHz and a wavelength of 43 cm. Its characteristics are based on electromagnetic wave sensing of the environment. Decimeter wave radiation can penetrate materials such as ceramics and plastics with very little attenuation. The time domain spectrum signal-to-noise ratio of decimeter wave is high, signal loss is less, and sensing accuracy is in the millimeter level. Based on CSI business filtering noise, the cost is lower. It is a new unmanned sensing method. Decimeter wave is used as the signal source for wireless sensing and is not affected by the surrounding light. The dynamic object to be positioned does not need to carry any equipment. The decimeter wave transceiver environment is used to realize non-contact sensing of the target to be sensed. The non-contact sensing of the surrounding environment has high sensing accuracy. The moving energy value is determined by windowing calculation to determine the confidence of the obstacle to confirm the target object. The position of the target object to be positioned is obtained by constructing the boundary intersection of the mutually perpendicular Fresnel zones, and the accurate positioning of the target object is realized. Figure 1 As shown in the following steps:

[0035] S1, arranging a decimeter wave signal transmitting end supporting measurement of channel state information and two decimeter wave signal receiving ends providing channel state information corresponding to the decimeter wave signal transmitting end on the top of the vehicle. The decimeter wave signal receiving end receives the environmental sensing data packet emitted by the decimeter wave signal transmitting end and extracts the channel state information CSI as the measured wireless sensing decimeter wave signal.

[0036] The basic principle of wireless sensing based on decimeter wave: the transmission of electromagnetic wave signal is constrained by physical space, resulting in the signal reaching the receiving end via multiple paths. On the one hand, the physical space constrains the propagation of electromagnetic waves, and on the other hand, the electromagnetic waves reaching the receiving end also record the characteristics of the physical space they pass through. When a person is in the physical space, additional paths will be introduced due to the reflection and diffraction of the signal by the person. Therefore, the influence of the behavior of the obstacle on the propagation of the electromagnetic wave signal will be characterized by the electromagnetic wave signal reaching the receiving end. By establishing a mapping relationship between the changes in these signals and the different behaviors of the person, the basic idea of human behavior sensing based on decimeter wave signal is established. The present application provides a decimeter wave sensing method based on a decimeter wave signal transmitting end. The dynamic object to be positioned does not need to carry any equipment. On the basis of not modifying the hardware, through two or more decimeter wave signal receiving ends, when there is an object in the environment, the object will be sensed. The arrangement of the decimeter wave signal transmitting end (referred to as the transmitting end) and the decimeter wave signal receiving end (referred to as the receiving end) is as shown in Figure 2 The specific arrangement diagram is as shown in Figure 3As shown, the embodiment includes one transmitting end and two receiving ends, three devices are arranged on the roof of the vehicle, two receiving ends are arranged on the two sides of the vehicle head respectively, and one transmitting end is arranged at the tail of the vehicle. The transmitting end is a decimeter wave signal transmitting device, the receiving end is a corresponding decimeter wave signal receiver of the transmitting end, each transmitting and receiving end is provided with a corresponding antenna, a uniform linear antenna array is built with the antennas, and the interval between the antennas is not more than half the wavelength of the used decimeter wave. The decimeter wave signal receiving end receives the environmental perception data packet emitted by the decimeter wave signal transmitting end by using the uniform linear antenna array, and extracts the channel state information from the environmental perception data packet as the measured wireless perception decimeter wave signal. The interval of the antenna array should be determined by the signal wavelength used by the specific decimeter wave signal transmitting and receiving device. If other frequency decimeter wave signals are used, the antenna spacing of the antenna array should also be adjusted. The receiving end receives the packet emitted by the transmitting end by using the antenna array, and measures the CSI information (which reflects the changes of amplitude and phase of the signal from the transmitting end to the receiving end) from the packet; the position of each receiving end is measured before the system runs.

[0037] Further, the application can filter noise by using the CSI quotient, more completely receive the message of the signal, improve the signal-to-noise ratio by using the CSI quotient model, and construct an orthogonal base signal. The amplitude and phase noise of the commercial decimeter wave device can be eliminated, the signal-to-noise ratio can be improved, the perception range can be improved, a new orthogonal base signal can be constructed, the accurate perception of small activities can be supported, the tracking accuracy can be improved to the decimeter level, and the complementary relationship between the antenna array and multiple CSI quotients is utilized, so that the application has great application prospects. Further, after the channel state information is extracted, the Kalman filtering algorithm can be used for filtering fusion and error supplement processing, and the processed CSI information is used as the measured wireless perception signal.

[0038] S2, the CSI signal measured after removing noise and filtering fusion and error supplement processing is divided into windows, the moving energy value of each window is calculated, the continuously appearing moving energy values are detected, the surrounding environment is perceived, and the possibility of collision is predicted, and it is judged that there is an obstacle around when the continuously appearing moving energy values exceed the set obstacle threshold.

[0039] Further, the calculation method of the moving energy value of each window is as follows: first, the mean value of a long window signal is calculated, and the CSI is subtracted from the mean value, so as to remove the direct current component in the CSI; then, the signal after the windowing is converted from time domain to frequency domain by FFT (Fast Fourier Transform), and each coefficient of the FFT result is obtained; finally, the moving energy is calculated according to each coefficient of the FFT result. Wherein m FFT is the FFT coefficient calculation value through the time window, and windowlength / 2 is the half wavelength; when the continuously appearing moving energy E exceeds the set threshold, it is judged that there is an obstacle.

[0040] S3. Calculate the confidence level of the surrounding obstacles. If the confidence level exceeds the set confidence level threshold, it is determined to be a target object sensed wirelessly without contact.

[0041] Preferably, a neural network can be used to calculate the confidence level and determine the specific type. The neural network is trained using a large amount of pre-collected, manually labeled CSI data. The input to the neural network is CSI information, and the output is the type of obstacle and its confidence level. In other words, step S2 above determines that the existing obstacle may be of different types such as people, vehicles, or objects. Then, this step can use the neural network to calculate the confidence level to determine the specific type, thereby identifying it as a target object detected wirelessly without contact.

[0042] S4. Locating the target object sensed wirelessly without contact: Two decimeter wave signal receivers simultaneously acquire the measured wireless sensing decimeter wave signals from the decimeter wave signal transmitters. These signals contain several pairs of carriers. At least one pair of carriers, with the positions of the decimeter wave signal transmitter and the first decimeter wave signal receiver as the two foci of an ellipse, constructs an M-layer Fresnel zone. At least another pair of carriers, with the positions of the decimeter wave signal transmitter and the second decimeter wave signal receiver as the two foci of an ellipse, constructs an N-layer Fresnel zone, where the N-layer Fresnel zone is perpendicular to the M-layer Fresnel zone. Based on the first decimeter wave signal received... The receiving end receives the wireless sensing decimeter wave signal within a certain time period. Within the M-layer Fresnel zone, based on the Fresnel phase difference and Fresnel phase shift, the Fresnel zone level of the target object is determined. Based on the wireless sensing decimeter wave signal received by the second decimeter wave signal receiving end within a corresponding time period, within the N-layer Fresnel zone, based on the Fresnel phase difference and Fresnel phase shift, the Fresnel zone level of the target object is determined. The boundary intersection point between the Fresnel zone level of the target object in the M-layer Fresnel zone and the Fresnel zone level of the target object in the N-layer Fresnel zone is determined, and the position of the target object is determined based on the boundary intersection point.

[0043] Specifically, the first decimeter wave signal receiver is... Figure 3 The receiver R1 shown is also called the first receiving device R1, and the second decimeter wave signal receiver is... Figure 3 The receiver R2 shown is also called the second receiving device R2, and the decimeter wave signal transmitter is... Figure 3 The transmitter T shown can also be called the transmitting device T.

[0044] Step S4 achieves precise positioning of the target object by establishing the boundary intersections of the Fresnel zone hierarchy within mutually perpendicular Fresnel zones. Figure 4The positioning procedure is shown as follows: first, the first receiving device R1 and the second receiving device R2 simultaneously receive wireless perception decimeter wave signals from the sending device T, and the wireless perception decimeter wave signals contain two or more carriers. Second, at least one pair of carriers A in the two or more carriers, with the position of the sending device T and the position of the first receiving device R1 as the foci of the ellipse, a plurality of M-layer Fresnel zones of the carriers are constructed. Next, for at least one pair of carriers B in the two or more carriers, with the position of the sending device T and the position of the second receiving device R2 as the foci of the ellipse, a plurality of N-layer Fresnel zones of the carriers are constructed. As shown in Figure 3 As shown, the M-layer Fresnel zone and the N-layer Fresnel zone are a series of concentric ellipses-Fresnel zones formed around the transceiver, and the N-layer Fresnel zone is perpendicular to the M-layer Fresnel zone. The innermost ellipse is defined as the first Fresnel zone (i.e., the first layer Fresnel zone), the ellipse ring between the first and second layers is defined as the second Fresnel zone (i.e., the second layer Fresnel zone), and the M(th) layer Fresnel zone (or the N(th) layer Fresnel zone) is defined in the same way, and each layer of ellipse is defined as the boundary of the corresponding Fresnel zone. The first layer Fresnel zone signal propagation path is more than half a wavelength away from the corresponding transceiver, the second layer Fresnel zone signal propagation path is more than half a wavelength away from the corresponding transceiver, each layer of Fresnel zone is more than half a wavelength away from the previous layer, and the space is filled with such Fresnel zones.

[0045] Some scholars have proposed a Fresnel zone model sensing theory for wireless non-contact sensing, which is the theoretical basis for decimeter wave wireless non-contact sensing. The Fresnel zone refers to a series of concentric ellipses with the two points of the transceiver as the foci. Due to the different paths and lengths of electromagnetic wave propagation paths, the electromagnetic waves propagating to the first Fresnel zone are in phase with the line-of-sight (LoS) propagation path, resulting in an enhanced electromagnetic wave signal at the observation point. The electromagnetic waves propagating to the second Fresnel zone are out of phase with the LoS, resulting in a weakened electromagnetic wave signal at the observation point. Subsequently, with the odd-even alternation of the Fresnel zone, the enhanced and weakened interference superposition results are obtained at the observation point. If reflection and frequency diversity are considered, a Fresnel zone reflection model is constructed, and the precise relationship between the small movement of the object and the signal fluctuation pattern is revealed. Based on this, the object motion behavior involving only sub-wavelength level small displacement can be captured, the limit of decimeter wave sensing is improved to centimeter level, and the penetration ability of decimeter wave is maintained, filtering out interference such as leaves and plastic bags.

[0046] The filtering capability refers to that the electromagnetic wave with a wavelength of 43 cm can penetrate the roadside plastic bags, leaves and other obstacles that have little influence on vehicle driving. When the electromagnetic wave penetrates the obstacles, the signal received by the decimeter wave signal receiving end is not affected, and therefore it will not be regarded as an obstacle. This better solves the problem that the automatic driving laser radar and millimeter wave radar will detect the area of leaves and plastic bags as an untravelable area.

[0047] Next, according to the wireless perception decimeter wave signals received by the first receiving device R1 within a time period, in the M-layer Fresnel zone, the Fresnel zone F1 in which the target object is located is determined according to the theoretical Fresnel phase difference and the Fresnel phase shift. Then, according to the wireless perception decimeter wave signals received by the second receiving device R2 within a corresponding time, in the N-layer Fresnel zone, the Fresnel zone F2 in which the target object is located is determined according to the theoretical Fresnel phase difference and the Fresnel phase shift. The boundary intersection point of the Fresnel zones F1 and F2 is determined. Finally, the position of the target is determined according to the boundary intersection point of the Fresnel zones F1 and F2. That is, the region in which the obstacle (target object) is located is determined according to the waveform of the echo after the decimeter wave hits the obstacle, and therefore it is the first point that hits in the decimeter wave transmission process, as shown in the figure. In the M-layer Fresnel zone, according to the theoretical Fresnel phase difference and the Fresnel phase shift, it is determined that the target object is located in the eighth Fresnel zone F1 whose Fresnel zone level is the eighth layer; in the N-layer Fresnel zone, according to the theoretical Fresnel phase difference and the Fresnel phase shift, it is determined that the target object is located in the seventh Fresnel zone F2 whose Fresnel zone level is the seventh layer, and then the boundary of the two Fresnel zone levels (the eighth Fresnel zone F1 and the seventh Fresnel zone F2) is used to determine the boundary intersection point, realizing the positioning requirement of the target object. Figure 3

[0048] ​The application also relates to a vehicle wireless non-contact sensing system based on channel state information, which corresponds to the vehicle wireless non-contact sensing method based on channel state information, and can be understood as an implementation system of the vehicle wireless non-contact sensing method based on channel state information, and comprises a decimeter wave wireless sensing module, an obstacle judgment module, a confidence degree calculation module and a Fresnel zone positioning module connected in sequence, and each module works cooperatively, wherein the decimeter wave wireless sensing module is arranged on the top of a vehicle and comprises a decimeter wave signal transmitting end for measuring channel state information and two decimeter wave signal receiving ends corresponding to the decimeter wave signal transmitting end and providing channel state information; the environment sensing data packet emitted by the decimeter wave signal transmitting end is received by the decimeter wave signal receiving end, and the channel state information is extracted from the environment sensing data packet as the measured wireless sensing decimeter wave signal; the obstacle judgment module divides the channel state information into windows, calculates the moving energy value of each window, detects the continuously appearing moving energy value, senses the surrounding environment and predicts the possibility of collision, and judges that there is an obstacle around when the continuously appearing moving energy value exceeds the set obstacle threshold value; the confidence degree calculation module calculates the confidence degree of the obstacle existing around, and determines that the target object sensed by the wireless non-contact sensing is the target object when the confidence degree exceeds the set confidence degree threshold value; and the Fresnel zone positioning module positions the target object sensed by the wireless non-contact sensing: the two decimeter wave signal receiving ends simultaneously obtain the measured wireless sensing decimeter wave signal containing a plurality of pairs of carriers from the decimeter wave signal transmitting end; at least one pair of carriers in the plurality of pairs of carriers takes the position of the decimeter wave signal transmitting end and the position of the first decimeter wave signal receiving end as the two foci of an ellipse to construct M layers of Fresnel zones; at least another pair of carriers in the plurality of pairs of carriers takes the position of the decimeter wave signal transmitting end and the position of the second decimeter wave signal receiving end as the two foci of an ellipse to construct N layers of Fresnel zones, and the N layers of Fresnel zones are perpendicular to the M layers of Fresnel zones; according to the wireless sensing decimeter wave signal received by the first decimeter wave signal receiving end within a time period, the Fresnel zone level where the target object is located is determined in the M layers of Fresnel zones according to the Fresnel phase difference and the Fresnel phase shift; according to the wireless sensing decimeter wave signal received by the second decimeter wave signal receiving end within a corresponding time period, the Fresnel zone level where the target object is located is determined in the N layers of Fresnel zones according to the Fresnel phase difference and the Fresnel phase shift; the boundary intersection point of the Fresnel zone level where the target object is located in the M layers of Fresnel zones and the Fresnel zone level where the target object is located in the N layers of Fresnel zones is determined, and the position of the target object is determined according to the boundary intersection point. Further, the Fresnel zone positioning module can send the determined position information of the target object to a decision module, and the decision module makes a corresponding response to ensure that the autonomous vehicle is safe and strategic.

[0049] Preferably, the decimeter wave wireless sensing module filters and fuses and error supplements the channel state information through Kalman filtering algorithm after extracting the channel state information, and the processed channel state information is taken as the measured wireless sensing signal; the obstacle judgment module divides the processed channel state information into windows.

[0050] Preferably, the obstacle judgment module calculates the moving energy value of each window, specifically: first, the mean value of a long window signal is calculated, and the channel state information is subtracted from the mean value to remove the direct current component in the CSI, then the signal after windowing is converted from time domain to frequency domain through FFT transformation to obtain each coefficient of the FFT transformation result, and then the moving energy is calculated according to each coefficient.

[0051] Preferably, the confidence calculation module calculates the confidence and determines the specific type by using a neural network, and the neural network is trained by a large amount of channel state information data with artificial annotation collected in advance, the input of the neural network is the channel state information, and the output is the type of the obstacle and the confidence thereof.

[0052] It should be noted that the above specific embodiments can enable those skilled in the art to more fully understand the present invention, but do not limit the present invention in any way. Therefore, although the present invention has been described in detail with reference to the drawings and examples, those skilled in the art should understand that the present invention can still be modified or replaced by equivalents, in short, all technical solutions and improvements that do not deviate from the spirit and scope of the present invention should be covered in the protection scope of the patent of the present invention.

Claims

1. A vehicle wireless non-contact sensing method based on channel state information, characterized by, The method comprises the following steps: S1, arranging a decimeter wave signal transmitting end for supporting measurement of channel state information and two decimeter wave signal receiving ends for providing channel state information corresponding to the decimeter wave signal transmitting end on the top of a vehicle, receiving environmental sensing data packets emitted by the decimeter wave signal transmitting end through the decimeter wave signal receiving end, and extracting channel state information from the environmental sensing data packets as measured wireless sensing decimeter wave signals; S2, dividing the channel state information into windows, calculating a moving energy value of each window, detecting continuously appearing moving energy values, sensing the surrounding environment, and predicting the possibility of collision, and judging that there is an obstacle around when the continuously appearing moving energy values exceed a set obstacle threshold; S3, calculating the confidence degree of the obstacle existing around, and determining a target object sensed by wireless non-contact sensing when the confidence degree exceeds a set confidence degree threshold; S4, positioning the target object sensed by wireless non-contact sensing: the two decimeter wave signal receiving ends simultaneously obtain measured wireless sensing decimeter wave signals containing a plurality of pairs of carriers from the decimeter wave signal transmitting end, at least one pair of carriers of the plurality of pairs of carriers constructs M layers of Fresnel zones with the position of the decimeter wave signal transmitting end and the position of the first decimeter wave signal receiving end as two foci of an ellipse, at least another pair of carriers of the plurality of pairs of carriers constructs N layers of Fresnel zones with the position of the decimeter wave signal transmitting end and the position of the second decimeter wave signal receiving end as two foci of an ellipse, and the N layers of Fresnel zones are perpendicular to the M layers of Fresnel zones; according to the wireless sensing decimeter wave signals received by the first decimeter wave signal receiving end within a time period, the Fresnel zone level in which the target object is located is determined in the M layers of Fresnel zones according to a Fresnel phase difference and a Fresnel phase shift; according to the wireless sensing decimeter wave signals received by the second decimeter wave signal receiving end within a corresponding time period, the Fresnel zone level in which the target object is located is determined in the N layers of Fresnel zones according to the Fresnel phase difference and the Fresnel phase shift; the boundary intersection point of the Fresnel zone level in which the target object is located in the M layers of Fresnel zones and the Fresnel zone level in which the target object is located in the N layers of Fresnel zones is determined, and the position of the target object is determined according to the boundary intersection point; In the S1 step, the channel state information is filtered and fused and error supplement processing is performed through a Kalman filtering algorithm after the channel state information is extracted, and the processed channel state information is used as the measured wireless sensing signal; in the S2 step, the processed channel state information is divided into windows; In the S2 step, the moving energy value of each window is calculated as follows: the mean value of a long window signal is calculated first, and the channel state information is subtracted from the mean value to remove the direct current component in the CSI, then the signal after windowing is converted from the time domain to the frequency domain through FFT transformation to obtain each coefficient of the FFT transformation result, and the moving energy is calculated according to each coefficient; In the S3 step, the confidence degree and the specific type are calculated by using a neural network, the neural network is trained through a large amount of channel state information data with artificial annotation collected in advance, the input of the neural network is the channel state information, and the output is the type of the obstacle and the confidence degree thereof. The M-layer Fresnel zone and the N-layer Fresnel zone are a series of concentric elliptical-Fresnel zones formed around the transmitting and receiving ends, and the signal propagation path of the first-layer Fresnel zone is more than half a wavelength away from the corresponding transmitting and receiving end, the signal propagation path of the second-layer Fresnel zone is more than half a wavelength away from the corresponding transmitting and receiving end, and each layer of Fresnel zone is more than half a wavelength away from the previous layer. The S4 step is to achieve accurate positioning of the target object by locating the target object at the intersection of the boundaries of the Fresnel zone layers in the Fresnel zone perpendicular to each other. The positioning process is as follows: first, the first receiving device R1 and the second receiving device R2 simultaneously receive wireless sensing decimeter wave signals from the transmitting device T, and the wireless sensing decimeter wave signals contain two or more carriers; second, at least one pair of carriers A in the two or more carriers, with the position of the transmitting device T and the position of the first receiving device R1 as the foci of the ellipse, an M-layer Fresnel zone of multiple carriers is constructed; Next, for at least one pair of carriers B in the two or more carriers, with the position of the transmitting device T and the position of the second receiving device R2 as the foci of the ellipse, an N-layer Fresnel zone of multiple carriers is constructed, and the M-layer Fresnel zone and the N-layer Fresnel zone are a series of concentric elliptical-Fresnel zones formed around the transmitting and receiving ends, and the N-layer Fresnel zone is perpendicular to the M-layer Fresnel zone; define the innermost ellipse as the first Fresnel zone (i.e. the first-layer Fresnel zone), the ellipse ring between the first and second layers is defined as the second Fresnel zone (i.e. the second-layer Fresnel zone), and so on to define the M(th) Fresnel zone (or the N(th) Fresnel zone), and each layer of ellipse is defined as the boundary of the corresponding Fresnel zone; the signal propagation path of the first-layer Fresnel zone is more than half a wavelength away from the corresponding transmitting and receiving end, the signal propagation path of the second-layer Fresnel zone is more than half a wavelength away from the corresponding transmitting and receiving end, and each layer of Fresnel zone is more than half a wavelength away from the previous layer, and the space is filled with such Fresnel zones.

2. The vehicle wireless non-contact sensing method according to claim 1, characterized by, Two decimeter wave signal receiving ends are arranged on both sides of the vehicle head, and a decimeter wave signal transmitting end is arranged at the vehicle tail. Each transmitting and receiving end is provided with a corresponding antenna, and a uniform linear antenna array is built with the antennas, and the interval between the antennas does not exceed half a wavelength of the used decimeter wave. The decimeter wave signal receiving end receives the environmental perception data packet emitted by the decimeter wave signal transmitting end and extracts the channel state information therefrom as the measured wireless sensing decimeter wave signal.

3. A vehicle wireless non-contact sensing system based on channel state information, characterized by, It comprises, in sequence, a decimeter wave wireless sensing module, an obstacle judgment module, a confidence calculation module, and a Fresnel zone positioning module, The decimeter wave wireless sensing module arranges a decimeter wave signal transmitting end supporting measurement of channel state information on the top of the vehicle, and two decimeter wave signal receiving ends corresponding to the decimeter wave signal transmitting end for providing channel state information. The decimeter wave signal receiving end receives the environmental perception data packet emitted by the decimeter wave signal transmitting end and extracts the channel state information therefrom as the measured wireless sensing decimeter wave signal. The obstacle judgment module divides the channel state information into windows, calculates the moving energy value of each window, detects the continuously appearing moving energy value, senses the surrounding environment and predicts the possibility of collision, and judges that there is an obstacle around when the continuously appearing moving energy value exceeds a set obstacle threshold value. The confidence calculation module calculates the confidence of the obstacle existing around, and determines that the target object sensed by wireless non-contact sensing is when the confidence exceeds a set confidence threshold value. The Fresnel zone positioning module positions the target object sensed by wireless non-contact sensing: two decimeter wave signal receiving ends simultaneously obtain the measured wireless sensing decimeter wave signal from the decimeter wave signal transmitting end, the wireless sensing decimeter wave signal contains a plurality of pairs of carriers, at least one pair of carriers in the plurality of pairs of carriers takes the position of the decimeter wave signal transmitting end and the position of the first decimeter wave signal receiving end as the two foci of an ellipse to construct M layers of Fresnel zones, and at least another pair of carriers in the plurality of pairs of carriers takes the position of the decimeter wave signal transmitting end and the position of the second decimeter wave signal receiving end as the two foci of an ellipse to construct N layers of Fresnel zones, and the N layers of Fresnel zones are perpendicular to the M layers of Fresnel zones; according to the wireless sensing decimeter wave signal received by the first decimeter wave signal receiving end within a time period, the Fresnel zone level in which the target object is located is determined according to the Fresnel phase difference and the Fresnel phase shift in the M layers of Fresnel zones; according to the wireless sensing decimeter wave signal received by the second decimeter wave signal receiving end within a corresponding time period, the Fresnel zone level in which the target object is located is determined according to the Fresnel phase difference and the Fresnel phase shift in the N layers of Fresnel zones; The boundary intersection point of the Fresnel zone level in which the target object is located in the M layers of Fresnel zones and the Fresnel zone level in which the target object is located in the N layers of Fresnel zones is determined, and the position of the target object is determined according to the boundary intersection point.

4. The wireless non-contact sensing system of claim 3, wherein The decimeter wave wireless sensing module performs filtering fusion and error supplement processing on the extracted channel state information through the Kalman filtering algorithm, and the processed channel state information is used as the measured wireless sensing signal; the obstacle judgment module divides the processed channel state information into windows.

5. The wireless non-contact sensing system of claim 3, wherein The obstacle judgment module calculates the moving energy value of each window: first, the mean value of a long window signal is calculated, and the channel state information is subtracted from the mean value to remove the direct current component in the CSI, then the divided signal is converted from the time domain to the frequency domain through FFT transformation to obtain each coefficient of the FFT transformation result, and the moving energy is calculated according to each coefficient.

6. The vehicle wireless non-contact sensing system according to one of claims 3 to 5, characterized in that The confidence calculation module uses a neural network to calculate the confidence and determine the specific type, trains the neural network through a large amount of channel state information data with artificial annotation collected in advance, the input of the neural network is the channel state information, and the output is the type and confidence of the obstacle.

Images