Method and device for locating people in a closed site under non-visual conditions without collision

Abstract

The application discloses a personnel positioning anti-collision method and system in a closed field under non-visible conditions, and the method comprises the following steps: receiving, by a personnel terminal, positioning signals of a first vehicle beacon and a second vehicle beacon on a vehicle through a multi-antenna array to obtain multiple frames of angle of arrival estimation values; when a dispersion degree index of the multiple frames of angle of arrival estimation values corresponding to the first vehicle beacon meets a preset non-direct-viewing determination condition, determining that a non-direct-viewing state exists between the personnel terminal and the first vehicle beacon; determining that the personnel is in a dangerous topological state relative to the vehicle in combination with a receiving state of the second vehicle beacon, and assigning a logical distance; obtaining a vehicle speed, determining a risk level according to the logical distance and the speed; and broadcasting a state message by the personnel terminal, and outputting a blocking control signal before starting an engine according to the risk level by a vehicle terminal. The scheme can identify dangerous topological states such as a vehicle bottom blind area in a metal multipath environment, and complete end-side blocking before starting the vehicle, thereby improving personnel anti-collision safety.

Description

Technical Field

[0001] This invention relates to the field of collision avoidance control technology, specifically to a method and device for personnel positioning and collision avoidance under non-visual conditions in enclosed spaces. Background Technology

[0002] In enclosed or semi-enclosed industrial settings such as open-pit mines, large vehicles, maintenance personnel, and pedestrians often coexist in the same work area, highlighting the need for personnel positioning and vehicle collision avoidance safety monitoring. In these scenarios, mining trucks and other vehicles have bodies composed of large areas of metal structures, and the working environment presents multiple adverse factors such as dust, noise, signal obstruction, and unstable geological conditions.

[0003] Existing pedestrian collision avoidance systems typically use the pedestrian's end to receive vehicle side beacon signals. The system estimates the distance between the pedestrian and the vehicle based on the received signal strength and uploads the location information to the cloud for risk assessment. In environments with strong multipath propagation caused by a metal vehicle body, the received signal strength fluctuates drastically due to the superposition of reflected signals, making it difficult to distinguish between the line-of-sight path and the reflected path. When a pedestrian is located under the vehicle or directly beneath the front of the vehicle—outside of blind spots—the line-of-sight path is completely blocked by the metal body. Current methods cannot differentiate between weak signals due to signal attenuation and non-line-of-sight conditions caused by physical obstruction, thus failing to reliably identify the pedestrian's hazardous topological location.

[0004] Furthermore, existing systems rely on cloud-based decision-making for vehicle startup safety. The moment a vehicle starts moving from a standstill is the most dangerous window, and the round-trip latency for cloud communication and decision-making typically exceeds 2 seconds. By the time a blocking command arrives, the vehicle has often already begun to move, failing to provide effective protection for people stationary in blind spots.

[0005] The aforementioned shortcomings manifest in two main ways: First, under the strong multipath influence of a metal vehicle body, existing signal strength estimation methods cannot reliably distinguish between line-of-sight paths and physical obstructions, resulting in the inability to identify individuals in non-visual blind spots such as under the vehicle or below the front. Second, the initial stage from vehicle standstill to start relies on cloud-based decision-making; communication and computational delays prevent blocking commands from arriving before engine start, creating a blind spot safety window. Therefore, it is necessary to provide a collision avoidance technology solution that can reliably identify non-visual blind spot states without relying on real-time ranging based on line-of-sight signal strength and complete end-side blocking control before vehicle start. Summary of the Invention

[0006] To address the issues of unreliable determination of non-visual blind spot conditions in multipath environments and the inability to fill the time window of vehicle start-up blind spots, this invention provides a method, system, electronic device, and computer-readable storage medium for personnel positioning and collision avoidance under non-visual conditions in enclosed spaces.

[0007] In one aspect, the present invention provides a method for personnel positioning and collision avoidance under non-visual conditions in an enclosed space, comprising: receiving positioning signals transmitted by a first vehicle-mounted beacon and a second vehicle-mounted beacon installed on a vehicle via a multi-antenna array, and obtaining multi-frame angle-of-arrival estimates corresponding to the first vehicle-mounted beacon and the second vehicle-mounted beacon respectively; calculating a dispersion index of the multi-frame angle-of-arrival estimates corresponding to the first vehicle-mounted beacon; determining that the personnel and the first vehicle-mounted beacon are in a non-visual state when the dispersion index meets a preset non-line-of-sight determination condition; and determining that the personnel are in a non-visual state with the first vehicle-mounted beacon when the personnel and the first vehicle-mounted beacon are in a non-visual state and the personnel receive the positioning signal corresponding to the second vehicle-mounted beacon within a preset time window. The system describes the vehicle as being in a dangerous topology state; assigns a preset logical distance to the dangerous topology state; acquires the vehicle's speed; determines a risk level based on the logical distance and the speed, wherein when the person is in the dangerous topology state and the speed is less than a preset stationary speed threshold, the blocking risk level is directly determined; otherwise, the risk level is determined based on the relationship between the quotient of the logical distance and the speed and a preset collision time threshold; the person broadcasts a status message containing the risk level via a local wireless link; the vehicle listens for and caches the status message while the vehicle is not running; the vehicle reads the latest cached status message during the vehicle startup phase, and outputs a blocking control signal before the engine starts when the risk level is the blocking risk level.

[0008] The aforementioned method distinguishes between line-of-sight and non-line-of-sight states by the dispersion of the angle-of-arrival estimate, eliminating reliance on received signal strength indicators for single distance estimation. This allows for the identification of non-visual blind spots caused by physical obstructions in metallic multipath environments. A preset logical distance is assigned to personnel corresponding to non-visual blind spots, ensuring distance representation does not depend on real-time path loss estimation and improving the reliability of hazardous topology state determination. Simultaneously, after determining the risk level, personnel broadcast a status message via a local wireless link. The vehicle continuously listens for and caches this status message while the vehicle is not running, and reads the latest cached status message during vehicle startup. When the risk level is classified as blocking risk, a blocking control signal is output before engine startup. This process bypasses the cloud, reducing communication and decision-making latency from vehicle standstill to startup, enabling blocking actions to be completed before engine start.

[0009] In one embodiment, the dispersion index includes variance. The preset non-line-of-sight determination condition includes at least one of the following: the variance is greater than a preset variance threshold, and is greater than the preset variance threshold for a consecutive preset number of periods; the angular distribution of the estimated angle of arrival values ​​in multiple frames satisfies a preset bimodal distribution condition. Through the dual confirmation of the variance threshold and multiple consecutive periods, misjudgments caused by instantaneous multipath fluctuations within a single period can be avoided, improving the stability of non-line-of-sight state discrimination.

[0010] In one embodiment, the method further includes: determining the vehicle axis direction based on the angle of arrival and azimuth of the first vehicle-mounted beacon and the second vehicle-mounted beacon; when the dispersion index does not meet the preset non-direct-view determination condition, and the relative azimuth of the person's end relative to the vehicle axis direction is less than a preset angle threshold, and the signal strength corresponding to the first vehicle-mounted beacon is greater than a preset signal strength threshold, determining that the person is in an extremely close-range contact hazard topology state relative to the vehicle; assigning a second preset logical distance to the extremely close-range contact hazard topology state; the hazard topology state also includes the extremely close-range contact hazard topology state. This can include the extremely close-range contact hazard of the person located in the fan-shaped area in front of the vehicle within the protection scope, avoiding the omission of dangerous scenarios that are directly visible but extremely close, due to reliance solely on non-direct-view determination.

[0011] In one embodiment, before determining the vehicle axis direction based on the azimuth angles of arrival corresponding to the first and second vehicle-mounted beacons, the method further includes: obtaining the pitch and roll angles at the personnel end; and performing a rotation transformation on the azimuth angles of arrival based on the pitch and roll angles to obtain the azimuth angles of arrival in the horizontal plane. When the personnel are in a squatting or lying position, causing the antenna array plane to tilt, the attitude compensation matrix can project the angle of arrival obtained from the tilted array onto the horizontal plane, which can reduce the impact of attitude changes on the calculation of relative azimuth angles.

[0012] In one implementation, obtaining the vehicle's speed includes: the vehicle terminal reading the speed via a vehicle bus and broadcasting the speed message via the local wireless link at a preset period; and the personnel terminal receiving and caching the latest speed message to obtain the vehicle's speed. This method enables the personnel terminal to continuously obtain vehicle speed information without relying on the cloud, and the speed message is transmitted via the local wireless link, ensuring good availability even in communication-restricted areas.

[0013] In one embodiment, the method further includes: receiving a geological score broadcast from an environmental terminal; determining the risk level based on the relationship between the quotient of the logical distance and the speed and a preset collision time threshold specifically includes: determining the blocking risk level when the quotient is less than a first collision time threshold; determining a second risk level when the quotient is not less than the first collision time threshold and not greater than a second collision time threshold, or when the geological score is less than a preset geological score threshold; and determining a third risk level when the quotient is greater than the second collision time threshold and the geological score is not less than the preset geological score threshold. By introducing a geological score, even with a long collision time, the risk level can be tightened under unstable geological conditions, improving environmental adaptability.

[0014] In one implementation, when the risk level is the blocking risk level, the method further includes: sending an emergency stop message via the local wireless link; triggering the vibration device and ring-shaped light indicator on the personnel end to provide a danger warning. The combined effect of passive blocking at the vehicle end and active warning at the personnel end forms a dual safety guarantee.

[0015] In one implementation, the local wireless link is a LoRa link; the status message includes at least one of personnel identification, the risk level, and the hazardous topology state; when the vehicle is not running, the vehicle is powered by the vehicle battery and listens to and caches the status message in a standby mode with standby power consumption less than a preset power consumption threshold. Using LoRa as the local wireless link allows for short-range communication between the vehicle and personnel when wide-area networks such as 4G are unavailable; the constant-power standby listening ensures that personnel status is continuously obtained even after the vehicle is turned off, providing a condition for querying the cached status when the vehicle restarts.

[0016] In one embodiment, the multi-antenna array includes three antennas arranged in an isosceles right triangle, with the spacing between adjacent antennas being 0.45 to 0.5 times the wavelength of the positioning signal carrier. This half-wavelength-level antenna spacing improves the resolution of the angle-of-arrival phase difference while avoiding phase ambiguity. The isosceles right triangle arrangement facilitates simultaneous calculation of azimuth and elevation angles and reduces the risk of image ambiguity.

[0017] On the other hand, the present invention also provides a personnel positioning and collision avoidance system under non-visual conditions in a closed environment. The system includes: an angle-of-arrival (AOA) measurement module, used to receive positioning signals transmitted by a first vehicle-mounted beacon and a second vehicle-mounted beacon mounted on a vehicle via a multi-antenna array at the personnel end, and obtain multi-frame AOA estimates corresponding to the first and second vehicle-mounted beacons respectively; a dispersion calculation module, used to calculate a dispersion index of the multi-frame AOA estimates corresponding to the first vehicle-mounted beacon; a non-line-of-sight (NOS) discrimination module, used to determine that the personnel end and the first vehicle-mounted beacon are in a non-line-of-sight state when the dispersion index meets a preset NOS discrimination condition; and a topology state determination module, used to determine the relative position of the personnel when the personnel end and the first vehicle-mounted beacon are in a non-line-of-sight state, and the personnel end receives the positioning signal corresponding to the second vehicle-mounted beacon within a preset time window. The vehicle is in a dangerous topology state; a logical distance assignment module is used to assign a preset logical distance to the dangerous topology state; a vehicle speed acquisition module is used to acquire the vehicle's speed; a risk level determination module is used to determine the risk level based on the logical distance and the speed, wherein when the person is in the dangerous topology state and the speed is less than a preset stationary speed threshold, the blocking risk level is directly determined; otherwise, the risk level is determined based on the relationship between the quotient of the logical distance and the speed and a preset collision time threshold; a status broadcast module is used to broadcast a status message containing the risk level through a local wireless link; a status caching and blocking control module is used to listen to and cache the status message on the vehicle side when the vehicle is not started, and to read the latest cached status message during the vehicle start-up phase, and to output a blocking control signal before the engine starts when the risk level is the blocking risk level. This system and method correspond one-to-one with each step in the claims, enabling non-visual blind spot identification and pre-start blocking control without relying on the cloud.

[0018] In some preferred embodiments, the topology state determination module is further configured to: determine the vehicle axis direction based on the angle of arrival and azimuth of the first vehicle beacon and the second vehicle beacon; when the dispersion index does not meet the preset non-direct line of sight determination condition, and the relative azimuth of the personnel end relative to the vehicle axis direction is less than a preset angle threshold, and the signal strength corresponding to the first vehicle beacon is greater than a preset signal strength threshold, determine that the personnel are in a dangerous topology state of extremely close contact with the vehicle; the logical distance assignment module is further configured to assign a second preset logical distance to the dangerous topology state of extremely close contact; the risk level determination module is further configured to, when the personnel are in the dangerous topology state of extremely close contact, determine the risk level based on the second preset logical distance and the speed in the manner described above.

[0019] In another aspect, the present invention also provides an electronic device including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the above-described method.

[0020] In another aspect, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0021] The present invention has the following beneficial effects: By calculating the dispersion index of the estimated angle of arrival, and determining the non-line-of-sight state when the dispersion index meets the preset non-line-of-sight judgment conditions, distance misjudgment caused by drastic fluctuations in the received signal strength indication in strong metal multipath environments is avoided. The non-line-of-sight state is combined with the received state of the second vehicle-mounted beacon signal to determine the dangerous topology state, and a preset logical distance is assigned to the dangerous topology state. This enables reliable identification of personnel in non-visual blind spots such as under the vehicle and below the front of the vehicle without relying on real-time physical ranging.

[0022] By determining the risk level at the personnel end and broadcasting status messages via the local wireless link, the vehicle end continuously listens for and caches status messages while the vehicle is not started. During the vehicle start-up phase, the latest cached status message is read, and when the risk level is the blocking risk level, a blocking control signal is output before the engine starts. This allows the blocking decision to be completed at the end side, shortening the response delay at the moment of vehicle start-up and avoiding the start-up blind zone safety window caused by cloud communication delay.

[0023] By setting up extremely close-range contact dangerous topology states, attitude compensation, geological scoring, LoRa local redundant links, active early warning at the personnel end, and constant power standby monitoring, the coverage of extremely close-range contact scenarios in front of vehicles is further improved, the adaptability to attitude changes and complex geological conditions is enhanced, and the dual guarantee capability of communication robustness and safety interruption is strengthened when there is no wide area network coverage. Attached Figure Description

[0024] Figure 1 A schematic diagram of the overall architecture of a personnel positioning and anti-collision system under non-visual conditions in a closed space, provided for an embodiment of this application; Figure 2 This is a flowchart illustrating a method for locating and preventing collisions of personnel in a closed environment under non-visual conditions, as provided in an embodiment of this application. Detailed Implementation

[0025] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0026] In the description of this application, the phrase "in one embodiment" does not necessarily refer to the same embodiment, although it may refer to the same embodiment. Similarly, the phrase "in some embodiments," as used herein, does not necessarily refer to the same embodiment when used multiple times, although it may refer to the same embodiment. As used herein, the term "or" is an inclusive "or" operator and is equivalent to the term "and / or," unless the context clearly specifies otherwise. The term "based on" is not exclusive and allows for reliance on additional factors not described, unless the context clearly specifies otherwise. The word "exemplary" herein means "serving as an example, instance, or illustration." Any embodiment described herein as "exemplary" is not necessarily to be construed as superior to or better than other embodiments. The scope of this application is limited only by the scope of the appended claims, and any examples set forth in this specification are not intended to be limiting, but merely to illustrate some of the many possible embodiments of the claimed application. The various embodiments provided in this application should not be construed as limiting the scope of protection of this application.

[0027] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0028] In this application, unless otherwise expressly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0029] As mentioned earlier, in enclosed areas such as open-pit mines, the metal bodies of large vehicles cause strong multipath propagation, making distance estimation based on received signal strength indicators prone to failure. Furthermore, relying on cloud-based decision-making at the moment a vehicle starts moving from a standstill cannot achieve a safe stop before the engine starts. This application provides a collision avoidance scheme that does not rely on cloud collaboration or real-time received signal strength ranging. It uses the dispersion of multi-antenna angle-of-arrival estimation to determine non-line-of-sight states, assigns logical distances based on topological states, determines the risk level based on logical distance and vehicle speed, and outputs a blocking control signal before vehicle startup through local state caching and startup phase reading.

[0030] See Figure 1 The personnel positioning and collision avoidance system provided in this application includes a personnel terminal 100, a vehicle terminal 200, and an environment terminal 300. The personnel terminal 100 is a wearable or portable positioning terminal; the vehicle terminal 200 is an in-vehicle terminal installed on a vehicle and its associated components; and the environment terminal 300 is an environmental sensing and broadcasting device deployed in the work area. The personnel terminal 100, the vehicle terminal 200, and the environment terminal 300 communicate with each other via a local wireless link and / or a wide-area wireless link.

[0031] The personnel terminal 100 includes a multi-antenna array 110, an inertial measurement unit 120, a first local wireless communication unit 130, a personnel terminal main control module 140, and a warning component 150. The multi-antenna array 110 includes at least three antennas, such as a first antenna, a second antenna, and a third antenna. The three antennas can be arranged in an isosceles right triangle, and the spacing between adjacent antennas can be set to 0.45 to 0.5 times the wavelength of the positioning signal carrier. The inertial measurement unit 120 includes an accelerometer and a magnetometer for measuring the pitch, roll, and yaw angles of the personnel terminal 100. The first local wireless communication unit 130 may include a Bluetooth communication module and a LoRa communication module. The Bluetooth communication module is used to receive positioning signals from vehicle beacons, and the LoRa communication module is used for local data broadcasting with the vehicle terminal 200 and the environmental terminal 300. The personnel terminal main control module 140 is used to perform angle of arrival estimation, dispersion calculation, non-line-of-sight discrimination, topology state determination, risk level determination, and status message generation. The warning component 150 includes a vibration motor and a ring-shaped luminous indicator. A ring-shaped luminous indicator may include multiple luminous units to provide directional warnings to personnel.

[0032] The vehicle-side 200 includes a first vehicle beacon 210, a second vehicle beacon 220, a vehicle-side local wireless communication component 230, a vehicle bus interface 240, a vehicle-side main control module 250, and a blocking execution module 260. The first vehicle beacon 210 can be installed above the front bumper of the vehicle, and the second vehicle beacon 220 can be installed at the center of the rear of the vehicle. The first vehicle beacon 210 and the second vehicle beacon 220 each periodically transmit a positioning signal with their own identifier. The positioning signal can be a Bluetooth extended broadcast frame, carrying a constant tone extension field at the end of the frame. The vehicle-side local wireless communication component 230 is used to listen to the status messages broadcast by the personnel-side 100 and broadcast vehicle speed messages. The vehicle bus interface 240 reads the vehicle speed through the vehicle bus, which includes, but is not limited to, the CAN bus. The blocking execution module 260 may include a relay control circuit for cutting off the starter motor main relay engagement circuit according to the control signal from the vehicle-side main control module 250, thereby preventing the engine from starting. The first vehicle beacon 210, the second vehicle beacon 220, the vehicle-side local wireless communication component 230, the vehicle-side main control module 250, and the blocking execution module 260 can be powered by the vehicle battery after the vehicle is turned off to maintain the monitoring capability.

[0033] The environmental terminal 300 includes an environmental beacon 310 and a roadside unit 320. The environmental beacon 310 is used to periodically broadcast environmental information, including environmental terminal identification and geological scores. The roadside unit 320 is used to receive and process environmental sensing data such as those from geological microseismic sensors. The environmental beacon 310 can be deployed at the boundaries of the work area and in hazardous areas, and its broadcast coverage can be less than or equal to a preset radius.

[0034] During operation, the first vehicle-mounted beacon 210 and the second vehicle-mounted beacon 220 transmit positioning signals respectively. The personnel terminal 100 receives the positioning signals via a multi-antenna array 110 and estimates the angle of arrival using the phase difference between different antennas. The personnel terminal main control module 140 combines the signal reception status of the second vehicle-mounted beacon 220 with the non-line-of-sight discrimination result corresponding to the first vehicle-mounted beacon 210 to determine the hazardous topology. The vehicle terminal 200 reads the vehicle speed via the vehicle bus interface 240 and broadcasts the speed message externally via the vehicle terminal local wireless communication component 230. The personnel terminal 100 receives the speed message and calculates the collision time and determines the risk level based on the logical distance. After generating a status message, the personnel terminal 100 broadcasts it externally via the first local wireless communication component 130. The vehicle terminal main control module 250 continuously listens to and caches this status message while the vehicle is not running. When the vehicle starts, the vehicle terminal main control module 250 reads the latest cached status message and outputs a blocking control signal via the blocking execution module 260 when the risk level meets the blocking conditions.

[0035] For ease of explanation, the following provides specific explanations of some terms used in the embodiments of this application. It should be understood that these explanations are for reading assistance only and are not intended to limit the scope of protection.

[0036] The angle of arrival estimate can be understood as the incident angle determined by the personnel based on the phase difference of the same signal received by the multi-antenna array.

[0037] The dispersion index may include at least one of variance, standard deviation, or mean square error, used to reflect the dispersion of the angle of arrival estimates across multiple frames.

[0038] The non-line-of-sight state refers to a state in which there is no line-of-sight path between the personnel and the first vehicle-mounted beacon, or the line-of-sight path is blocked.

[0039] A hazardous topology refers to a positional relationship between an individual and a vehicle where there is a significant risk of being crushed in a collision, including at least situations where an individual is located under a vehicle or in the vehicle's blind spot. Logical distance is a conservative distance preset for hazardous topologies.

[0040] The collision time is the quotient of the logical distance and the vehicle speed, representing the time margin before a collision occurs between a person and a vehicle under the current relative motion trend.

[0041] The first and second vehicle-mounted beacons are used to transmit positioning signals, and they are installed in different positions on the vehicle, so that personnel can determine the vehicle's axle or the position of the front and rear of the vehicle based on the signal relationship between the two beacons.

[0042] The specific values, durations, and component models in the embodiments of this application are all exemplary and are not limited thereto.

[0043] This application provides a method for personnel positioning and collision avoidance under non-visual conditions in an enclosed space. This method can be executed collaboratively by a personnel terminal 100 and a vehicle terminal 200. The method includes: The personnel terminal receives positioning signals transmitted by the first and second vehicle-mounted beacons via a multi-antenna array, obtaining multi-frame angle-of-arrival (AOA) estimates for the first and second vehicle-mounted beacons respectively. The dispersion index of the multi-frame AOA estimates for the first vehicle-mounted beacon is calculated. When the dispersion index meets a preset non-line-of-sight (NOS) determination condition, the personnel terminal and the first vehicle-mounted beacon are determined to be in a NOS state. When the personnel terminal and the first vehicle-mounted beacon are in a NOS state, and the personnel terminal receives the positioning signal corresponding to the second vehicle-mounted beacon within a preset time window, the personnel are determined to be in a dangerous topology state relative to the vehicle. A preset logical distance is assigned to the dangerous topology state, and the vehicle's speed is obtained. The risk level is determined based on the logical distance and speed. The personnel terminal broadcasts a status message containing the risk level via a local wireless link. The vehicle terminal listens for and caches the status message while the vehicle is not running. During the vehicle startup phase, the vehicle terminal reads the latest cached status message and, if the risk level is a blocking risk level, outputs a blocking control signal before engine startup.

[0044] In one embodiment, the non-line-of-sight determination condition can be determined through pre-calibration. The calibration method can be as follows: A person is placed in a line-of-sight scenario between themselves and the vehicle's first onboard beacon. Multiple frames of angle-of-arrival estimates are collected within a preset calibration period (e.g., 10 consecutive seconds). The variance distribution of the angle-of-arrival in this line-of-sight scenario is calculated to obtain the mean and upper limit of the line-of-sight variance. Next, the person is placed in a typical non-line-of-sight scenario location, such as a vehicle under-vehicle inspection area or a blind spot directly below the vehicle's front. Multiple frames of angle-of-arrival estimates are collected within the same calibration period. The variance distribution of the angle-of-arrival in this non-line-of-sight scenario is calculated to obtain the mean and lower limit of the non-line-of-sight variance. The upper limit of the line-of-sight variance is compared with the lower limit of the non-line-of-sight variance. If there is a gap between them that is greater than a preset discrimination threshold (e.g., 10 degrees squared), the value within this gap is selected as the preset variance threshold. For example, this threshold can be 1.5 to 2 times the upper limit of the line-of-sight variance. For the preset angle interval and preset ratio in the preset bimodal distribution condition, histogram statistics are performed on the estimated angle of arrival values ​​of multiple frames in non-direct-view scenarios to extract the angle difference distribution between two significant peaks and the amplitude ratio distribution of the weaker peak to the stronger peak. The minimum value of the angle difference distribution is taken as the preset angle interval, and the maximum value of the amplitude ratio distribution is taken as the preset ratio. The determination of the number of consecutive preset periods is based on the statistical analysis of the duration of occasional deviation exceeding the standard in direct-view scenarios, and an integer greater than the number of periods corresponding to the occasional duration is selected as the number of consecutive preset periods. Through the above calibration method, the non-direct-view judgment condition can be quantitatively configured before specific implementation, avoiding reliance on subjective experience to set parameters and improving the stability and repeatability of the judgment.

[0045] In one embodiment, the preset logical distance is determined based on the vehicle's physical dimensions, the space occupied by personnel, and a safety margin. Specifically, the ground clearance of large vehicles such as mining trucks is typically between 1.5 meters and 2.5 meters. When personnel are performing maintenance in a squatting or lying position, the height of their center of gravity from the ground does not exceed 0.5 meters, and the extension length of the tools they may hold does not exceed 0.3 meters. The maximum horizontal projection distance from the personnel's center of gravity to the vehicle's chassis metal structure, plus the extension length of the tools, is used as the base distance, and a preset safety margin (e.g., 0.2 meters) is then added to obtain the preset logical distance.

[0046] The method provided in this application distinguishes between line-of-sight and non-line-of-sight paths by differentiating the dispersion of the angle of arrival estimation using multiple antennas, thus circumventing the problem of distorted received signal strength indications in metal multipath environments. After determining a non-line-of-sight state, the method does not attempt to accurately measure distance using an attenuation model, but instead assigns a logical distance based on the topological state of the person relative to the vehicle. Therefore, even if the signal is blocked by the metal vehicle body, people in non-visual blind spots such as under the vehicle or below the front of the vehicle can still be reliably identified as being in dangerous topological states. Furthermore, the risk level determination is completed at the personnel end, and the status message is broadcast via a local wireless link. The vehicle end continues to listen to and cache this status message while the vehicle is not running. When the vehicle starts, the vehicle end does not need to wait for a response from the cloud; it can directly read the latest cached status message and determine whether to output a blocking control signal. In this way, the decision-making link from a stationary state to the moment of starting is shortened to the millisecond level at the end side, without relying on wide area network round-trip latency, thus filling the safety window in the vehicle start-up blind spot. Regarding the characteristic of receiving positioning signals and obtaining angle of arrival estimates, each antenna in the multi-antenna array at the personnel end receives the positioning signal transmitted by the same vehicle-mounted beacon at different spatial locations. Because the path length of electromagnetic waves to different antennas varies, the same signal exhibits different phases on different antennas. When the positioning signal is a Bluetooth extended broadcast frame, its frame end can carry a constant tone extension field. During the duration of the constant tone extension field, the personnel switch receiving channels between antennas in a preset order to collect in-phase and quadrature sampling data from each antenna. Taking three antennas as an example, three sets of sampling data are obtained within one constant tone extension window. By extracting the phase from the in-phase and quadrature sampling data of each antenna pair, the phase difference between the antenna pairs can be obtained. Substituting the phase difference, antenna spacing, and signal wavelength into the angle of arrival (AHA) formula, the candidate AHA estimate is obtained. If the three antennas form an isosceles right triangle layout, the azimuth and elevation components can be obtained simultaneously, and a consistent solution is selected from multiple candidate angles as the current AHA estimate through a triangulation consistency test. Repeating the above estimation process within one heartbeat cycle yields multiple frames of AHA estimates. The length of the heartbeat cycle, the number of estimations in each cycle, and the antenna switching cycle can all be configured according to actual hardware capabilities and power consumption requirements; this embodiment does not limit these.

[0047] In practical implementation, if three antennas are used to form an array with a spacing on the order of half a wavelength, the antenna spacing can be set to approximately 6 centimeters in the 2.45 GHz band. The antenna switching period can be set to 1 microsecond, so that the sampling point of each antenna corresponds to a continuous symbol in the constant tone extension field. Since the above process does not rely on the received signal strength indication but utilizes the stability of the phase difference, the angle of arrival estimate can remain relatively stable even when the metal vehicle body forms a strong multipath but a direct line of sight still exists. Compared with directly using the received signal strength indication for distance estimation, the phase difference is sensitive to multipath superposition in a different way, especially when the energy of the direct line of sight is dominant, the angle of arrival estimate is less affected by the perturbation of the reflected path. This provides a feasible signal basis for subsequently using the dispersion index to distinguish non-direct line of sight states.

[0048] After obtaining the multi-frame angle of arrival (AHA) estimates corresponding to the first vehicle-mounted beacon, the operator-side main control module calculates the dispersion index. Taking variance as an example, the operator first calculates the mean of the multi-frame AHA estimates, and then calculates the average of the sum of squares of the differences between each frame's AHA estimate and the mean. The dispersion index can also use statistical measures such as standard deviation, mean square error, and interquartile range. The dispersion index characterizes the degree of clustering of the multi-frame AHA estimates around their central position. Under line-of-sight conditions, the multi-frame AHA estimates should be concentrated within a narrow angular range; when the line-of-sight path is blocked and only a reflection path exists, multipath signals arrive at the antenna array from different angles, causing significant differences in the frame estimates. Therefore, the magnitude of the dispersion index reflects whether the line-of-sight component is stably dominant in the currently received signal.

[0049] In practice, multiple angle of arrival (AHA) estimates can be obtained at equal intervals, such as 10 estimates every 200ms. In line-of-sight scenarios, since the energy of the direct wave is much higher than that of the reflected wave, the phase difference is dominated by the direct wave, and the dispersion of the 10 estimates is usually small. In non-line-of-sight scenarios, such as under a vehicle, the signal undergoes multiple metal reflections before reaching the antenna array, and the estimates in each frame show significant fluctuations. Compared with directly determining the location based on a single AHA estimate, calculating the dispersion index before determining the non-line-of-sight state can reduce random errors by utilizing statistical information within a short period of time. Even if an AHA estimate jumps due to sudden interference, it will not immediately change the determination of the non-line-of-sight state.

[0050] For features used to determine non-line-of-sight states based on a dispersion index, the personnel end compares the calculated dispersion index with preset non-line-of-sight determination conditions. For example, when the dispersion index includes variance, if the variance is greater than a preset variance threshold, the line-of-sight path of the vehicle-mounted beacon can be considered unreliable, thus determining a non-line-of-sight state between the personnel end and the vehicle-mounted beacon. To avoid misjudgments caused by random fluctuations within a single period, it can also be required that the variance is greater than the preset variance threshold for a consecutive preset number of periods. In another example, histogram statistics can be performed on the estimated angle of arrival values ​​of multiple frames. When the angle distribution shows at least two significant peaks, and the angle interval between the two peaks exceeds a preset difference, and the ratio of the weaker peak to the stronger peak exceeds a preset ratio, it is determined that a preset bimodal distribution condition is met. A bimodal distribution indicates that the signal may be formed by the superposition of multiple reflection paths from different directions. The above preset non-line-of-sight determination conditions can be used individually or in combination. When the preset non-line-of-sight determination conditions are not met, it can be determined that the personnel end and the first vehicle-mounted beacon are in a line-of-sight state or close to a line-of-sight state. The result of this determination will be used for subsequent topology state determination.

[0051] In practical implementation, when using a variance threshold of 25 degrees squared and exceeding the threshold for three consecutive heartbeat cycles as the non-line-of-sight criterion, the variance in a line-of-sight scenario is usually less than 8 degrees squared, while the variance in a non-line-of-sight scenario may exceed 35 degrees squared, thus providing sufficient differentiation. Compared to directly judging obstruction based on the absolute value or fluctuation range of the received signal strength indication, this scheme transforms the judgment object from scalar strength to the statistical dispersion of the angle of arrival, which can better distinguish between situations where the signal weakens due to increased distance and situations where the signal loses its line-of-sight path due to physical obstruction. Especially in non-visual blind spots such as under a vehicle, the received signal strength may still be high due to the short reflection path, which is easily misjudged as safe by traditional schemes; however, when using the dispersion of the angle of arrival for judgment, because the direct path is obstructed by the vehicle body, the estimated values ​​of multiple frames will show significant dispersion, thereby triggering non-line-of-sight judgment.

[0052] After determining a non-line-of-sight situation, the personnel further determines the dangerous topology state by combining the signal reception status of the second vehicle-mounted beacon. If the personnel are in a non-line-of-sight situation with the first vehicle-mounted beacon and receive the positioning signal corresponding to the second vehicle-mounted beacon within a preset time window, it can be assumed that the personnel are located in a blind spot such as in front of or below the vehicle where the first vehicle-mounted beacon is located, while still being able to receive the signal from the rear side of the vehicle. In this case, the personnel are in a dangerous topology state relative to the vehicle. The preset time window can be set to the most recent second, the most recent number of broadcast cycles, etc. The signal reception status of the second vehicle-mounted beacon can be determined by scanning the identifier in the broadcast frame of the second vehicle-mounted beacon. If, although the personnel are in a non-line-of-sight situation with the first vehicle-mounted beacon, they do not receive the signal corresponding to the second vehicle-mounted beacon within the preset time window, it can be assumed that the personnel may be far from the vehicle or not in an area covered by both beacons, and this is not considered a dangerous topology state.

[0053] In practice, when personnel are inspecting under the vehicle, the first vehicle-mounted beacon, installed above the front bumper, may have its direct signal blocked by the front metal structure of the vehicle, resulting in significant dispersion in the estimated angle of arrival (ACO) value. Simultaneously, the rear beacon, located diagonally behind the personnel, can receive its signal via the side or rear of the vehicle, meaning personnel may still receive the second vehicle-mounted beacon signal. Combining these two factors allows for the differentiation between the blind spot under the vehicle and the simple attenuation of the first vehicle-mounted beacon signal. Relying solely on the non-direct line-of-sight status of the first vehicle-mounted beacon might misjudge personnel as being in the front blind spot; conversely, relying solely on the presence of the second vehicle-mounted beacon signal fails to distinguish whether personnel are directly in front of the vehicle. By using both the non-direct line-of-sight status of the first vehicle-mounted beacon and the receptivity of the second vehicle-mounted beacon as criteria for determining hazardous topology states, the specificity and reliability of topology determination can be improved.

[0054] Once a hazardous topology is identified, a preset logical distance is assigned to it by the personnel. This preset logical distance is not derived in real-time from the path loss model, but rather is a conservative distance value pre-set for the hazardous topology. For example, if a person is located under the vehicle or in the vehicle's blind spot, a logical distance of 0.5 meters can be assigned. When a person is in very close proximity to a hazardous topology, a smaller logical distance can be assigned. The logical distance is assigned based on the safety boundary corresponding to the person's topological position relative to the vehicle, not the instantaneous intensity of the current signal. In subsequent collision time calculations, using the preset logical distance instead of the real-time estimated physical distance avoids errors from real-time ranging under non-line-of-sight or strong multipath conditions from being included in the risk assessment.

[0055] In practice, setting the logical distance for the blind spot under the vehicle to 0.5 meters is more conservative and stable than estimating the distance using the received signal strength indicator. Since the received signal strength indicator can fluctuate by more than 10 dBm under metal multipath propagation, extrapolating the distance based on this fluctuation could result in significant variations between 0.3 meters and several meters, causing the risk level to fluctuate wildly. The preset logical distance anchors the distance to a conservative value corresponding to the hazardous topology, ensuring that even when the vehicle is stationary near a person, a clear and reliable distance can be assigned to that person. This eliminates the interference of real-time distance estimation errors in risk level determination, making it easier to trigger blocking conditions before the vehicle starts moving.

[0056] For vehicle speed acquisition, the vehicle end reads the vehicle speed via the vehicle bus. The vehicle bus can be a CAN bus connected to the OBD-II interface, and the vehicle speed can be read from standard parameter identifiers. After reading the vehicle speed, the vehicle end broadcasts a vehicle speed message via a local wireless link at a preset period. The vehicle speed message can include the vehicle identifier, vehicle speed, and transmission timestamp. The personnel end receives and caches the latest vehicle speed message via the local wireless link. When the personnel end needs to calculate the collision time, it uses the most recently cached vehicle speed value. When the vehicle speed is less than a preset stationary speed threshold, the vehicle can be considered stationary. This preset stationary speed threshold can be set according to the actual vehicle operating conditions, for example, 0.1 m / s. If personnel are in a dangerous topology state while the vehicle is stationary, the blocking protection will not be lost because the collision time cannot be calculated; instead, the blocking risk level will be directly determined.

[0057] In practice, the broadcast period for vehicle speed messages can be set to 100ms, matching the risk assessment within one heartbeat cycle on the personnel end. Compared to relying on cloud-based vehicle speed reports or personnel indirectly calculating vehicle speed based on signal changes, directly reading the vehicle speed from the vehicle bus on the vehicle end provides a value consistent with the actual vehicle's motion state, avoiding collision time distortion caused by communication link delays or signal estimation errors. Especially when the vehicle is in a rapid transition from stationary to moving, the short-cycle update of the local broadcast link ensures that the speed cached on the personnel end is sufficiently close to the actual situation.

[0058] In determining the risk level, when personnel are in a dangerous topological state and the vehicle speed is less than a preset stationary speed threshold, the blocking risk level is directly determined without relying on collision time calculation. This is because when the vehicle is stationary, if personnel are already in a dangerous topological state such as under the vehicle or in the blind spot in front of the vehicle, the safety window is extremely short once the vehicle starts moving, so blocking should be triggered directly before starting. When the vehicle speed is not less than the preset stationary speed threshold, the personnel end calculates the collision time based on the quotient of the logical distance and the vehicle speed, and compares the collision time with the preset collision time threshold. For example, if the collision time is less than the first collision time threshold, the blocking risk level is determined; if the collision time is between the first and second collision time thresholds, the second risk level is determined; if the collision time is greater than the second collision time threshold, the third risk level is determined. If the vehicle speed is less than the preset stationary speed threshold but personnel are not in a dangerous topological state, the collision time can be considered infinite. In this way, the risk level reflects both the topological relationship between personnel and vehicles and the time urgency of relative motion.

[0059] In practice, when a vehicle moves towards a person in the blind spot under the vehicle at a speed of 0.5 m / s, and the logical distance corresponding to this dangerous topology is 0.5 m, the collision time is 1 second, which is less than the first collision time threshold of 2 seconds. Therefore, the blocking risk level is directly triggered. However, when the vehicle is stationary and its speed is 0, the collision time cannot be mathematically calculated effectively, but a person may still be under the vehicle. If only the collision time is used for judgment, a stationary vehicle would be mistakenly judged as safe, and ignition could not be blocked. This solution first determines whether the vehicle speed is less than the preset stationary speed threshold in the dangerous topology state before deciding whether to directly determine the blocking risk level, thus avoiding this logical flaw. Compared with solutions based solely on distance thresholds or solely on collision time, this method is more in line with actual safety decisions: it considers both the current distance and whether the vehicle is about to move.

[0060] After the risk level is determined, personnel broadcast status messages via a local wireless link. These status messages can include at least one of the following: personnel identification, risk level, and hazardous topology status. Broadcasting of status messages can be performed within each heartbeat cycle, or the broadcast frequency can be adjusted according to the risk level. When the risk level is high, personnel can initiate additional immediate broadcasts without waiting for a fixed period. Through local wireless link broadcasting, vehicles and other devices in the same area can directly learn about personnel status without cloud relay. This ensures that even if the wide area network is interrupted, local risk status can still be transmitted between vehicles and personnel.

[0061] In practice, the personnel broadcast status messages during each heartbeat cycle. When the risk level reaches the blocking risk level, they also immediately send an emergency stop message via the local wireless link. Compared to uploading only to the cloud platform via 4G, the local broadcast link has shorter link latency and stronger availability. The vehicle can continuously listen to the status messages broadcast locally, and even if the vehicle is turned off and in standby mode, it can refresh the cached content in a timely manner, thus providing a basis for blocking judgment during the vehicle startup phase.

[0062] While the vehicle is not running, the vehicle-mounted system listens for status messages broadcast by personnel and caches the most recently acquired status message in its main control module. Statuses during which the vehicle is not running include engine off, ignition switch off, and engine stopped. The vehicle-mounted system can be powered by the vehicle's battery, maintaining its listening capability in a low-power standby mode with standby power consumption below a preset threshold. Because the vehicle-mounted system is continuously listening, even if personnel enter under the vehicle several hours after the engine is off, the system can update its cache as long as the personnel's broadcast is captured. This caching mechanism ensures that the vehicle does not need to establish communication with newly emerging statuses upon startup, but instead directly reads the existing cache, thereby reducing decision-making delays.

[0063] In practice, the standby current at the vehicle end can be controlled below the milliampere level to ensure that the vehicle battery is not significantly burdened during long-term parking. When the vehicle starts, the ignition switch closing powers on the vehicle system, but the vehicle's main control module is already operational or can be quickly woken up, reading the latest cached status message within milliseconds. Compared to cloud-based decision-making, this process does not require waiting for the vehicle to connect to the wide area network and upload data, nor does it require waiting for a blocking command from the cloud; therefore, the time window from ignition to decision is compressed.

[0064] During vehicle startup, the vehicle reads the latest status message from the cache. If the risk level in the cache is a blocking risk level, the vehicle outputs a blocking control signal before engine startup, for example, by pulling down a general-purpose input / output pin to drive the relay control circuit to disconnect the starter motor main relay's engagement circuit. Since engine startup and vehicle movement typically occur after the starter motor is engaged, disconnecting the starter motor main relay earlier can prevent engine startup. This blocking control signal can be maintained until the risk status is eliminated or manually reset. The entire judgment and blocking process does not involve the cloud and can be completed within a preset time range. For example, the time from ignition switch closure to blocking control signal output can be completed within 150 milliseconds. Since engine startup itself requires several hundred milliseconds, this blocking action can occur before the engine actually starts.

[0065] In practice, the vehicle-side main control module can immediately query the risk level in the cache after initialization. If the risk level is a blocking risk level, it pulls low on the general-purpose input / output pins connected to the relay control circuit. After the relay control circuit responds, the starter motor main relay cannot engage because its engagement circuit is cut off, the starter motor receives no power, and the engine cannot start. Compared with solutions that remotely cut off fuel or power via the cloud, this solution's response link does not pass through an external network, is not limited by network-free areas such as deep mines or tunnels, and can cover the most dangerous window of vehicle startup.

[0066] To reduce the false alarm rate of non-line-of-sight condition discrimination, in one optional implementation, the dispersion index includes variance; the preset non-line-of-sight determination condition includes at least one of the following: the variance is greater than a preset variance threshold, and is greater than the preset variance threshold for a consecutive preset number of periods; the angle distribution of the multi-frame angle of arrival estimates satisfies a preset bimodal distribution condition. The specific values ​​of the variance threshold and the consecutive periods can be calibrated according to the scene. The bimodal distribution condition can be determined by the angle difference between the two peaks being greater than a preset difference and the ratio of the weaker peak to the stronger peak being greater than a preset ratio. Since metal multipath environments may experience staged reflection occlusion, confirmation over multiple consecutive periods can filter out sudden fluctuations; while the bimodal distribution condition is for situations where multiple reflection paths are simultaneously strong, avoiding multiple reflections canceling each other out in a single angle of arrival estimate, which would result in low variance.

[0067] To further cover the extremely close contact hazard of personnel located in the fan-shaped area in front of the vehicle, in an optional embodiment, the method further includes determining the vehicle axis direction based on the azimuth angles of arrival corresponding to the first and second vehicle-mounted beacons. The personnel end can measure the angles of arrival of the first and second vehicle-mounted beacons respectively, forming two direction vectors in the personnel end's local reference frame. The vehicle axis direction can be determined by the direction from the rear beacon to the front beacon. Then, the relative azimuth angle of the personnel end relative to the vehicle axis direction is calculated. When the dispersion index does not meet the preset non-line-of-sight judgment condition, it indicates that the first vehicle-mounted beacon and the personnel end may still be in a line-of-sight relationship. However, if the relative azimuth angle is less than a preset angle threshold, and the signal strength corresponding to the first vehicle-mounted beacon is greater than a preset signal strength threshold, it indicates that the personnel are located in the extremely close area in front of the vehicle. In this case, the extremely close contact hazard topology is determined, and a second preset logical distance is assigned. The relative azimuth angle can be set to within 30 degrees, and the signal strength can be set to greater than -60 dBm. This optimization measure incorporates the forward fan-shaped area into a dangerous topology, enabling the identification of people in or near the front of the vehicle even when the direct line of sight is unobstructed.

[0068] In one embodiment, the second preset logical distance is determined based on the vehicle's physical dimensions, the space occupied by occupants, and a safety margin. Specifically, the boundary of the fan-shaped danger zone in front of the vehicle is defined as the distance extending outward from the front of the vehicle's bumper. This distance should be less than the maximum displacement within the vehicle's start-up response time at a person's walking speed. Considering that the forearm length of an adult is approximately 0.3 to 0.4 meters, the minimum distance at which a person's limbs may touch the bumper when they are in front of the vehicle and the relative azimuth angle is less than 30 degrees is the second preset logical distance.

[0069] To reduce the impact of personnel's squatting or lying postures on the angle of arrival (AH) and azimuth, the pitch and roll angles can be obtained from the personnel before determining the vehicle's axis of orientation. The pitch and roll angles can be provided by the accelerometers in the inertial measurement unit (IMU). A rotation matrix is ​​constructed based on the pitch and roll angles, and the AH vector is rotated to project the incident direction vector from the tilted antenna array onto the horizontal plane, yielding the AH and azimuth in the horizontal plane. For example, the antenna array for a squatting personnel may be tilted forward. Without compensation, the azimuth of the vehicle's beacon will deviate due to the array tilt, affecting the calculation of the relative azimuth and vehicle axis of orientation. Compensation reduces topology misjudgments caused by changes in personnel posture.

[0070] To obtain accurate vehicle speed information, in one optional implementation, acquiring vehicle speed specifically includes the vehicle end reading the speed via the vehicle bus and broadcasting the speed report via a local wireless link at a preset period, as well as the personnel receiving and buffering the latest speed report. The vehicle bus can read standard vehicle speed parameters based on the OBD-II interface. This method does not rely on indirect estimation of the vehicle's motion state by the personnel via visual or radio signals, nor does it rely on cloud-based data transmission, thereby reducing speed errors. The broadcast period of the speed report can be adjusted according to the rate of change of vehicle speed, for example, 100 milliseconds.

[0071] To adapt risk levels to changes in geological conditions, in one optional implementation, personnel receive a geological score broadcast from the environmental unit. This geological score, calculated by the environmental unit based on data from geological microseismic sensors, characterizes the geological stability of the work area. Under direct line-of-sight conditions, if the collision time exceeds a first collision time threshold, the risk should ideally be reduced. However, if the geological score is below a preset geological score threshold, it indicates geological instability, and the risk level cannot be simply reduced. Therefore, a second risk level is determined when the collision time is not less than the first collision time threshold and not greater than the second collision time threshold, or when the geological score is less than the preset threshold; a third risk level is determined when the collision time exceeds the second collision time threshold and the geological score is not less than the preset threshold. This ensures that collision risks in geologically unstable areas are not underestimated, thereby improving environmental adaptability.

[0072] To enhance the alert intensity in hazardous situations, in one optional implementation, when the risk level is a blocking risk level, the personnel send an emergency stop message via a local wireless link, simultaneously triggering a vibration device and a ring-shaped luminous indicator. The vibration device can activate high-frequency vibration; the luminous unit in the ring-shaped luminous indicator can flash in the direction of the vehicle. This way, even if personnel are unable to notice the audible and visual alarms in dusty or noisy environments, they can still be alerted promptly through tactile and nearby visual cues. Simultaneously, the emergency stop message can be transmitted via the local wireless link to both the vehicle and environmental terminals, providing redundant sources for vehicle-side judgment and area broadcasting.

[0073] To enhance the robustness of local communication in environments without a wide area network, in one optional implementation, the local wireless link is a LoRa link. Compared to Bluetooth links, LoRa links have advantages in line-of-sight and penetration performance, making them particularly suitable for local redundant communication within the operational area. Status messages may include at least one of personnel identification, risk level, and hazardous topology status. When the vehicle is not running, the vehicle is powered by the vehicle battery and listens for and buffers status messages in a standby mode with standby power consumption below a preset power consumption threshold. Through LoRa and constant-power standby, the vehicle can maintain status monitoring capabilities during long-term parking.

[0074] To improve the phase difference resolution and angle calculation capability of angle of arrival estimation, in one optional implementation, the multi-antenna array includes three antennas arranged in an isosceles right triangle, with the spacing between adjacent antennas being 0.45 to 0.5 times the wavelength of the positioning signal carrier. Taking a 2.45 GHz signal as an example, the antenna spacing can be approximately 6 centimeters. This half-wavelength spacing ensures that the phase difference between two antennas will not exceed 180 degrees within a range of ±90 degrees, and that 2π ambiguity will not occur when the phase difference is between -90 and +90 degrees. The third antenna in the vertical direction can be used for elevation angle calculation, and combined with triangulation consistency checks, left-right mirror ambiguity can be eliminated.

[0075] In a specific application scenario, a worker is performing maintenance under a mining truck. The vehicle is off, and the worker's positioning card is located under the truck. The first vehicle-mounted beacon is installed above the front bumper, but its direct signal is blocked by the front metal structure of the vehicle, resulting in significant dispersion in the estimated angle of arrival (ACO) measured by the worker. For example, within three consecutive 200ms heartbeat cycles, the variance is greater than 25 degrees squared, leading the worker to determine that the first vehicle-mounted beacon is not in a line-of-sight state. Simultaneously, a second vehicle-mounted beacon is installed at the center of the rear of the truck, and its signal reaches the worker from the rear or side of the truck. This beacon was detected by the worker within the last second, thus determining that the worker is in a dangerous topology state, and assigning a logical distance of 0.5 meters to this dangerous topology state. During the vehicle's parking period, the worker continuously broadcasts a speed message via LoRa, indicating a current speed of 0 m / s, which is less than the preset stationary speed threshold of 0.1 m / s. Based on the dangerous topology state and the stationary speed, the worker directly determines the blocking risk level and broadcasts a status message via LoRa. The vehicle's terminal caches this status message while it is not running. When the driver enters the cab and closes the ignition switch, the vehicle's terminal reads the latest cached status message after initialization, detects a risk level of blocking, and immediately pulls low on the relay control pin. Within tens of milliseconds, the relay control circuit cuts off the starter motor's main relay engagement circuit, preventing the starter motor from operating and the engine from starting. The entire process does not involve the 4G network, and the blocking is successful even if there is no cloud connection in the work area.

[0076] In another scenario, personnel walk along a work surface while a vehicle approaches them at a certain speed. The estimated angle of arrival (AHA) of the first vehicle-mounted beacon obtained by the personnel via a multi-antenna array has low dispersion and does not meet the non-line-of-sight (NOS) determination condition. The personnel calculate the vehicle's axis direction based on the AHA and azimuth angles of the first and second vehicle-mounted beacons, determining that the relative azimuth angle of the personnel to the vehicle's axis direction is approximately 25 degrees, less than 30 degrees. The signal strength of the first vehicle-mounted beacon is greater than -60 dBm, therefore, the personnel are determined to be in a state of extremely close contact with a dangerous topology, and a second preset logical distance of 0.3 meters is assigned. The vehicle broadcasts a speed of 1 meter per second. The personnel calculate the collision time as 0.3 seconds, which is less than the first collision time threshold of 2 seconds, determining the blocking risk level, and broadcasting a status message and an emergency stop message via LoRa. After reading the status message during the start-up phase or movement, the vehicle can output a blocking control signal or trigger an emergency stop. Simultaneously, the vibration motor and ring-shaped indicator on the personnel activate, prompting the personnel to evacuate away from the vehicle. In this scenario, although a direct line of sight exists, the person is still in the very close area in front of the vehicle. Relying solely on non-direct line of sight detection is insufficient to cover the risk. Therefore, the extremely close contact with the dangerous topology state supplements the risk outside the blind spot.

[0077] In deep mine tunnels, 4G signals are absent, and only local wireless links exist in the work area. When personnel are in the blind spot under the vehicle, the personnel terminal determines the risk level to be a blocking risk level. Status messages cannot be uploaded to the cloud platform via 4G, but can be broadcast via LoRa. The vehicle terminal remains in a standby listening state, able to receive and buffer status messages via LoRa. When the driver starts the engine, the vehicle terminal reads the buffered status and completes the blocking, indicating that the entire safety logic does not rely on cloud connectivity. Furthermore, when the personnel card detects a blocking risk level, even if the vehicle terminal fails to receive the broadcast in time, the personnel terminal can still send an emergency stop command via LoRa, triggering a local response from nearby roadside units or the vehicle terminal. This scenario demonstrates that, with local wireless links and end-side buffering, the system still possesses secure blocking capabilities in areas without wide area network coverage.

[0078] To ensure system safety, this method can also set several boundary conditions. For the vehicle, the blocking control signal is only output when the risk level is at the blocking risk level; once the risk is eliminated and the vehicle re-enters the inactive listening state, the blocking control signal can be revoked. For personnel, the emergency alarm can be stopped when the vehicle speed exceeds a preset safe speed threshold and personnel are in a safe, distant area. For topology status determination, if personnel fail to receive any vehicle beacon signal within the most recent preset time window, no dangerous topology status determination is performed to avoid unfounded safety judgments due to signal loss. During vehicle startup, if the timestamp corresponding to the cached status message exceeds a preset period, it can be updated by monitoring over-the-air broadcasts. If updates are still unavailable, a conservative strategy can be adopted, which may include outputting a vehicle startup restriction warning message before startup without outputting the blocking control signal. Furthermore, once the geological score returns to normal and the collision time exceeds the second collision time threshold, the risk level can be reduced from the second risk level to the third risk level. The above boundary conditions can prevent the system from malfunctioning and repeatedly blocking vehicle startup due to individual sensor failures or signal loss.

[0079] Next, refer to Figure 2 A complete process reduction of the above method will be described. For ease of description, the steps in the process will be numbered as steps S101 to S110.

[0080] Step S101: The personnel end receives the positioning signals transmitted by the first vehicle beacon and the second vehicle beacon through a multi-antenna array, and obtains the multi-frame angle of arrival estimates corresponding to the first vehicle beacon and the second vehicle beacon respectively.

[0081] Step S102: The personnel end calculates the dispersion index of the multi-frame angle of arrival estimates corresponding to the first vehicle beacon.

[0082] Step S103: The personnel end determines whether the dispersion index meets the preset non-direct-view judgment condition. If it meets the condition, proceed to step S104; if it does not meet the condition, proceed to step S108 to continue the extremely close-range contact topology state judgment, or return to step S101 to continue listening.

[0083] Step S104: The personnel determine whether they have received the positioning signal corresponding to the second vehicle beacon within the preset time window. If yes, proceed to step S105; if no, return to step S101 to continue listening.

[0084] Step S105: The personnel end determines that the personnel are in a dangerous topology state relative to the vehicle, and assigns a preset logical distance to the dangerous topology state.

[0085] Step S106: The personnel end obtains the vehicle speed, wherein the vehicle speed is read from the vehicle bus by the vehicle end and broadcast through the local wireless link.

[0086] Step S107: The personnel end determines whether the personnel are in a dangerous topology state and whether the vehicle speed is less than a preset stationary speed threshold. If yes, directly determine the blocking risk level and proceed to step S109; if no, determine the risk level based on the relationship between the quotient of logical distance and vehicle speed and a preset collision time threshold.

[0087] Step S108: The personnel end determines the vehicle axis direction based on the arrival angle and azimuth angle corresponding to the first vehicle beacon and the second vehicle beacon. When the dispersion index does not meet the preset non-direct line of sight judgment condition, the relative azimuth angle is less than the preset angle threshold, and the signal strength of the first vehicle beacon is greater than the preset signal strength threshold, the extremely close contact dangerous topology state is determined, a second preset logical distance is assigned, and it participates in the risk level determination.

[0088] Step S109: The personnel broadcast a status message containing the risk level via the local wireless link.

[0089] Step S110: The vehicle listens for and caches status messages while the vehicle is not started; during the vehicle start-up phase, it reads the latest cached status message, and if the risk level is the blocking risk level, it outputs a blocking control signal before the engine starts.

[0090] Regarding the step numbers and order in the above process, those skilled in the art will understand that the execution order and conditional branches of each step do not constitute a unique limitation. In specific implementations, the execution order of some steps can be adjusted, or certain steps can be merged or split, all of which are reasonable variations within the scope of protection of this application. The various thresholds, cycles, counts, and other parameters are only illustrative examples and can be specifically determined according to on-site conditions.

[0091] Corresponding to the above method, this application embodiment also provides a control device. The device's functional modules include an angle-of-arrival measurement module, a dispersion calculation module, a non-line-of-sight discrimination module, a topology state determination module, a logical distance assignment module, a vehicle speed acquisition module, a risk level determination module, a status broadcasting module, and a status buffering and blocking control module. The angle-of-arrival measurement module is used to implement the steps in the above method of receiving positioning signals from the first vehicle-mounted beacon and the second vehicle-mounted beacon and obtaining multi-frame angle-of-arrival estimates; the dispersion calculation module is used to implement the step in the above method of calculating the dispersion index; the non-line-of-sight discrimination module is used to implement the step in the above method of determining non-line-of-sight states; the topology state determination module is used to implement the step in the above method of determining dangerous topology states; the logical distance assignment module is used to implement the step in the above method of assigning logical distances to dangerous topology states; the vehicle speed acquisition module is used to implement the step in the above method of acquiring vehicle speed; the risk level determination module is used to implement the step in the above method of determining risk levels; the status broadcasting module is used to implement the step in the above method of broadcasting status messages; and the status buffering and blocking control module is used to implement the step in the above method of buffering status messages at the vehicle end and outputting blocking control signals during the startup phase. The data flow between the modules described above is consistent with the input-output relationship in the aforementioned method steps.

[0092] In summary, the above embodiments form a local security closed loop that does not rely on the cloud. The personnel end obtains the angle of arrival estimate through an array antenna, identifies non-line-of-sight states through a dispersion index, and determines the dangerous topology state by combining the relationship between multiple beacon signals. This solves the problem of reliable identification of personnel located in non-visual blind spots such as under or under the vehicle in metal multipath environments. The introduction of logical distance avoids introducing real-time ranging errors in non-line-of-sight states into risk assessment; collision time unifies logical distance and vehicle speed into a time dimension, giving the risk level a clear physical meaning. When the vehicle is stationary, the combination criterion of dangerous topology state and preset stationary speed threshold directly triggers the blocking risk level, compensating for the failure of traditional collision time-based calculations at zero speed. The personnel end broadcasts status messages, the vehicle end constantly listens and buffers during the inactive period, and outputs blocking control signals before engine start during vehicle startup, enabling decision-making and blocking to be independent of a wide area network and achieving millisecond-level end-side response. Optional features such as attitude compensation, extremely close-range contact status, geological scoring, LoRa local redundancy, and active early warning from personnel further cover additional scenarios and boundary conditions, enhancing the system's adaptability in complex and enclosed environments from different dimensions. This solution improves collision avoidance safety for personnel and vehicles while reducing the risk of missed detections due to multipath ranging errors and cloud latency, demonstrating significant engineering application value.

[0093] Based on the same inventive concept, this application also provides a personnel positioning and collision avoidance system under non-visual conditions in a closed environment. The system includes a personnel end and a vehicle end. The personnel end includes a multi-antenna array, a discreteness calculation module, a non-line-of-sight discrimination module, a topology state determination module, a logical distance assignment module, a vehicle speed acquisition module, a risk level determination module, and a status broadcasting module. The vehicle end includes a status buffering and blocking control module. These modules can be configured according to the aforementioned system architecture. The system's operation is as follows: the multi-antenna array on the personnel end receives positioning signals transmitted by a first vehicle-mounted beacon and a second vehicle-mounted beacon; the discreteness calculation module calculates the discreteness index; the non-line-of-sight discrimination module determines the non-line-of-sight state based on the discreteness index; the topology state determination module determines the dangerous topology state based on the non-line-of-sight state and the reception status of the second vehicle-mounted beacon; the logical distance assignment module assigns a logical distance to the dangerous topology state; the vehicle speed acquisition module acquires the vehicle speed; the risk level determination module determines the risk level; the status broadcasting module broadcasts status messages; the vehicle end's status buffering and blocking control module buffers status messages during the inactive period and outputs blocking control signals based on the risk level during the activation phase. The modules in the system embodiment can be implemented through software, hardware, or a combination of both. When implemented through software, the modules can be integrated into the personnel-side main control module or the vehicle-side main control module.

[0094] In the above embodiments of this application, the descriptions of each embodiment have their own emphasis. Parts not described in detail in a certain embodiment can be referred to in the relevant descriptions of other embodiments. The steps illustrated in the related flowcharts can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowcharts, in some cases, the steps shown or described may be performed in a different order than that shown here. In other words, the order of steps described in the foregoing embodiments is merely an example. Reasonable adjustments to the order of steps based on the content of the embodiments of this application are also within the protection scope of the embodiments of this application.

[0095] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application. Furthermore, those skilled in the art can combine and integrate the different embodiments or examples described in this specification.

Claims

1. A method for personnel positioning and collision avoidance under non-visual conditions in an enclosed space, characterized in that, include: The personnel receive the positioning signals transmitted by the first vehicle-mounted beacon and the second vehicle-mounted beacon installed on the vehicle through a multi-antenna array, and obtain the multi-frame angle of arrival estimates corresponding to the first vehicle-mounted beacon and the second vehicle-mounted beacon respectively; Calculate the dispersion index of the multi-frame angle of arrival estimates corresponding to the first vehicle-mounted beacon; When the dispersion index meets the preset non-direct line of sight determination condition, it is determined that the personnel end and the first vehicle beacon are in a non-direct line of sight state. When the personnel terminal and the first vehicle beacon are in a non-line-of-sight state, and the personnel terminal receives the positioning signal corresponding to the second vehicle beacon within a preset time window, it is determined that the personnel are in a dangerous topology state relative to the vehicle, and the dangerous topology state has a preset logical distance; Obtain the speed of the vehicle; The risk level is determined based on the logical distance and the speed. When the personnel are in the dangerous topology and the speed is less than a preset stationary speed threshold, the blocking risk level is directly determined. Otherwise, the risk level is determined based on the relationship between the quotient of the logical distance and the speed and a preset collision time threshold. The personnel broadcast a status message containing the risk level via a local wireless link. The vehicle-side system listens for and caches the status messages while the vehicle is not running. The vehicle terminal reads the latest status message from the cache during the vehicle startup phase, and outputs a blocking control signal before the engine starts when the risk level is the blocking risk level.

2. The method according to claim 1, characterized in that, The dispersion index includes variance; the preset non-direct observation judgment condition includes at least one of the following: The variance is greater than a preset variance threshold, and is greater than the preset variance threshold for a consecutive preset number of periods. The angle distribution of the multi-frame angle of arrival estimates satisfies the preset bimodal distribution condition.

3. The method according to claim 1, characterized in that, The method further includes: The vehicle axis direction is determined based on the arrival angle and azimuth angle corresponding to the first vehicle beacon and the second vehicle beacon; When the dispersion index does not meet the preset non-direct view determination condition, and the relative azimuth angle of the personnel end relative to the vehicle axis direction is less than the preset angle threshold, and the signal strength corresponding to the first vehicle beacon is greater than the preset signal strength threshold, it is determined that the personnel are in a dangerous topology state of extremely close contact with the vehicle relative to the vehicle. Assign a second preset logical distance to the aforementioned extremely close contact with dangerous topological states; The hazardous topology states also include the extremely close-range contact hazardous topology states; When the personnel are in the extremely close contact with the dangerous topology, the risk level is determined according to the second preset logical distance and the speed, in accordance with the method of determining the risk level based on the logical distance and the speed.

4. The method according to claim 3, characterized in that, Before determining the vehicle axis direction based on the arrival angle and azimuth angle corresponding to the first vehicle beacon and the second vehicle beacon, the method further includes: Obtain the pitch and roll angles of the personnel end; The azimuth angle of arrival is rotated and transformed based on the pitch angle and the roll angle to obtain the azimuth angle of arrival in the horizontal plane.

5. The method according to claim 1, characterized in that, The process of obtaining the speed of the vehicle includes: The vehicle reads the speed via the vehicle bus and broadcasts the vehicle speed message via the local wireless link at a preset period. The personnel end receives and caches the latest vehicle speed message to obtain the vehicle's speed.

6. The method according to claim 1, characterized in that, The method further includes: Geological scores received from environmental broadcasts; The step of determining the risk level based on the relationship between the quotient of the logical distance and the speed and a preset collision time threshold specifically includes: When the quotient is less than the first collision time threshold, the blocking risk level is determined; When the quotient is not less than the first collision time threshold and not greater than the second collision time threshold, or when the geological score is less than the preset geological score threshold, a second risk level is determined. When the quotient is greater than the second collision time threshold and the geological score is not less than the preset geological score threshold, a third risk level is determined.

7. The method according to claim 1, characterized in that, When the risk level is the blocking risk level, the method further includes: Send an emergency stop message via the local wireless link; The vibration device and the ring-shaped light indicator at the personnel end are triggered to issue a danger warning.

8. The method according to claim 1, characterized in that, The local wireless link is a LoRa link; the status message includes at least one of the personnel identifier, the risk level, and the dangerous topology state; when the vehicle is not started, the vehicle terminal is powered by the vehicle battery and listens to and buffers the status message in a standby mode with standby power consumption less than a preset power consumption threshold.

9. The method according to claim 1, characterized in that, The multi-antenna array includes three antennas arranged in an isosceles right triangle, with the spacing between adjacent antennas being 0.45 to 0.5 times the wavelength of the positioning signal carrier.

10. An electronic device, characterized in that, It includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the method according to any one of claims 1-9.

Images