Scenario-based intelligent helmet module perception switching method and system

Abstract

The present invention discloses a scene-based intelligent helmet module perception switching method and system, which relates to data processing related fields. The method includes: obtaining positioning information through multiple perception modules to perform cluster analysis on multiple radar scanning data sequences, obtaining multiple scanning point clustering result set sequences and associating them with multiple dynamic target sets, obtaining multiple dynamic target scanning result sequence sets to perform multiple dynamic target speed identification, obtaining multiple dynamic target speed sets, calculating multiple overall dynamic speed analysis to obtain multiple risk factors, determining key perception modules and multiple secondary perception modules and switching them, and adjusting the perception frequencies of multiple secondary perception modules. The method solves the technical problem that the user's safety and efficiency are reduced due to the interference of the environment in the perception switching and response speed, and realizes the technical effect of accurately perceiving the helmet module based on the scene and improving the user's safety and efficiency.

Description

Technical Field

[0001] The present application relates to data processing related fields, and in particular to a scene-based intelligent helmet module perception switching method and system. Background Art

[0002] With the rapid development of science and technology, helmet module sensing and switching methods have gradually become one of the core technologies of smart helmets. Helmet module sensing and switching technology is a key feature of modern smart helmets. By integrating various sensors and advanced algorithms, it achieves real-time perception of user status and environmental information, automatically switching the helmet's operating mode or adjusting relevant parameters accordingly. However, existing technologies have technical problems such as environmental interference, which can affect the sensing and response speed, reducing user safety and efficiency. Summary of the Invention

[0003] This application solves the technical problem of reduced user safety and efficiency due to environmental interference in perception switching and response speed by providing a scene-based smart helmet module perception switching method and system, and achieves the technical effect of accurate scene-based perception of the helmet module and improved user safety and efficiency.

[0004] The present application provides a scenario-based smart helmet module perception switching method and system method, which is applied to a scenario-based smart helmet module perception switching system, comprising: using multiple perception modules in the smart helmet to collect radar scanning data sequences and noise data in multiple azimuth areas according to a preset perception frequency, wherein the multiple perception modules are arranged in a ring at multiple positions in the smart helmet;

[0005] Acquiring positioning information through the smart helmet, and the multiple perception modules collecting multiple historical accident data in the multiple azimuth areas according to the positioning information;

[0006] Performing cluster analysis on multiple radar scanning data sequences to obtain multiple scanning point clustering result set sequences, performing dynamic target fitting on multiple first scanning point clustering result sets within a first moment to obtain multiple dynamic target sets, and associating the dynamic target sets with the multiple scanning point clustering result set sequences to obtain multiple dynamic target scanning result sequence sets;

[0007] According to the multiple dynamic target scanning result sequences, multiple dynamic target speeds are identified to obtain multiple dynamic target speed sets, and multiple overall dynamic speeds are calculated;

[0008] Processing and obtaining multiple dynamic target densities based on the multiple dynamic target sets and the positioning information, and analyzing and obtaining multiple hazard coefficients for multiple azimuth areas by combining the multiple overall dynamic speeds, multiple noise data, and multiple historical accident data;

[0009] According to the multiple risk factors, a key perception module and multiple secondary perception modules are determined and switched, and the perception frequencies of the multiple secondary perception modules are adjusted.

[0010] In a possible implementation, positioning information is obtained through a smart helmet, and the multiple perception modules collect multiple historical accident data in the multiple azimuth areas based on the positioning information, and perform the following processing: obtaining current positioning information through a positioning unit in the smart helmet;

[0011] Determining a plurality of target areas for historical accident data extraction based on the plurality of orientation areas and positioning information;

[0012] The number of accidents in the multiple target areas within a historical period is extracted to obtain multiple historical accident data.

[0013] In a possible implementation, cluster analysis is performed on multiple radar scan data sequences to obtain multiple scan point cluster result set sequences, dynamic target fitting is performed on multiple first scan point cluster result sets within a first moment to obtain multiple dynamic target sets, and the dynamic target sets are associated with the multiple scan point cluster result set sequences, and the following processing is performed: cluster analysis is performed on the multiple radar scan data sequences, and scan points with a distance less than a distance threshold are grouped into one cluster result to obtain multiple scan point cluster result set sequences, where each scan point cluster result set sequence includes scan point cluster result sets at multiple moments;

[0014] Perform indexing based on the first moment to obtain a plurality of first scanning point clustering result sets;

[0015] Obtain a dynamic target anchor frame set, perform iterative dynamic target fitting on all scanning point clustering results in the plurality of first scanning point clustering result sets, obtain a dynamic target anchor frame with the greatest fitting degree, and obtain multiple dynamic target sets, wherein the dynamic target anchor frame set includes all dynamic target anchor frames of multiple dynamic targets at different angles;

[0016] According to the multiple dynamic target sets, fitting association is performed on the multiple scanning point clustering result set sequences to obtain multiple dynamic target scanning result sequence sets, each dynamic target scanning result including the multiple dynamic targets.

[0017] In a possible implementation, iterative dynamic target fitting is performed on all scanning point clustering results in the plurality of first scanning point clustering result sets to obtain a dynamic target anchor frame with the greatest fitting degree, thereby obtaining a plurality of dynamic target sets, and performing the following processing: obtaining a dynamic target anchor frame set based on all dynamic target anchor frames at different angles of the plurality of dynamic targets;

[0018] Randomly selecting a first dynamic target anchor frame, selecting all scanning point clustering results in the plurality of first scanning point clustering result sets, and calculating and obtaining a plurality of first fit degrees according to a proportion of scanning points in each scanning point clustering result that fall within the first dynamic target anchor frame;

[0019] Randomly selecting a second dynamic target anchor frame again, selecting all scanning point clustering results in the plurality of first scanning point clustering result sets, and calculating to obtain a plurality of second fit degrees;

[0020] Continue to perform iterative fitting, output the dynamic target corresponding to the dynamic target anchor frame with the largest fitting degree in the iterative fitting of each scanning point clustering result, and obtain multiple dynamic target sets.

[0021] In a possible implementation, multiple dynamic target speed identifications are performed based on multiple dynamic target scanning result sequences to obtain multiple dynamic target speed sets, and multiple overall dynamic speeds are calculated. The following processing is performed: according to the first step length, the multiple dynamic target scanning result sequences are extracted to obtain multiple first dynamic target scanning result sets;

[0022] extracting the plurality of dynamic target scanning result sequences according to a second step length to obtain a plurality of second dynamic target scanning result sets, wherein the second step length is greater than the first step length;

[0023] performing multiple dynamic target speed identifications based on the multiple first dynamic target scanning result sets and the multiple second dynamic target scanning result sets to obtain multiple dynamic target speed sets;

[0024] The dynamic target speed means of the plurality of dynamic target speed sets are calculated to obtain a plurality of overall dynamic speeds.

[0025] In a possible implementation, multiple dynamic target speed identifications are performed based on the multiple first dynamic target scanning result sets and the multiple second dynamic target scanning result sets, and the following processing is performed: based on historical perception data, a sample dynamic target scanning result sequence set and a multiple sample dynamic target speed sets are acquired;

[0026] Processing the sample dynamic target scanning result set according to the first step length and the second step length to obtain a plurality of sample first dynamic target scanning result sets and a plurality of sample second dynamic target scanning result sets;

[0027] Based on the slowfast network, a dynamic target speed identifier is trained using the plurality of sample first dynamic target scanning result sets, the plurality of sample second dynamic target scanning result sets and the plurality of sample dynamic target speed sets;

[0028] Based on the dynamic target speed identifier, the plurality of first dynamic target scanning result sets and the plurality of second dynamic target scanning result sets are identified to obtain a plurality of dynamic target speed sets.

[0029] In a possible implementation, based on the multiple dynamic target sets and the positioning information, multiple dynamic target densities are obtained, and combined with the multiple overall dynamic speeds, multiple noise data, and multiple historical accident data, multiple hazard coefficients of multiple azimuth areas are analyzed and obtained, and the following processing is performed: based on the multiple dynamic target sets, a number of dynamic targets is obtained by statistics, and based on the road characteristics of the multiple azimuth areas obtained according to the positioning information, a number of dynamic target densities are calculated;

[0030] Based on historical road safety detection data, a sample dynamic target density set, a sample overall dynamic speed set, a sample noise data set, a sample historical accident data set, and a sample hazard coefficient set are obtained;

[0031] The sample dynamic target density set, the sample overall dynamic speed set, the sample noise data set, and the sample historical accident data set are respectively used in combination with the sample hazard coefficient set to construct multiple hazard identification branches to obtain a hazard identifier;

[0032] Based on the hazard identifier, the multiple dynamic target densities, multiple overall dynamic speeds, multiple noise data and multiple historical accident data are identified to obtain multiple branch hazard coefficient sets, and multiple hazard coefficients are calculated.

[0033] In a possible implementation, based on the multiple risk factors, a key sensing module and multiple secondary sensing modules are determined and switched, and the sensing frequencies of the multiple secondary sensing modules are adjusted. The following processing is performed: the sensing module corresponding to the azimuth area with the highest risk factor is selected as the key sensing module, and the other multiple sensing modules are switched as multiple secondary sensing modules;

[0034] Setting the perception frequency of the key perception module to the maximum perception frequency;

[0035] According to the ratio of the risk coefficients of the orientation areas of the other multiple secondary perception modules to the risk coefficient of the orientation area of ​​the key perception module, the preset perception frequency is corrected and calculated to obtain multiple secondary perception frequencies for adjustment.

[0036] The present application also provides a scenario-based smart helmet module perception switching system, comprising: a first data acquisition module, the first data acquisition module being configured to collect radar scanning data sequences and noise data of multiple azimuth areas according to a preset perception frequency through multiple perception modules within the smart helmet, wherein the multiple perception modules are arranged in a ring at multiple locations within the smart helmet;

[0037] A positioning information acquisition module, the positioning information acquisition module is used to obtain positioning information through the smart helmet, and the multiple perception modules collect multiple historical accident data in the multiple azimuth areas according to the positioning information;

[0038] a cluster analysis module, the cluster analysis module being configured to perform cluster analysis on multiple radar scan data sequences to obtain multiple scan point cluster result set sequences, perform dynamic target fitting within multiple first scan point cluster result sets within a first moment to obtain multiple dynamic target sets, and associate the dynamic target sets with the multiple scan point cluster result set sequences to obtain multiple dynamic target scan result sequence sets;

[0039] A speed identification module, the speed identification module is used to identify multiple dynamic target speeds according to multiple dynamic target scanning result sequences, obtain multiple dynamic target speed sets, and calculate multiple overall dynamic speeds;

[0040] a first analysis module, configured to process and obtain multiple dynamic target densities based on the multiple dynamic target sets and the positioning information, and analyze and obtain multiple risk factors for multiple azimuth areas by combining the multiple overall dynamic speeds, multiple noise data, and multiple historical accident data;

[0041] A switching module is used to determine a key sensing module and multiple secondary sensing modules according to the multiple risk factors and switch them, and to adjust the sensing frequencies of the multiple secondary sensing modules.

[0042] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0043] The scenario-based intelligent helmet module perception switching method and system provided in this application relate to the field of data processing technology. They solve the technical problem that the perception switching and response speed are affected by environmental interference, resulting in reduced user safety and efficiency. They achieve the technical effect of accurately perceiving the helmet module based on the scenario, thereby improving user safety and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention are briefly introduced below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in precise order. Instead, various steps may be processed in reverse order or simultaneously as needed. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0045] Figure 1 A flow chart of a scenario-based smart helmet module perception switching method and system method provided in an embodiment of the present application;

[0046] Figure 2 A schematic diagram of the structure of the scenario-based smart helmet module perception switching system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0047] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0048] In order to make the purpose, technical solutions and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limiting this application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0049] In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict, and the terms “first\second” involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. The terms “including” and “having” and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs. The terms used herein are for the purpose of describing the embodiments of this application only.

[0050] The embodiment of the present application provides a scene-based intelligent helmet module perception switching method and system method, the method is applied to a scene-based intelligent helmet module perception switching system, such as Figure 1 As shown, the method includes:

[0051] Step A100, collecting radar scan data sequences and noise data from multiple azimuth areas according to a preset sensing frequency using multiple sensing modules within the smart helmet, wherein the multiple sensing modules are arranged in a ring shape at multiple locations within the smart helmet;

[0052] Multiple sensing modules are arranged in a circular shape at different positions in the smart helmet to ensure that all directions of the helmet can be covered by the sensing modules, thereby realizing all-round data collection. First, multiple sensing modules in the smart helmet are started. The multiple sensing modules are arranged in a circular shape at different positions in the smart helmet to ensure that all directions of the helmet can be covered by the sensing modules, thereby realizing all-round data collection. Each sensing module has radar scanning and noise detection functions, and collects radar scanning data sequences and noise data at the same time. Further, multiple sensing modules are used to monitor multiple directions of the user according to a preset sensing frequency. The preset sensing frequency is used to determine the density and real-time performance of subsequent data collection. At the same time, the sensing frequency can be dynamically adjusted according to the scenario. For example, the sensing frequency can be increased in scenarios requiring high precision, and the sensing frequency can be reduced in scenarios that do not require frequent updates to save energy. The multiple directions of the user can be radar scanning data and noise data in the front, back, left and right areas. The radar scanning data includes radar data generated by vehicles on the road, which can identify vehicle speed. The noise data can be the average of noise at multiple moments.

[0053] Executing step A200, obtaining positioning information through the smart helmet, and the multiple perception modules collecting multiple historical accident data in the multiple azimuth areas based on the positioning information;

[0054] In one possible implementation, step A200 further includes step A210, obtaining current positioning information through a positioning unit in the smart helmet; and executing step A220, determining multiple target areas for historical accident data extraction based on the multiple orientation areas and the positioning information.

[0055] Before positioning, a fixed area can be divided based on geographical location, for example, a circular area with a certain radius or an area of ​​a specific shape can be set with the current location as the center. When the smart helmet is worn and started, the built-in positioning unit (such as GPS, Beidou, RTK, etc.) starts working and receives signals from the global positioning satellite system. The received satellite signals can then be analyzed by the positioning unit to calculate the current wearer's precise location information, including longitude, latitude, and altitude, etc. The current positioning information obtained is matched with multiple pre-defined orientation areas to determine the specific area to which the current location belongs. Further, based on the current positioning information and possible movement direction (if available), multiple target areas where historical accident data needs to be extracted are determined. The multiple target areas can include the current area, surrounding adjacent areas, etc.

[0056] Execute step A230 to extract the number of accidents in the multiple target areas within a historical period to obtain multiple historical accident data.

[0057] Define a historical time range for extraction, such as one week, one month, one year, etc., and use the geographic location information of the multiple target areas obtained above to filter out the accident data located in the target area through geographic location query or spatial database query. After filtering out the accident data in the target area, further filter out the accident data within the historical time range based on the historical time range determined above, and perform statistics on the filtered historical accident data. The counting function in the database query statement can also be used to calculate the number of accidents in each target area within the specified historical time range, and output multiple historical accident data.

[0058] Executing step A300, performing cluster analysis on multiple radar scan data sequences to obtain multiple scan point cluster result set sequences, performing dynamic target fitting within multiple first scan point cluster result sets within a first moment to obtain multiple dynamic target sets, and associating the dynamic target sets with the multiple scan point cluster result set sequences to obtain multiple dynamic target scan result sequence sets; in one possible implementation, step A300 further includes step A310, performing cluster analysis on the multiple radar scan data sequences, classifying scan points having a distance less than a distance threshold into one cluster result, and obtaining multiple scan point cluster result set sequences, each scan point cluster result set sequence including scan point cluster result sets at multiple moments;

[0059] Clustering of radar scanning data is completed by performing cluster analysis on multiple radar scanning data sequences, and dense scanning points (such as all scanning points of a vehicle or a pedestrian) are formed into a clustering result. This means that a distance threshold is first set according to application requirements and data characteristics. The distance threshold is used to determine the clustering classification of the scanning points. Cluster analysis can be performed on the radar scanning data sequence at each moment through hierarchical clustering. This means that according to the selected clustering algorithm and distance threshold, scanning points with a distance less than the threshold are classified into a clustering result. For the clustering result at each moment, a scanning point clustering result set is generated. The scanning point clustering result set contains the clustering information of all scanning points at the corresponding moment, and the scanning point clustering result sets at different moments are arranged in chronological order to form a scanning point clustering result set sequence.

[0060] Executing step A320, indexing based on the first moment to obtain multiple first scanning point clustering result sets; executing step A330, obtaining a dynamic target anchor frame set, performing iterative dynamic target fitting on all scanning point clustering results in the multiple first scanning point clustering result sets, obtaining a dynamic target anchor frame with the maximum fitting degree, and obtaining multiple dynamic target sets, wherein the dynamic target anchor frame set includes all dynamic target anchor frames of multiple dynamic targets at different angles;

[0061] First, scanning data at a first moment is selected from a radar scanning data sequence, and cluster analysis is performed on the scanning data at the first moment to obtain multiple first scanning point clustering result sets. Then, in one possible implementation, step A330 further includes step A331, obtaining a dynamic target anchor frame set based on all dynamic target anchor frames of multiple dynamic targets at different angles; executing step A332, randomly selecting a first dynamic target anchor frame, selecting all scanning point clustering results within the multiple first scanning point clustering result sets, and calculating multiple first fits based on the proportion of scanning points in each scanning point clustering result that fall within the first dynamic target anchor frame;

[0062] Based on the dynamic targets contained in the dynamic environment, which can be dynamic vehicles, dynamic pedestrians, etc., dynamic target anchor frames are defined for various dynamic targets at different angles. The dynamic target anchor frames are used to characterize the typical shapes and sizes of dynamic targets in radar scanning, thereby obtaining a set of dynamic target anchor frames.

[0063] Furthermore, an anchor frame is randomly selected from the dynamic target anchor frame set as the first dynamic target anchor frame, and the first dynamic target anchor frame is used to select the scanning points in the clustering result of each scanning point clustering result in the multiple first scanning point clustering result sets. At the same time, the RANSAC algorithm is used to fit and calculate the ratio of the number of scanning points falling into the first dynamic target anchor frame to the total number of scanning points in the clustering result. The calculated ratio data is used to reflect the degree of fit between the scanning point clustering result and the selected dynamic target anchor frame, and it is recorded as the first fit degree for output.

[0064] Execute step A333, randomly select the second dynamic target anchor frame again, select all the scanning point clustering results in the multiple first scanning point clustering result sets, and calculate to obtain multiple second fit degrees; execute step A334, continue iterative fitting, output the dynamic target corresponding to the dynamic target anchor frame with the largest fit degree for each scanning point clustering result in the iterative fitting, and obtain multiple dynamic target sets.

[0065] Randomly select an anchor frame from the dynamic target anchor frame set as the dynamic target anchor frame in the current iteration (the first dynamic target anchor frame in the first iteration, the second, third, etc. dynamic target anchor frames in subsequent iterations), so the second dynamic target anchor frame is a dynamic target anchor frame different from the first dynamic target anchor frame, and then use the dynamic target anchor frame selected in the current iteration to perform a frame selection operation on each scanning point clustering result in the multiple first scanning point clustering result sets, and then calculate the ratio of the number of scanning points falling into the current anchor frame to the total number of scanning points in the clustering result for each scanning point clustering result, as the current The fit between the previous anchor frame and the clustering result is calculated to obtain the second fit, and the above steps of frame selection and fit calculation are repeated. A new dynamic target anchor frame is randomly selected in each iteration until the iteration termination condition is met. The iteration termination condition can be that the preset number of iterations is reached, or the fit improvement is no longer obvious. Then, the ransac algorithm is used for fitting, and the shape anchor frames of different vehicles (such as cars and trucks) at different angles (such as left front and front) are used for fitting to obtain the shape anchor frame with the largest fit of the clustering result of each scanning point, and then multiple dynamic target sets are obtained.

[0066] Execute step A340 to perform fitting association on the plurality of scanning point clustering result set sequences based on the plurality of dynamic target sets to obtain a plurality of dynamic target scanning result sequence sets, each dynamic target scanning result including the plurality of dynamic targets.

[0067] Multiple dynamic target sets can be stored in a dynamic target scanning result sequence set corresponding to the scanning results at different moments, and then each scanning point clustering result in the scanning point clustering result set at the current moment is traversed. By calculating the similarity between the scanning point clustering result and the dynamic target anchor frame, for each scanning point clustering result, the multiple dynamic target sets previously determined are used for matching, and the dynamic target with the highest similarity is selected as the dynamic target corresponding to the scanning point clustering result. The matched dynamic target is added to the corresponding dynamic target scanning result sequence. If the dynamic target appears for the first time in the scanning result at the current moment, a new entry is created in the scanning result sequence of the dynamic target.

[0068] By fitting and associating all the multiple scanning point clustering result set sequences with time series, multiple dynamic target scanning result sequence sets can be obtained, which means that after traversing the scanning point clustering result sets at all moments, each dynamic target should have a corresponding scanning result sequence. The multiple dynamic target scanning result sequence sets can be arranged in chronological order to form multiple dynamic target scanning result sequence sets.

[0069] Executing step A400, performing multiple dynamic target speed identifications based on the multiple dynamic target scan result sequences to obtain multiple dynamic target speed sets, and calculating multiple overall dynamic speeds; in one possible implementation, step A400 further includes step A410, extracting the multiple dynamic target scan result sequences according to a first step length to obtain multiple first dynamic target scan result sets; executing step A420, extracting the multiple dynamic target scan result sequences according to a second step length to obtain multiple second dynamic target scan result sets, wherein the second step length is greater than the first step length;

[0070] First, the first and second step sizes are set. The first step size is a shorter step size, used to extract a more detailed dynamic target scan result set; the second step size is a longer step size, used to extract a coarser dynamic target scan result set. Furthermore, starting from the starting point of the scan result sequence, the scan results of the dynamic target are sequentially extracted, using the first step size as an interval. The extracted scan results are combined into a new sequence to form the first dynamic target scan result set. If the length of the scan result sequence is not an integer multiple of the first step size, the last step size may be smaller than the first step size. In this case, the remaining scan results can be ignored or added to the last extracted scan result set.

[0071] Then, with the second step length as the interval, starting from the starting point of the scan result sequence, the scan results of the dynamic target are extracted in sequence, and the extracted scan results are combined into a new sequence to form the second dynamic target scan result. Similarly, if the length of the scan result sequence is not an integer multiple of the second step length, the remaining scan results can be ignored or added to the last extracted scan result set as needed.

[0072] Executing step A430, performing multiple dynamic target speed identification based on the multiple first dynamic target scanning result sets and the multiple second dynamic target scanning result sets to obtain multiple dynamic target speed sets;

[0073] In one possible implementation, step A430 further includes step A431, obtaining a sample dynamic target scanning result sequence set and a plurality of sample dynamic target speed sets based on historical perception data; executing step A432, processing the sample dynamic target scanning result set according to the first step length and the second step length to obtain a plurality of sample first dynamic target scanning result sets and a plurality of sample second dynamic target scanning result sets;

[0074] First, historical perception data is collected through the scanning results of radar, laser radar (LiDAR) or other sensors at different time points, and the scanning results related to dynamic targets are extracted from the historical perception data to form a sample dynamic target scanning result sequence set. Based on the target scanning result sequence set, the speed information of the corresponding dynamic target is calculated to form multiple sample dynamic target speed sets.

[0075] According to the pre-set first step length, starting from the starting point of the sequence, a scan result is extracted every first step length to form a new sequence, namely the sample first dynamic target scan result set. Similarly, according to the pre-set second step length (where the second step length is greater than the first step length), starting from the starting point of the sequence, a scan result is extracted every second step length to form another new sequence, namely the sample second dynamic target scan result set.

[0076] Execute step A433, based on the slowfast network, use the multiple sample first dynamic target scanning result sets, the multiple sample second dynamic target scanning result sets and the multiple sample dynamic target speed sets to train a dynamic target speed identifier; execute step A434, based on the dynamic target speed identifier, identify the multiple first dynamic target scanning result sets and the multiple second dynamic target scanning result sets to obtain multiple dynamic target speed sets.

[0077] Match multiple sample dynamic target speed sets with corresponding scan result sets to form a complete training sample set, where each sample includes a scan result and a corresponding target speed, and then construct a SlowFast network. The SlowFast network can include two branches: a Slow branch and a Fast branch. The Slow branch is used to capture spatial semantic information, while the Fast branch is used to capture temporal detail information. The preprocessed training sample set is used to train the SlowFast network model, and a loss function such as the mean square error (MSE) or the mean absolute error (MAE) is defined to measure the difference between the model-predicted speed and the actual speed. Multiple rounds of training are performed until the performance of the model on the validation set reaches a preset threshold or no longer improves. Finally, the performance of the trained dynamic target speed identifier is evaluated on the validation set, thereby outputting the dynamic target speed identifier.

[0078] Multiple first dynamic target scanning result sets and multiple second dynamic target scanning result sets are input into a trained dynamic target speed identifier to predict each scanning result set, output the corresponding dynamic target speed, and organize the prediction results into multiple dynamic target speed sets.

[0079] Execute step A440 to calculate the average dynamic target speeds of the multiple dynamic target speed sets to obtain multiple overall dynamic speeds.

[0080] Each speed set in the multiple dynamic target speed sets is traversed, and the speed value of each speed set is extracted and summed, and then the sum is divided by the number of data in the multiple dynamic target speed sets to obtain the speed values ​​and averages corresponding to the multiple dynamic target speed sets, and output multiple overall dynamic speeds.

[0081] Next, step A500 is executed to process and obtain multiple dynamic target densities based on the multiple dynamic target sets and the positioning information, and analyze and obtain multiple risk factors for multiple azimuth areas by combining the multiple overall dynamic speeds, multiple noise data, and multiple historical accident data.

[0082] In one possible implementation, step A500 further includes step A510, obtaining a number of dynamic targets based on the multiple dynamic target sets, and calculating a density of dynamic targets based on the road features of the multiple orientation areas obtained based on the positioning information.

[0083] By traversing each dynamic target set by calculating the number of elements in the set and counting the number of dynamic targets therein, the number of multiple dynamic targets is obtained. Furthermore, in combination with obtaining the road features of multiple orientation areas based on the positioning information, it means that the road length, area or other relevant parameters can be obtained from the geographic information system (GIS), the map database or the special road feature database, so as to obtain the road feature information of each orientation area. Finally, the number of targets in each dynamic target set is combined with the road features of the corresponding orientation area to calculate the density of multiple dynamic targets. The multiple dynamic target densities refer to the number of targets per unit road length or unit area.

[0084] Execute step A520 to obtain, based on historical road safety detection data, a sample dynamic target density set, a sample overall dynamic speed set, a sample noise data set, a sample historical accident data set, and a sample hazard coefficient set;

[0085] First, the positioning and speed data of all dynamic targets are extracted from historical road safety inspection data. On this basis, the number of dynamic targets in each area (such as road sections, intersections, etc.) is calculated. Then, combined with the road length or area of ​​each area, the dynamic target density of each area is calculated. Finally, the dynamic target density of all areas is summarized into a sample dynamic target density set for output.

[0086] Secondly, the speed data of all dynamic targets are extracted from the historical data. For each area or time period, the average speed of all dynamic targets is calculated, and the average speed values ​​of all dynamic targets are summarized into the overall dynamic speed set of the sample.

[0087] The historical road safety inspection data was then cleaned to remove obvious errors and incomplete items. Statistical methods, business rules, and expert judgment were then used to identify data points that deviated from the norm or could not be reasonably explained as noise. This noise data was then categorized, organized, and compiled into a sample noise data set.

[0088] All accident records are then extracted from historical road safety inspection data. The accident records may include the time, location, type (such as rear-end collision, side collision, etc.) of the accident, as well as information about the vehicles and people involved. The accident data is classified and organized to form a sample historical accident data set.

[0089] The calculation of the final risk factor can be defined based on specific needs and available data. Risk indicators can include accident frequency, accident severity (e.g., number of injuries, number of fatalities), dynamic target density, and speed variation. A weighting unit is also set to combine these risk indicators into a comprehensive risk factor. For each region or time period, a risk factor is calculated based on historical data and the set weighting unit. The risk factors for all regions are then aggregated into a sample risk factor set.

[0090] Execute step A530, respectively use the sample dynamic target density set, the sample overall dynamic speed set, the sample noise data set, and the sample historical accident data set, combined with the sample hazard coefficient set, to construct multiple hazard identification branches and obtain a hazard identifier; execute step A540, based on the hazard identifier, identify the multiple dynamic target densities, multiple overall dynamic speeds, multiple noise data, and multiple historical accident data, obtain multiple branch hazard coefficient sets, and calculate to obtain multiple hazard coefficients.

[0091] Multiple hazard identification branches can be constructed based on the sample dynamic target density set, the sample overall dynamic speed set, the sample noise data set, and the sample historical accident data set, respectively, in combination with the sample hazard coefficient set. This means that the sample dynamic target density set is used to train the dynamic target density branch by using machine learning or deep learning algorithms (such as random forests, neural networks, etc.). The dynamic target density branch takes the dynamic target density feature as input and outputs the hazard coefficient. Similarly, the sample overall dynamic speed set is used to train the overall dynamic speed branch with the overall dynamic speed feature as input and outputs the hazard coefficient. Secondly, the relevant data within the noise time period, that is, the sample noise data set (such as flow, speed, etc.) is used to train the noise data branch, and the output is the hazard coefficient. Then, the historical accident data branch is trained based on the sample historical accident data set (such as accident frequency, type, severity, etc.), and the output is the hazard coefficient. Further, a hazard identifier is constructed based on the dynamic target density branch, the overall dynamic speed branch, the noise data branch, and the historical accident data branch.

[0092] Based on the hazard identifier, the multiple dynamic target densities, multiple overall dynamic speeds, multiple noise data and multiple historical accident data are identified, which means that the sample hazard coefficient can be labeled first. For example, the probability of a dangerous event occurring at the same location under a set of sample dynamic target densities, sample overall dynamic speeds, sample noise data, and sample historical accident data is labeled as a hazard coefficient for training, that is, new dynamic target densities, overall dynamic speeds, noise data and historical accident data are preprocessed and feature extracted, and then the extracted features are respectively input into the corresponding hazard identification branches to obtain the hazard coefficient output by each branch. The hazard coefficient can be calculated by averaging or weighted averaging the hazard levels of multiple branches.

[0093] Optionally, weights can be set based on the ratios of multiple dynamic target densities, multiple overall dynamic speeds, multiple noise data, and multiple historical accident data to standard dynamic target density, standard overall dynamic speed, standard noise data, and standard historical accident data under standard safety conditions (e.g., the average number of historical accidents on multiple roads per unit time). A weighted calculation can be performed on these multiple ratios to obtain a risk factor, with the greater the risk factor, the higher the probability of a safety incident. The standard dynamic target density, standard overall dynamic speed, and standard noise data can be set based on the average dynamic target density, average overall dynamic speed, and average noise data under a non-congested, safe road condition. The weights for the dynamic target density ratio, the overall dynamic speed ratio, the noise data ratio, and the historical accident data ratio can be, for example, 0.3, 0.3, 0.2, and 0.2.

[0094] Finally, step A600 is executed to determine a key sensing module and multiple secondary sensing modules according to the multiple risk factors and switch them, and to adjust the sensing frequencies of the multiple secondary sensing modules.

[0095] In one possible implementation, step A600 further includes step A610, selecting a perception module corresponding to a direction area with the highest risk factor as a key perception module, and switching the other perception modules as secondary perception modules;

[0096] Since vehicle density, vehicle speed, noise, and historical accidents are all factors related to the degree of danger on the road, it is necessary to analyze the danger coefficients of multiple azimuth areas separately. Furthermore, the danger coefficients of each azimuth area calculated previously are sorted in descending order, and the azimuth area with the highest danger coefficient is selected as the azimuth area with the highest danger coefficient. At the same time, the perception module corresponding to the azimuth area with the highest danger coefficient is selected as the key perception module, and the perception frequency is increased. Except for the key perception module, all other perception modules are regarded as secondary perception modules. The secondary perception module adjusts the perception frequency according to the danger coefficient to improve the accuracy and speed of perception of the dangerous direction. When danger occurs, it can improve the user, such as voice reminders. The module switching can include static switching, dynamic switching, etc. Static switching refers to directly determining the key perception module and the secondary perception module based on the initial correspondence between the azimuth area and the perception module, and keeping this relationship unchanged. Dynamic switching refers to dynamically adjusting the key perception module and the secondary perception module based on the danger coefficient calculated in real time or updated regularly. When the risk factor of a certain location exceeds a certain threshold or changes significantly, the corresponding perception module can be switched to a priority perception module, and the original priority perception module may need to be downgraded to a secondary perception module. The entire method can be executed periodically, for example, once at each intersection.

[0097] Execute step A620 to set the perception frequency of the key perception module to the maximum perception frequency; execute step A630 to correct and calculate the preset perception frequency based on the ratio of the danger coefficients of the orientation areas of multiple other secondary perception modules and the danger coefficient of the orientation area of ​​the key perception module, obtain multiple secondary perception frequencies, and make adjustments.

[0098] First, the key sensing module's sensing frequency is set to the maximum to ensure the most intensive and timely monitoring of that area. For each secondary sensing module, the preset sensing frequency is modified based on the ratio of the risk factor of its corresponding area to the risk factor of the key sensing module's area. This risk factor ratio serves as a weighting factor to adjust the sensing frequency of the secondary sensing module. The system is then ensured to operate at the new sensing frequency, regularly reassessing the risk factor of each area and adjusting the sensing module's sensing frequency as needed. System performance and resource usage are monitored to ensure that adjustments to the sensing frequency do not overburden the system or affect its normal operation.

[0099] The embodiments of the present application solve the technical problem that the perception switching and response speed are affected by environmental interference, which leads to reduced user safety and efficiency. It achieves the technical effect of accurately perceiving the helmet module based on the scene, thereby improving user safety and efficiency.

[0100] In the above, refer to Figure 1 The scene-based intelligent helmet module perception switching method and system method according to the embodiment of the present invention are described in detail. Figure 2 The present invention describes a scenario-based intelligent helmet module perception switching system according to an embodiment of the present invention.

[0101] The scenario-based intelligent helmet module perception switching system according to an embodiment of the present invention is used to solve the technical problem that the perception switching and response speed are affected by environmental interference, which reduces user safety and efficiency. It achieves the technical effect of accurately perceiving the helmet module based on the scenario, thereby improving user safety and efficiency. The scenario-based intelligent helmet module perception switching system includes: a first data acquisition module 10, a positioning information acquisition module 20, a cluster analysis module 30, a speed recognition module 40, a first analysis module 50, and a switching module 60.

[0102] A first data acquisition module 10 is configured to collect radar scanning data sequences and noise data in multiple azimuth areas according to a preset sensing frequency through multiple sensing modules in the smart helmet, wherein the multiple sensing modules are arranged in a ring-shaped manner at multiple positions in the smart helmet;

[0103] A positioning information acquisition module 20, wherein the positioning information acquisition module 20 is configured to acquire positioning information through a smart helmet, and the plurality of perception modules collect a plurality of historical accident data within the plurality of azimuth areas according to the positioning information;

[0104] A cluster analysis module 30 is configured to perform cluster analysis on multiple radar scan data sequences to obtain multiple scan point cluster result set sequences, perform dynamic target fitting within multiple first scan point cluster result sets within a first moment to obtain multiple dynamic target sets, and associate the dynamic target sets with the multiple scan point cluster result set sequences to obtain multiple dynamic target scan result sequence sets;

[0105] A speed identification module 40 is configured to identify multiple dynamic target speeds based on multiple dynamic target scanning result sequences, obtain multiple dynamic target speed sets, and calculate multiple overall dynamic speeds;

[0106] A first analysis module 50 is configured to process and obtain multiple dynamic target densities based on the multiple dynamic target sets and the positioning information, and analyze and obtain multiple risk factors for multiple azimuth areas by combining the multiple overall dynamic speeds, multiple noise data, and multiple historical accident data;

[0107] The switching module 60 is used to determine the key sensing module and multiple secondary sensing modules according to the multiple risk factors and switch them, and adjust the sensing frequency of the multiple secondary sensing modules.

[0108] The specific configuration of the positioning information acquisition module 20 will be described in detail below. As described above, positioning information is acquired through the smart helmet, and the multiple perception modules collect multiple historical accident data within the multiple azimuth areas based on the positioning information. The positioning information acquisition module 20 may further include: a positioning unit for acquiring current positioning information through the positioning unit within the smart helmet; a first extraction unit for determining multiple target areas for historical accident data extraction based on the multiple azimuth areas and the positioning information; and a second extraction unit for extracting the number of accidents within the multiple target areas within a historical period to obtain multiple historical accident data.

[0109] The specific configuration of the cluster analysis module 30 will be described in detail below. As described above, cluster analysis is performed on multiple radar scanning data sequences to obtain multiple scanning point clustering result set sequences, dynamic target fitting is performed on multiple first scanning point clustering result sets within a first moment to obtain multiple dynamic target sets, and the dynamic target sets are associated with the multiple scanning point clustering result set sequences. The cluster analysis module 30 may further include: a second analysis unit for performing cluster analysis on multiple radar scanning data sequences, classifying scanning points with a distance less than a distance threshold into one cluster result, and obtaining multiple scanning point clustering result set sequences, each of which includes scanning point clustering result sets at multiple moments; a first indexing unit for indexing the first scanning point clustering result set based on the first scanning point clustering result set. The method comprises the following steps: indexing at each moment to obtain a plurality of first scanning point clustering result sets; a first fitting unit is used to obtain a dynamic target anchor frame set, iteratively perform dynamic target fitting on all scanning point clustering results in the plurality of first scanning point clustering result sets, obtain a dynamic target anchor frame with the largest fitting degree, and obtain a plurality of dynamic target sets, wherein the dynamic target anchor frame set includes all dynamic target anchor frames of a plurality of dynamic targets at different angles; and a second fitting unit is used to perform fitting association on the plurality of scanning point clustering result set sequences according to the plurality of dynamic target sets, and obtain a plurality of dynamic target scanning result sequence sets, wherein each dynamic target scanning result includes the plurality of dynamic targets.

[0110] The specific configuration of the first fitting unit will be described in detail below. As described above, a dynamic target anchor frame set is obtained, and iterative dynamic target fitting is performed on all scanning point clustering results within the multiple first scanning point clustering result sets to obtain a dynamic target anchor frame with the largest fit, thereby obtaining multiple dynamic target sets. The first fitting unit may further include: an anchor frame determination unit for obtaining a dynamic target anchor frame set based on all dynamic target anchor frames of multiple dynamic targets at different angles; a first frame selection unit for randomly selecting a first dynamic target anchor frame, frame-selecting all scanning point clustering results within the multiple first scanning point clustering result sets, and calculating multiple first fits based on the proportion of scanning points within each scanning point clustering result falling within the first dynamic target anchor frame; a second frame selection unit for again randomly selecting a second dynamic target anchor frame, frame-selecting all scanning point clustering results within the multiple first scanning point clustering result sets, and calculating multiple second fits; the second fitting unit is for continuing to perform iterative fitting, outputting the dynamic target corresponding to the dynamic target anchor frame with the largest fit in the iterative fitting for each scanning point clustering result, thereby obtaining multiple dynamic target sets.

[0111] The specific configuration of the speed identification module 40 will be described in detail below. As described above, based on multiple dynamic target scan result sequences, multiple dynamic target speed identifications are performed to obtain multiple dynamic target speed sets, and multiple overall dynamic speeds are calculated. The speed identification module 40 may further include: a first extraction unit for extracting the multiple dynamic target scan result sequences according to a first step length to obtain multiple first dynamic target scan result sets; a second extraction unit for extracting the multiple dynamic target scan result sequences according to a second step length to obtain multiple second dynamic target scan result sets, wherein the second step length is greater than the first step length; a first identification unit for performing multiple dynamic target speed identifications based on the multiple first dynamic target scan result sets and the multiple second dynamic target scan result sets to obtain multiple dynamic target speed sets; and a first calculation unit for calculating the dynamic target speed mean of the multiple dynamic target speed sets to obtain multiple overall dynamic speeds.

[0112] Below, the specific configuration of the first recognition unit will be described in detail. As described above, based on the multiple first dynamic target scanning result sets and the multiple second dynamic target scanning result sets, multiple dynamic target speed identifications are performed, and the first recognition unit may further include: a dynamic information acquisition unit for acquiring a sample dynamic target scanning result sequence set and a multiple sample dynamic target speed sets based on historical perception data; a data processing unit for processing the sample dynamic target scanning result set according to the first step length and the second step length to obtain a multiple sample first dynamic target scanning result set and a multiple sample second dynamic target scanning result set; a recognizer training unit for training a dynamic target speed recognizer based on a slowfast network using the multiple sample first dynamic target scanning result sets, the multiple sample second dynamic target scanning result sets and the multiple sample dynamic target speed sets; and a second recognition unit for recognizing the multiple first dynamic target scanning result sets and the multiple second dynamic target scanning result sets based on the dynamic target speed recognizer to obtain multiple dynamic target speed sets.

[0113] The specific configuration of the first analysis module 50 will be described in detail below. As described above, based on the multiple dynamic target sets, a plurality of dynamic target numbers are statistically obtained, and multiple dynamic target densities are calculated based on the road characteristics of the multiple azimuth areas obtained based on the positioning information. The first analysis module 50 may further include: a first calculation unit for statistically obtaining the plurality of dynamic target numbers based on the multiple dynamic target sets, and calculating multiple dynamic target densities based on the road characteristics of the multiple azimuth areas obtained based on the positioning information; a detection acquisition unit for acquiring a sample dynamic target density set, a sample overall dynamic speed set, a sample noise data set, a sample historical accident data set, and a sample hazard coefficient set based on historical road safety detection data; a branch construction unit for constructing multiple hazard identification branches using the sample dynamic target density set, the sample overall dynamic speed set, the sample noise data set, and the sample historical accident data set, in combination with the sample hazard coefficient set, to obtain a hazard identifier; and a third identification unit for identifying the multiple dynamic target densities, the multiple overall dynamic speeds, the multiple noise data, and the multiple historical accident data based on the hazard identifier to obtain multiple branch hazard coefficient sets and calculate multiple hazard coefficients.

[0114] The specific configuration of the switching module 60 will be described in detail below. As described above, based on the multiple risk factors, a key sensing module and multiple secondary sensing modules are determined and switched, and the sensing frequencies of the multiple secondary sensing modules are adjusted. The switching module 60 may further include: a key sensing unit for selecting the sensing module corresponding to the azimuth area with the largest risk factor as the key sensing module, and the other multiple sensing modules as multiple secondary sensing modules for module switching; a frequency setting unit for setting the sensing frequency of the key sensing module to the maximum sensing frequency; and a second calculation unit for performing a correction calculation on the preset sensing frequency based on the ratio of the risk factors of the azimuth areas of the other multiple secondary sensing modules to the risk factor of the azimuth area of ​​the key sensing module, to obtain multiple secondary sensing frequencies for adjustment.

[0115] The scenario-based smart helmet module perception switching system provided by the embodiment of the present invention can execute the scenario-based smart helmet module perception switching method and system method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0116] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other and are not used to limit the scope of protection of the present invention.

[0117] The above specific embodiments do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application shall be included within the scope of protection of this application.

Claims

1. A scene-based intelligent helmet module perception switching method, characterized in that: The method comprises: The radar scanning data sequences and noise data of multiple azimuth areas are collected according to a preset sensing frequency through multiple sensing modules in the smart helmet, wherein the multiple sensing modules are arranged in a ring shape at multiple positions in the smart helmet; Acquiring positioning information through the smart helmet, wherein the plurality of sensing modules collect a plurality of historical accident data in the plurality of azimuth areas according to the positioning information; Performing cluster analysis on multiple radar scanning data sequences to obtain multiple scanning point clustering result set sequences, performing dynamic target fitting in multiple first scanning point clustering result sets within a first moment to obtain multiple dynamic target sets, and associating them with the multiple scanning point clustering result set sequences to obtain multiple dynamic target scanning result sequence sets; According to the multiple dynamic target scanning result sequences, multiple dynamic target speed identifications are performed to obtain multiple dynamic target speed sets, and multiple overall dynamic speeds are calculated; According to the multiple dynamic target sets and the positioning information, multiple dynamic target densities are processed and obtained, and multiple risk factors of multiple azimuth areas are analyzed and obtained by combining the multiple overall dynamic speeds, multiple noise data and multiple historical accident data; According to the multiple risk factors, determining and switching a key sensing module and multiple secondary sensing modules, and adjusting the sensing frequencies of the multiple secondary sensing modules; According to the plurality of dynamic target sets and the positioning information, a plurality of dynamic target densities are processed and obtained, and a plurality of risk factors of a plurality of azimuth areas are analyzed and obtained by combining the plurality of overall dynamic speeds, a plurality of noise data and a plurality of historical accident data, including: According to the plurality of dynamic target sets, a plurality of dynamic target numbers are obtained by counting, and a plurality of dynamic target densities are obtained by calculating in combination with the road features of the plurality of orientation areas obtained according to the positioning information; Based on historical road safety detection data, obtain a sample dynamic target density set, a sample overall dynamic speed set, a sample noise data set, a sample historical accident data set and a sample risk factor set; The sample dynamic target density set, the sample overall dynamic speed set, the sample noise data set, and the sample historical accident data set are respectively used, combined with the sample hazard coefficient set, to construct multiple hazard identification branches to obtain a hazard identifier; Based on the hazard identifier, the multiple dynamic target densities, multiple overall dynamic speeds, multiple noise data and multiple historical accident data are identified to obtain multiple branch hazard coefficient sets, and multiple hazard coefficients are calculated.

2. The method according to claim 1, characterized in that The positioning information is obtained by the smart helmet, and the plurality of perception modules collect a plurality of historical accident data in the plurality of azimuth areas according to the positioning information, including: Obtain current positioning information through the positioning unit in the smart helmet; Determining multiple target areas for historical accident data extraction based on the multiple azimuth areas and positioning information; The number of accidents in the multiple target areas within a historical period is extracted to obtain multiple historical accident data.

3. The method according to claim 1, characterized in that Performing cluster analysis on multiple radar scanning data sequences to obtain multiple scanning point clustering result set sequences, performing dynamic target fitting in multiple first scanning point clustering result sets within a first moment to obtain multiple dynamic target sets, and associating them to the multiple scanning point clustering result set sequences, including: Perform cluster analysis on multiple radar scanning data sequences, classify scanning points whose distance is less than a distance threshold into one cluster result, and obtain multiple scanning point cluster result set sequences, each of which includes scanning point cluster result sets at multiple moments; Performing indexing based on the first moment to obtain a plurality of first scanning point clustering result sets; Acquire a dynamic target anchor frame set, perform iterative dynamic target fitting on all scanning point clustering results in the plurality of first scanning point clustering result sets, acquire a dynamic target anchor frame with the maximum fitting degree, and obtain a plurality of dynamic target sets, wherein the dynamic target anchor frame set includes all dynamic target anchor frames of a plurality of dynamic targets at different angles; According to the multiple dynamic target sets, fitting association is performed on the multiple scanning point clustering result set sequences to obtain multiple dynamic target scanning result sequence sets, each dynamic target scanning result includes the multiple dynamic targets.

4. The method according to claim 3, characterized in that Acquire a dynamic target anchor frame set, perform iterative dynamic target fitting on all scanning point clustering results in the plurality of first scanning point clustering result sets, acquire a dynamic target anchor frame with the maximum fitting degree, and obtain a plurality of dynamic target sets, including: Based on all dynamic target anchor frames of multiple dynamic targets at different angles, a dynamic target anchor frame set is obtained; Randomly selecting a first dynamic target anchor frame, selecting all scanning point clustering results in the plurality of first scanning point clustering result sets, and calculating and obtaining a plurality of first degrees of fit according to a proportion of scanning points in each scanning point clustering result falling into the first dynamic target anchor frame; Randomly selecting a second dynamic target anchor frame again, selecting all scanning point clustering results in the plurality of first scanning point clustering result sets, and calculating to obtain a plurality of second fitting degrees; Continue to perform iterative fitting, output the dynamic target corresponding to the dynamic target anchor frame with the largest fitting degree in the iterative fitting of each scanning point clustering result, and obtain multiple dynamic target sets.

5. The method according to claim 1, characterized in that According to the multiple dynamic target scanning result sequences, multiple dynamic target speed identifications are performed to obtain multiple dynamic target speed sets, and multiple overall dynamic speeds are calculated, including: According to the first step, extracting the plurality of dynamic target scanning result sequences to obtain a plurality of first dynamic target scanning result sets; extracting the plurality of dynamic target scanning result sequences according to a second step length to obtain a plurality of second dynamic target scanning result sets, wherein the second step length is greater than the first step length; According to the plurality of first dynamic target scanning result sets and the plurality of second dynamic target scanning result sets, multiple dynamic target speed identifications are performed to obtain multiple dynamic target speed sets; The dynamic target speed averages of the plurality of dynamic target speed sets are calculated to obtain a plurality of overall dynamic speeds.

6. The method according to claim 5, characterized in that According to the plurality of first dynamic target scanning result sets and the plurality of second dynamic target scanning result sets, a plurality of dynamic target speed identifications are performed, including: Based on the historical perception data, a sample dynamic target scanning result sequence set and a plurality of sample dynamic target speed sets are obtained; Processing the sample dynamic target scanning result set according to the first step length and the second step length to obtain a plurality of sample first dynamic target scanning result sets and a plurality of sample second dynamic target scanning result sets; Based on the slowfast network, the plurality of sample first dynamic target scanning result sets, the plurality of sample second dynamic target scanning result sets and the plurality of sample dynamic target speed sets are used to train a dynamic target speed identifier; Based on the dynamic target speed identifier, the plurality of first dynamic target scanning result sets and the plurality of second dynamic target scanning result sets are identified to obtain a plurality of dynamic target speed sets.

7. The method according to claim 1, characterized in that According to the multiple risk factors, determining a key sensing module and multiple secondary sensing modules and switching them, and adjusting the sensing frequencies of the multiple secondary sensing modules, including: The perception module corresponding to the direction area with the greatest risk factor is selected as the key perception module, and the other multiple perception modules are used as multiple secondary perception modules for module switching; Setting the perception frequency of the key perception module to the maximum perception frequency; According to the ratio of the danger coefficients of the orientation areas of the other multiple secondary perception modules to the danger coefficient of the orientation area of ​​the key perception module, the preset perception frequency is corrected and calculated to obtain multiple secondary perception frequencies for adjustment.

8. The scene-based intelligent helmet module perception switching system is characterized by: The system is used to implement the scenario-based smart helmet module perception switching method according to any one of claims 1 to 7, and the system includes: A first data acquisition module, the first data acquisition module is used to collect radar scanning data sequences and noise data of multiple azimuth areas according to a preset sensing frequency through multiple sensing modules in the smart helmet, wherein the multiple sensing modules are arranged in a ring shape at multiple positions in the smart helmet; A positioning information acquisition module, the positioning information acquisition module is used to acquire positioning information through a smart helmet, and the multiple perception modules collect multiple historical accident data in the multiple azimuth areas according to the positioning information; A cluster analysis module, the cluster analysis module is used to perform cluster analysis on multiple radar scanning data sequences to obtain multiple scanning point cluster result set sequences, perform dynamic target fitting in multiple first scanning point cluster result sets within a first moment to obtain multiple dynamic target sets, and associate them with the multiple scanning point cluster result set sequences to obtain multiple dynamic target scanning result sequence sets; A speed identification module, the speed identification module is used to identify multiple dynamic target speeds according to multiple dynamic target scanning result sequences, obtain multiple dynamic target speed sets, and calculate and obtain multiple overall dynamic speeds; A first analysis module, the first analysis module is used to process and obtain multiple dynamic target densities according to the multiple dynamic target sets and the positioning information, and analyze and obtain multiple risk factors of multiple azimuth areas in combination with the multiple overall dynamic speeds, multiple noise data and multiple historical accident data; A switching module is used to determine a key sensing module and multiple secondary sensing modules according to the multiple risk factors and switch them, and to adjust the sensing frequency of the multiple secondary sensing modules.

Images