Complex city environment sensing and collision early warning system based on intelligent network connection automobile

Abstract

The invention discloses a complex city environment sensing and collision early warning system based on an intelligent network connection automobile, and relates to the technical field of electric automobile control. The complex city environment perception and collision early warning system based on the intelligent connected automobile comprises an obstacle positioning identification analysis module, a data fusion synchronous analysis module and a collision risk early warning analysis module. According to the invention, in the obstacle sensing stage of the intelligent network connection platform, positioning identification analysis is carried out on the data sensing process to determine whether radar wave adjustment optimization is carried out, and then in the data processing stage, data fusion synchronous analysis is carried out to determine whether data capacity window optimization is carried out. And finally, collision risk assessment analysis is carried out in the collision risk early warning stage to judge whether filtering cache optimization is carried out or not. The problem that in the prior art, the collision early warning false alarm rate is increased due to the fact that corresponding obstacle sensing delay of the intelligent networked automobile in the peak period road section is caused is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electric vehicle control, and in particular to a complex urban environment perception and collision warning system based on intelligent connected vehicles. BACKGROUND

[0002] In the field of intelligent connected vehicles (ICV), the complex urban environment perception and collision warning system is one of the core technologies to ensure driving safety and improve driving experience. The urban environment has characteristics such as fast dynamic changes, multiple types of obstacles, and complex traffic rules, which puts high requirements on the perception and decision-making capabilities of vehicles. Intelligent connected vehicles integrate advanced sensors, big data processing, artificial intelligence algorithms, and other technologies to achieve accurate perception and risk assessment of the surrounding environment, and then issue warnings to the driver before potential collisions occur, or even automatically take evasive measures such as emergency braking or steering in emergency situations, thereby significantly reducing the incidence of traffic accidents.

[0003] Specifically, by collecting static or dynamic information in the urban environment through multiple source heterogeneous sensors, such as using a vehicle-mounted camera to collect visible light images on the driving section to detect vehicles, pedestrians, and traffic signs and warning signs, and performing regional division of the road area; using a millimeter wave radar to detect obstacles such as range, speed, and angle, and converting the complex urban environment around the vehicle into a three-dimensional point cloud form presented in the intelligent connected platform; when the vehicle is close to a point, using an ultrasonic radar to detect obstacles in close proximity, such as the distance between vehicles before and after a traffic jam period, and a sudden pedestrian crossing the road; on the data fusion, under the premise of ensuring the integrity of the multi-source data stream, the obtained urban environment data is matched and fused, and the recognition probability is evaluated; on risk assessment, the collision risk is calculated based on the obtained environment data through a reinforcement learning model, and the corresponding prepared warning strategy is output in the model, and the warning mode is selected according to the collision risk calculation result and the risk level at this time (such as sound, visual, and touch prompts, and if necessary, the platform can be forced to intervene, such as forced braking).

[0004] For example, the Chinese invention patent with publication number CN118182510B discloses a driving safety warning system suitable for intelligent connected vehicles, which includes a driver evaluation module that monitors the behavior and physical state of the driver in the vehicle during the detection period, an environment perception module that analyzes the road risk conditions of the perceived section, and captures and analyzes the abnormal operating state of the vehicle during driving, and sends the corresponding alarm information to the driving warning end based on the generated driver warning signal, road environment perception warning signal, or vehicle operating abnormality warning signal.

[0005] For example, the Chinese invention patent with publication number CN118387135B discloses an intelligent networked vehicle automatic driving detection control method and its control system, which includes: continuously reinforcing learning planning target trajectory according to steering decision information and lane information, and judging whether to adjust the target trajectory according to the obtained front wheel steering angle loss value after the regulation of reinforcement learning, and combining the front wheel steering angle loss to adjust the planned target trajectory in time during trajectory planning.

[0006] The above-mentioned technology at least has the following technical problems: In the prior art, during the peak period, the traffic flow increases by several times, the road is densely filled with a large number of vehicles, and pedestrians, non-motor vehicles and other obstacles also frequently shuttle therein, and their motion trajectories are complex and variable, and are rapidly dynamically adjusted. Intelligent networked vehicles rely on multiple sensors to perceive the surrounding environment. In the face of such a complex scene, the sensors need to collect a large amount of data, which undoubtedly brings great pressure to the data collection link, increasing the time consumption of the collection process. After the data collection is completed, the data needs to be transmitted to the vehicle-mounted computing platform through the network. A large number of vehicles simultaneously transmit data during the peak period, which easily causes network congestion, and the data may appear queuing, packet loss and other situations during the transmission process, further aggravating the transmission delay. After receiving the data, the vehicle-mounted computing platform needs to fuse and process the multi-source heterogeneous data for analysis and identification to obtain accurate environmental information and perform collision risk assessment. The huge amount of data makes the processing task of the computing platform heavy, and the processing time becomes longer. This series of delays causes the environmental information obtained by the system to deviate from the actual scene, which easily causes misjudgment when assessing the collision risk, and misjudges the safe scene as a dangerous scene, thereby triggering the collision warning, and increasing the false alarm rate. There is a problem of increasing the collision warning false alarm rate caused by the delay of the corresponding obstacle perception of the intelligent networked vehicle in the peak period. SUMMARY

[0007] In order to solve the technical problem of increasing the collision warning false alarm rate caused by the delay of the corresponding obstacle perception of the intelligent networked vehicle in the peak period in the prior art, the embodiments of the present application provide a complex urban environment perception and collision warning system based on intelligent networked vehicles. The technical solution is as follows: A complex urban environment perception and collision warning system based on an intelligent connected vehicle is provided, and the system comprises: an obstacle positioning identification analysis module for performing positioning identification analysis on an obstacle perception process of an intelligent connected platform according to obtained obstacle perception data, to determine whether to perform radar wave adjustment optimization, the positioning identification analysis being used to quantify the position perception accuracy of the intelligent connected platform in the intelligent connected vehicle on obstacles in a collision risk area, and the radar wave adjustment optimization being used to improve the perception ability of the intelligent connected platform on obstacles; a data fusion and synchronization analysis module for performing data fusion and synchronization analysis on an obstacle data processing process of the intelligent connected platform according to obtained obstacle synchronization identification time, to determine whether to perform data capacity window optimization, the data fusion and synchronization analysis being used to quantify the time alignment degree of the multi-source sensor in the intelligent connected platform in processing obstacle data, and the data capacity window optimization being used to improve the fusion and synchronization ability of the intelligent connected platform on obstacle data; and a collision risk warning module for performing collision risk evaluation analysis on an obstacle motion trajectory of the intelligent connected platform according to obtained collision risk prediction data, to determine whether to perform filter cache optimization, the collision risk evaluation analysis being used to quantify the prediction accuracy of the intelligent connected platform on the collision risk of the intelligent connected vehicle and obstacles, and the filter cache optimization being used to improve the accuracy of collision risk prediction and the reliability of collision risk warning.

[0008] The technical scheme provided by the embodiment of the application has at least the following beneficial effects: 1. The obstacle edge feature extraction capability is improved, and the positioning resolution of low-speed micro obstacles is improved by performing positioning identification analysis on the obstacle perception process of the intelligent connected platform and determining whether to perform radar wave adjustment optimization according to the analysis result; the continuity of obstacle motion trajectory prediction is ensured by performing data fusion and synchronization analysis on the processing process in the obstacle data processing stage and determining whether to perform data capacity window optimization according to the analysis result; and the trajectory prediction mutation caused by noise interference is smoothed by performing collision risk evaluation analysis on the obstacle motion trajectory in the collision risk warning stage and determining whether to perform filter cache optimization according to the analysis result.

[0009] 2. The feature suppression effect caused by ignoring the dimension mismatch in the traditional method is avoided, and the cumulative problem of hysteresis error caused by linear superposition is overcome by fusing the trajectory prediction delay time and the ranging delay time into a collision risk prediction index through harmonic mean; compared with the traditional fixed weight fusion method, the feedforward compensation ability of the system on the two types of delay is improved through the dynamic balance mechanism, so that the collision risk prediction index has the ability of dynamic alignment deviation while ensuring real-time.

[0010] 3. By fusing the coupling analysis mechanism of the antenna array response time length, the multi-path interference ranging time length and the radar receiving response time length, compared with the existing technology for the management of multi-path scattering noise and response inconsistency interference, the method weakens the time delay mismatch problem caused by traditional perception through the influence of environmental correction value and inherent characteristics on characteristic value; compared with the conventional linear superposition compensation strategy, the three-level difference coupling mechanism not only can suppress the path tracking dispersion effect caused by strong reflection objects, but also can strengthen the adaptive compensation ability in dynamic scenes through hierarchical weight distribution, and in typical complex scenes such as dense pedestrian flow, the risk of motion target trajectory sticking and feature point matching error caused by the traditional dimensionless method is avoided. BRIEF DESCRIPTION OF DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0012] Figure 1 The structural schematic diagram of the complex urban environment perception and collision warning system based on intelligent networked vehicles provided by the embodiments of the present application is shown in the figure. Figure 2 The execution flowchart of the obstacle positioning and identification analysis module provided by the embodiments of the present application is shown in the figure. Figure 3 The execution flowchart of the data fusion and synchronization analysis module provided by the embodiments of the present application is shown in the figure. Figure 4 The execution flowchart of the collision risk warning analysis module provided by the embodiments of the present application is shown in the figure. DETAILED DESCRIPTION

[0013] The technical solutions in the present application will be described below with reference to the drawings.

[0014] In the embodiments of the present application, the words such as "example", "for example" are used to represent as an example, illustration or description. Any embodiment or design scheme described as "example" in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. In fact, the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present application, the meaning expressed by "and / or" can be both, or can be one of the two.

[0015] In the embodiments of the present application, "image" and "picture" can be used interchangeably, and it should be pointed out that their meanings are consistent when their differences are not emphasized. "Of", "corresponding" and "corresponding" can be used interchangeably, and it should be pointed out that their meanings are consistent when their differences are not emphasized.

[0016] In order to make the technical problems, technical solutions and advantages to be solved by the present application clearer, the following will be described in detail in combination with the drawings and specific embodiments.

[0017] The embodiments of the present application provide a complex urban environment perception and collision warning system based on intelligent connected vehicles, as shown in Figure 1 The structure diagram of the complex urban environment perception and collision warning system based on intelligent connected vehicles, the system comprises the following modules: the obstacle positioning identification analysis module is used for positioning identification analysis of the obstacle perception process of the intelligent connected platform in the obstacle perception stage of the intelligent connected platform, according to the obtained obstacle perception data, to judge whether to perform radar wave adjustment optimization, the positioning identification analysis is used for quantifying the position perception accuracy of the intelligent connected platform in the intelligent connected vehicle to the obstacles in the collision risk area, the radar wave adjustment optimization is used for improving the perception ability of the intelligent connected platform to the obstacles; the data fusion synchronization analysis module is used for data fusion synchronization analysis of the obstacle data processing process of the intelligent connected platform in the obstacle data processing stage of the intelligent connected platform, according to the obtained obstacle synchronization identification time, to judge whether to perform data capacity window optimization, the data fusion synchronization analysis is used for quantifying the time alignment degree of the multi-source sensor in the intelligent connected platform to process the obstacle data, the data capacity window optimization is used for improving the fusion synchronization ability of the intelligent connected platform to the obstacle data; the collision risk warning module is used for collision risk assessment analysis of the obstacle motion trajectory of the intelligent connected platform in the collision risk warning stage of the intelligent connected platform, according to the obtained collision risk prediction data, to judge whether to perform filter cache optimization, the collision risk assessment analysis is used for quantifying the prediction accuracy of the intelligent connected platform to the collision risk of the intelligent connected vehicle and the obstacle, and the filter cache optimization is used for improving the accuracy of the collision risk prediction and the reliability of the collision risk warning.

[0018] In the present embodiment, the obstacle perception stage focuses on improving the perception ability, through the obstacle positioning identification analysis, the position perception precision of the platform to the obstacles in the collision risk area is quantified, and it is accurately judged whether the radar wave adjustment optimization is needed. This stage effectively solves the positioning deviation problem caused by the obstacle shielding and signal interference in the urban environment, and the radar wave coverage range and resolution are significantly enhanced after optimization, which reduces the spatial position perception error of the vehicle to the surrounding obstacles, and provides reliable basic data for subsequent trajectory prediction.

[0019] The data processing stage enhances the data fusion efficiency, synchronously analyzes the time alignment of multi-source sensors (such as cameras and millimeter wave radars), and dynamically adjusts the data capacity window. This stage breaks through the bottleneck of asynchronous timing of heterogeneous sensor data, ensures the real-time of dynamic parameters such as obstacle speed and direction, and avoids false alarms caused by data misplacement.

[0020] The collision warning stage improves the reliability of risk judgment, collision risk assessment analysis quantifies the collision prediction accuracy, and filtering cache optimization enhances the stability of trajectory prediction by suppressing noise signals. This stage shortens the response time of the system to sudden collision scenarios and reduces the false alarm rate.

[0021] This example systematically explains the advantages of the system from three aspects: function implementation, technical breakthrough, and application effect. In the obstacle perception stage, the positioning analysis and radar optimization improve the accuracy of position perception. In the data processing stage, synchronous analysis and window adjustment break through the timing bottleneck and ensure the real-time of dynamic parameters. In the collision warning stage, with the help of evaluation analysis and filtering optimization, the response time is shortened and the false alarm rate is reduced, finally realizing the efficient and reliable cooperation of the whole chain of "perception-processing-warning".

[0022] Further, the obstacle perception process of the intelligent networked platform is positioned and identified according to the obtained obstacle perception data. The specific steps are as follows: at the end of the obstacle perception period, the obtained obstacle perception data and the pre-set obstacle perception data in the database are compared in terms of difference, and the difference comparison results are corrected by combining the obstacle perception data correction value to obtain the obstacle perception data score, and the coupling processing is performed to obtain the obstacle perception influence index.

[0023] Among them, the obstacle perception data includes the antenna array response time, the multipath interference ranging time, and the radar receiving response time, the obstacle perception data correction value includes the antenna array response time correction value, the multipath interference ranging time correction value, and the radar receiving response time correction value, the obstacle perception data score includes the antenna array response time score, the multipath interference ranging time score, and the radar receiving response time score, the pre-set obstacle perception data includes the pre-set antenna array response time, the multipath interference ranging time, and the radar receiving response time, and the obstacle perception influence index is used to quantify the influence degree of the obstacle perception data on the perception ability of the intelligent networked platform.

[0024] Specifically, the antenna array response duration is used to reflect the time required for the antenna array to respond to the signal from receiving to responding during the process of perceiving the obstacle. The shorter the response duration, the faster the antenna array can respond to the obstacle signal, which helps to obtain the position, speed and other information of the obstacle in time. The multipath interference ranging duration is used to reflect the time required for the intelligent network platform to complete accurate ranging operation in the multipath interference environment (the signal encounters obstacles during transmission and is reflected, refracted, etc., resulting in multiple signals received at the receiving end). Multipath interference increases the complexity and uncertainty of ranging. This duration reflects the ability of the intelligent network platform to overcome multipath interference for accurate ranging. The radar receiving response duration is used to reflect the time from signal receiving state to signal processing and giving effective response after the millimeter wave radar receives the signal reflected by the obstacle. The millimeter wave radar has high resolution and good anti-interference ability. Its receiving response duration directly affects the speed and accuracy of the intelligent network platform in perceiving the detailed features of the obstacle.

[0025] The specific expression of the antenna array response duration score W1 is: In the formula, W1 represents the antenna array response duration score corresponding to the intelligent network platform at the end of the obstacle perception period, C1 represents the antenna array response duration correction value, W AR represents the antenna array response duration corresponding to the intelligent network platform at the end of the obstacle perception period, W AR1 represents the preset antenna array response duration.

[0026] The specific expression of the multipath interference ranging duration score W2 is: In the formula, W2 represents the multipath interference ranging duration score corresponding to the intelligent network platform at the end of the obstacle perception period, C2 represents the multipath interference ranging duration correction value, W MI represents the multipath interference ranging duration corresponding to the intelligent network platform at the end of the obstacle perception period, W MI1 represents the preset multipath interference ranging duration.

[0027] The specific expression of the radar receiving response duration score W3 is: In the formula, W3 represents the radar receiving response duration score corresponding to the intelligent network platform at the end of the obstacle perception period, C3 represents the radar receiving response duration correction value, W MR represents the radar receiving response duration corresponding to the intelligent network platform at the end of the obstacle perception period, W MR1 represents the preset radar receiving response duration.

[0028] The specific expression of the obstacle perception influence index P D is: In the formula, P DThe intelligent network connection platform corresponds to the obstacle perception influence index at the end of the obstacle perception period.

[0029] In the embodiment, the antenna array response duration, the multipath interference ranging duration and the radar receiving response duration are obtained by the timer built in the intelligent network connection platform. The preset antenna array response duration, the preset multipath interference ranging duration and the preset radar receiving response duration are respectively represented by the results obtained by summing and averaging the historical antenna array response duration, the historical multipath interference ranging duration and the historical radar receiving response duration at the end of the historical obstacle perception period of the historical intelligent network connection platform. The antenna array response duration correction value, the multipath interference ranging duration correction value and the radar receiving response duration correction value are the influence degrees of the antenna array response duration, the multipath interference ranging duration and the radar receiving response duration on the obstacle perception influence index obtained in the database. In the formula, the database stores the corresponding correction values for each parameter, and the value ranges of the three correction values are all 0-1, and the sum is equal to 1. In actual application, the real-time antenna array response duration, the real-time multipath interference ranging duration and the real-time radar receiving response duration can be input to obtain the corresponding correction values, thereby providing accurate quantitative data for evaluating the obstacle perception influence index.

[0030] It should be noted that the obstacle perception influence index increases with the increase of the antenna array response duration, the multipath interference ranging duration and the radar receiving response duration. The antenna array response duration is prolonged with the increase of the multipath interference ranging duration, because more time is needed for dynamic beamforming optimization to suppress mixed reflection signals. The radar receiving response duration is simultaneously restricted by the first two, and when the antenna array beam is lowered, the response duration will increase. The calculation amount of the multipath interference increases, which increases the multipath interference ranging duration. Meanwhile, the radar needs to ensure the signal-to-noise ratio of the echo signal, thereby prolonging the process duration of the received signal.

[0031] Considering the above mutual influence mechanism, the relationship between the obstacle perception influence index and each variable can be more comprehensively understood. The accurate evaluation of these relationships in the obstacle perception influence data acquisition process is important. By optimizing the antenna array response duration, the multipath interference ranging duration and the radar receiving response duration, the obstacle perception influence can be effectively reduced, the accuracy and stability of the obstacle perception can be improved, the obstacle perception data obtained is more in line with the actual effect, and the problem that the intelligent network connection platform cannot reliably perceive obstacles is effectively solved.

[0032] As Figure 2As shown, the execution flowchart of the obstacle positioning identification analysis module provided by the embodiment of the present application is shown, if the acquired obstacle perception influence index is not greater than the preset value in the database, data fusion synchronous analysis is performed, if it is greater, the radar wave emission power is adjusted and the obstacle perception influence index is output, it is judged whether the newly acquired index is not greater than the preset value in the database, if not, data fusion synchronous analysis is performed, if yes, the beam width is adjusted and the newly acquired index is compared with the preset index in the database, if it is greater than the preset value in the database, obstacle perception risk warning is performed, if it is not greater than, data fusion synchronous analysis is performed.

[0033] It is further understood that whether to perform radar wave adjustment optimization, the specific steps are: comparing the acquired obstacle perception influence index with the preset obstacle perception influence index in the database: if the acquired obstacle perception influence index is greater than the preset obstacle perception influence index in the database, it is recorded as unqualified obstacle perception and radar wave adjustment optimization is performed, which means that the radar wave emission power and the beam width are adjusted to improve the intelligent network platform's perception response to obstacles; if the acquired obstacle perception influence index is not greater than the preset obstacle perception influence index in the database, it is recorded as qualified obstacle perception and data fusion synchronous analysis is performed.

[0034] Among them, the radar wave adjustment optimization, the specific steps are: based on the radar wave emission power adjustment value obtained by mapping the deviation of the acquired obstacle perception influence index in the database, which is used to quantify the specific degree of increasing the radar wave intensity at present, so as to reduce the obstacle perception failure probability caused by signal interference, the deviation of the obstacle perception influence index represents the difference between the acquired obstacle perception influence index and the preset obstacle perception influence index in the database, the preset obstacle perception influence index is represented by the result of summing and averaging the historical obstacle perception influence index of the intelligent network platform at the end of the obstacle perception period, after completing the radar wave emission power adjustment, the intelligent network platform will send the emission power verification instruction to the preset personnel, which is used to prompt the preset personnel to perform positioning identification verification test based on the radar wave emission power after the radar wave emission power adjustment; after the positioning identification verification test is completed, the obstacle perception process is performed again, if the newly acquired obstacle perception influence index is greater than the preset obstacle perception influence index, obstacle perception warning is performed, otherwise, the radar wave emission power adjustment is completed and data fusion synchronous analysis is performed.

[0035] It needs to be understood that the radar wave adjustment optimization also includes beam width optimization, and the specific steps are: based on the deviation of the re-acquired obstacle perception influence index at the end of the re-performed obstacle perception process, the beam width adjustment value is mapped in the database, which is used to quantify the degree of current need to reduce the beam width to reduce the signal mixed interference caused by dense obstacles; after adjusting the beam width, if the re-acquired obstacle perception influence index is not greater than the preset obstacle perception influence index in the database, the radar wave adjustment optimization is completed and data fusion synchronization analysis is performed, otherwise obstacle perception risk warning is performed.

[0036] In the embodiment, the adaptive power control algorithm is used to increase the power intensity and beam width in time based on the acquired current transmission power and beam width, improve the intelligent network connection platform's perception ability of obstacles, until the obstacle perception is qualified, and the historical power intensity and beam width are input into the reinforcement learning model as sample data, and the transmission power-beam width reinforcement learning model is obtained based on the adaptive power control algorithm through training, and the acquired power intensity and beam width are input into the transmission power-beam width reinforcement learning model to output the adjustment value of the corresponding transmission power and beam width.

[0037] The example enhances the signal penetration and anti-interference ability, compensates for the perception signal attenuation caused by sudden changes in obstacle material or environmental noise, and reduces the false detection and missed detection probability; the beam width optimization suppresses the delay caused by noise signal interference, improves the target resolution accuracy and boundary recognition clarity in the dense obstacle group, ensures the accurate matching of the power and beam parameter adjustment amount, and avoids signal saturation or blind area coverage caused by over-optimization.

[0038] As shown in Figure 3 The execution flowchart of the data fusion synchronization analysis module provided by the embodiment of the application is shown in the figure, whether the acquired obstacle synchronization identification time is greater than the preset value is judged, if not, collision risk assessment analysis is performed, if yes, the obstacle synchronization identification time is obtained through the data compression ratio adjustment value, and the same comparison with the preset value in the database is performed, if not satisfied, the buffer capacity size is adjusted and the obstacle synchronization identification time is re-acquired, whether the obtained value meets the preset requirement is judged, if not satisfied, the parallel processing window size is adjusted, whether the obstacle synchronization identification time obtained on this basis is greater than the preset value in the database is judged, if yes, data fusion warning is performed, if satisfied, collision risk assessment analysis is performed.

[0039] It is further understood that whether to perform data capacity window optimization is determined by comparing the obtained obstacle synchronous identification duration with the preset obstacle synchronous identification duration in the database. The obstacle synchronous identification duration represents the time length from when the intelligent network connection platform receives the environment perception data to when the obstacle positioning is completed by fusion, which is obtained by an internal timer. If the obtained obstacle synchronous identification duration is greater than the preset obstacle synchronous identification duration in the database, it is recorded as unqualified synchronous identification and data compression optimization is performed. Data compression optimization means that the data compression ratio is adjusted to improve the intelligent network connection platform's identification response to the obstacle position. If the obtained obstacle synchronous identification duration is not greater than the preset obstacle synchronous identification duration in the database, it is recorded as qualified synchronous identification and collision risk assessment analysis is performed.

[0040] The data compression optimization specifically includes the following steps: based on the obtained obstacle synchronous identification duration deviation, a corresponding data compression ratio adjustment value is mapped in the database to reduce the identification deviation caused by environment perception data ambiguity. The obstacle synchronous identification duration deviation represents the difference between the obtained obstacle synchronous identification duration and the preset obstacle synchronous identification duration in the database. The preset obstacle synchronous identification duration is represented by the result of summing and averaging the historical obstacle synchronous identification duration at the end of the historical intelligent network connection platform's obstacle data processing period. After the data compression ratio is adjusted, if the newly obtained obstacle synchronous identification duration is greater than the preset obstacle synchronous identification duration in the database, capacity window optimization is performed. Otherwise, data capacity window optimization is completed and collision risk assessment analysis is performed.

[0041] The capacity window optimization specifically includes the following steps: based on the obstacle synchronous identification duration deviation obtained after data compression optimization, a buffer capacity adjustment value is mapped in the database to quantify the degree to which the buffer capacity needs to be increased, so as to improve the integrity of the intelligent network connection platform's identification data and thus improve the identification accuracy. After the buffer capacity adjustment is completed, if the newly obtained obstacle synchronous identification duration is not greater than the preset obstacle synchronous identification duration in the database, data capacity window optimization is completed. Otherwise, based on the obstacle synchronous identification duration deviation, a corresponding parallel processing window adjustment value is mapped in the database to quantify the degree to which the parallel processing window size needs to be increased, so as to reduce the task queuing delay while increasing the processing rate of multiple parallel threads. After the parallel processing window adjustment, if the newly obtained obstacle synchronous identification duration is greater than the preset obstacle synchronous identification duration in the database, data fusion warning is performed. Otherwise, data capacity window optimization is completed and collision risk assessment analysis is performed.

[0042] In the embodiment, the statistical optimization algorithm is used to adaptively adjust the data compression ratio based on the obtained current data compression ratio until the synchronization identification is qualified, and the historical data compression ratio is taken as sample data and input into the convolutional neural network for training to obtain a data compression-neural network model. The obtained data compression ratio is input into the data compression-neural network model to output a corresponding data compression ratio adjustment value.

[0043] Similarly, the time series prediction control algorithm is used to timely increase the capacity and window size based on the obtained current buffer capacity size and parallel processing window size until the synchronization identification is qualified, and the historical buffer capacity size and parallel processing window size are input into the convolutional neural network for training based on the time series prediction control algorithm to obtain a buffer capacity-parallel processing neural network model. The obtained buffer capacity size and parallel processing window size are input into the model to output a corresponding adjustment value.

[0044] The example reduces the processing resources occupied by non-key data (such as the sky) by compressing data transmission and analysis delay. Secondly, the buffer expansion mechanism balances the data integrity constraint and real-time requirement, ensures the alignment of multiple data streams in time delay through elastic memory allocation, and improves the continuity and smoothness of the obstacle space extraction. The dynamic expansion of the parallel processing window optimizes the scheduling of multi-modal data processing tasks, reduces the queuing delay, and improves the real-time response speed in high-density obstacle scenarios. The adaptive matching of the data fusion load fluctuation of the complex urban environment ensures that the synchronization identification time is stably lower than the preset value, and provides high-confidence time synchronization feature input for collision risk assessment.

[0045] Further, the collision risk prediction data is obtained, and the obstacle motion trajectory of the intelligent networked platform is analyzed for collision risk assessment. Specifically, the obtained trajectory prediction delay time is compared with the preset trajectory prediction delay time in the database to obtain a trajectory prediction delay time score, and the obtained ranging delay time is compared with the preset ranging delay time in the database to obtain a ranging delay time score. The ranging delay time score and the trajectory prediction delay time score are harmonically averaged to obtain a collision risk prediction index. The collision risk prediction index is used to quantify the influence of the collision risk prediction data on the risk warning capability of the intelligent networked platform.

[0046] The collision risk prediction data includes trajectory prediction delay and distance measurement delay, both acquired through a timer built into the intelligent connected platform. The trajectory prediction delay represents the delay in predicting the future trajectory based on the target's historical state (such as position, speed, acceleration, and direction of motion). The shorter this delay, the lower the delay in the intelligent connected platform's ability to predict the activity state of obstacles in subsequent periods, which is beneficial for the platform to react quickly. The distance measurement delay represents the delay in measuring the distance between the obstacle and the vehicle. The shorter this delay, the lower the delay in the intelligent connected platform's measurement of the distance between the obstacle and the vehicle, which is beneficial for the platform's risk assessment.

[0047] In this embodiment, the preset trajectory prediction delay and ranging delay are represented by the sum of the historical trajectory prediction delay and ranging delay at the end of the collision risk warning period of the historical intelligent connected platform. In this example, the increase in trajectory prediction delay will cause the cumulative error of motion state to rise, requiring a longer ranging time to complete trajectory verification, resulting in an extension of ranging delay. At the same time, the aggravation of ranging delay reduces the update frequency of original distance information, and the trajectory prediction model needs to rely on more historical data interpolation to compensate for the lack of real-time performance, further amplifying the prediction delay.

[0048] The assessment of trajectory prediction delay score enhances the sensitivity to sudden lane changes or rapidly decelerating obstacles by shortening the time loss between historical feature extraction and future trajectory generation; the assessment of ranging delay score reduces data processing redundancy in distance measurement, enhancing the timeliness and continuity of dynamic obstacle ranging data; the coupled collision risk prediction index reflects the comprehensive risk situation; it can quickly capture urgent collision threats and accurately identify the evolution trend of potential risks.

[0049] like Figure 4 The diagram shows the execution flowchart of the collision risk warning analysis module provided in this embodiment of the invention. It determines whether the obtained collision risk prediction index meets the requirements, i.e., the obtained value is greater than the preset value. If it meets the requirements, a collision risk warning is issued. If it does not meet the requirements, the mean filtering intensity is adjusted and it is determined whether the collision risk prediction index obtained on this basis meets the conditions. If it does not meet the requirements, the historical data cache length is adjusted and it is determined whether the new index is greater than the preset value in the database. If it is greater than the preset value, a collision risk warning is issued. If it is not greater than the preset value, a "No collision risk instruction" is issued.

[0050] It is further understood that whether to perform filter cache optimization is determined by: if the obtained collision risk prediction index is greater than the preset collision risk prediction index in the database, it is determined that there is a collision risk and a collision risk warning is performed, otherwise, mean filter strength optimization is performed, which is specifically: based on the obtained collision risk prediction index deviation, a mean filter strength adjustment value is obtained by mapping in the database, which is used to quantify the degree of need to increase the Kalman filter strength to reduce the target prediction deviation amplitude due to noise interference. After adjusting the filter strength, if the re-obtained collision risk prediction index is greater than the preset collision risk prediction index in the database, a collision risk warning is performed, otherwise, the cache length is adjusted. The collision risk prediction index deviation represents the difference between the obtained collision risk prediction index and the preset collision risk prediction index in the database. The preset collision risk prediction index is represented by the result of summing and averaging the historical collision risk prediction index at the end of the historical intelligent network connection platform collision risk warning period.

[0051] The cache length adjustment is specifically: based on the re-obtained collision risk prediction index deviation after adjusting the filter strength, a read historical data cache length adjustment value is obtained by mapping in the database, which is used to quantify the current need to increase the historical cache length to reduce the deviation error caused by short-term data fluctuations. After adjusting the cache length, if the re-obtained collision risk prediction index is greater than the preset collision risk prediction index in the database, a collision risk warning is performed, otherwise, a currently no collision risk instruction is sent.

[0052] In this embodiment, the filter strength and the cache length are increased in time based on the obtained current obstacle motion state and the cache read length by the trajectory prediction algorithm, until the collision risk prediction is completed. At the same time, the historical cache length and the filter strength are input as sample data into the recurrent neural network to train the filter strength-cache length neural network model based on the trajectory prediction algorithm. The obtained cache length and filter strength are input into the filter strength-cache length neural network model to output the corresponding cache length and filter strength adjustment value.

[0053] It is understood that sensors in complex urban environments (such as peak period road sections) are easily affected by noise, shielding or short-term data fluctuations, resulting in instantaneous value being too high (false alarm) or too low (missed alarm). The core purpose of this step of optimization is to solve the problems of "instantaneous value misjudgment" and "data fluctuation interference" in the conventional collision risk judgment. Through the "filter-cache" double-layer optimization closed loop, both real-time performance and prediction stability are improved, and ultimately the collision risk assessment is realized from instantaneous value judgment to spatiotemporal continuity analysis, thereby effectively reducing the collision warning false alarm rate of intelligent network connected vehicles in peak period road sections.

[0054] It should be understood that the memory in the embodiments of the present application can be a volatile memory or a nonvolatile memory, or can include both volatile and nonvolatile memory. Among them, the nonvolatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically EPROM (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM) used as an external cache. By way of example, and not limitation, a number of forms of random access memory (RAM) can be used, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchlink DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0055] The above-described embodiments can be implemented in whole or in part by software, hardware (such as a circuit), firmware, or any combination thereof. When implemented in software, the above-described embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are wholly or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center through a wired (such as infrared, wireless, microwave, etc.) manner. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. containing one or more available medium collections. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state disk.

[0056] It should be understood that the term "and / or" herein merely describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can represent the following three cases: A exists alone, A and B exist together, and B exists alone, where A and B can be singular or plural. In addition, the character " / " herein generally represents that the associated objects before and after it are in an "or" relationship, but it can also represent an "and / or" relationship, which can be understood according to the context before and after it.

[0057] In the present application, "at least one" means one or more, and "multiple" means two or more. "At least one of the following" or the like means any combination of the items, including any combination of single or multiple items. For example, at least one of a, b, or c can represent a, b, c, a-b, a-c, b-c, or a-b-c, where a, b, and c can be single or multiple.

[0058] It should be understood that in various embodiments of the present application, the size of the sequence number of the above-described processes does not mean the order of execution, and the execution order of the processes should be determined according to their functions and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0059] Those skilled in the art can clearly understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0060] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working processes of the devices, apparatuses and units described above can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

[0061] In several embodiments provided by the present application, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely schematic, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0062] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0063] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically independently, or two or more units can be integrated into one unit.

[0064] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the parts of the present application that essentially contribute to the prior art or the parts of the technical solutions can be embodied in the form of software products. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0065] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A complex urban environment perception and collision warning system based on intelligent connected vehicles, characterized in that, Comprise: Obstacle positioning identification analysis module, data fusion synchronization analysis module and collision risk early warning analysis module; Among them, the obstacle positioning identification analysis module is used for positioning identification analysis of the obstacle perception process of the intelligent network platform according to the obtained obstacle perception data, to determine whether to perform radar wave adjustment optimization, the positioning identification analysis is used for quantifying the position perception accuracy of the intelligent network platform in the intelligent networked vehicle to the obstacle in the collision risk area, and the radar wave adjustment optimization is used for improving the perception ability of the intelligent network platform to the obstacle; The data fusion synchronization analysis module is used for data fusion synchronization analysis of the obstacle data processing process of the intelligent network platform according to the obtained obstacle synchronization identification time, to determine whether to perform data capacity window optimization, the data fusion synchronization analysis is used for quantifying the time alignment degree of the multi-source sensor in the intelligent network platform to process obstacle data, and the data capacity window optimization is used for improving the fusion synchronization ability of the intelligent network platform to obstacle data; The collision risk early warning module is used for collision risk evaluation analysis of the obstacle motion trajectory of the intelligent network platform according to the obtained collision risk prediction data, to determine whether to perform filter cache optimization, the collision risk evaluation analysis is used for quantifying the prediction accuracy of the intelligent network platform to the collision risk of the intelligent networked vehicle and the obstacle, and the filter cache optimization is used for improving the accuracy of collision risk prediction and the reliability of collision risk early warning. 2.The complex urban environment perception and collision warning system based on intelligent vehicles according to claim 1, wherein, The specific steps of the positioning identification analysis of the obstacle perception process of the intelligent network platform according to the obtained obstacle perception data are as follows: At the end of the obstacle perception period, the obtained obstacle perception data is compared with the preset obstacle perception data in the database in terms of difference degree, and the difference degree comparison results are modified by combining the obstacle perception data correction value to obtain obstacle perception data scores, which are coupled to obtain obstacle perception influence indicators; The obstacle perception data includes antenna array response time, multipath interference ranging time and radar receiving response time, the antenna array response time is used to reflect the time required for the antenna array to respond after receiving the signal during the process of perceiving the obstacle, the multipath interference ranging time is used to reflect the time required for the intelligent network platform to complete accurate ranging operation in the multipath interference environment, the radar receiving response time is used to reflect the time from signal receiving state to completing signal processing and giving effective response after the millimeter wave radar receives the signal reflected by the obstacle, and the obstacle perception influence indicator is used to quantify the influence degree of obstacle perception data on the perception ability of the intelligent network platform. 3.The complex urban environment perception and collision warning system based on intelligent vehicles according to claim 2, wherein, The specific steps of determining whether to perform radar wave adjustment optimization are as follows: If the obtained obstacle perception influence indicator is greater than the preset obstacle perception influence indicator in the database, it is recorded as unqualified obstacle perception and radar wave adjustment optimization is performed, the radar wave adjustment optimization means adjusting the radar wave transmitting power and beam width to improve the perception response of the intelligent network platform to the obstacle; If the obtained obstacle perception influence index is not greater than the preset obstacle perception influence index in the database, it is recorded as obstacle perception qualified and data fusion synchronization analysis is performed. 4.The complex urban environment perception and collision warning system based on intelligent vehicles according to claim 3, wherein, The radar wave adjustment optimization includes the following specific steps: Based on the radar wave transmission power adjustment value obtained by mapping the obtained obstacle perception influence index deviation in the database, the specific degree of increasing the radar wave intensity is quantified to reduce the obstacle perception failure probability caused by signal interference; After completing the radar wave transmission power adjustment, a transmission power verification instruction is sent to prompt a preset personnel to perform positioning identification verification test based on the radar wave transmission power after the radar wave transmission power adjustment; After the positioning identification verification test is completed, the obstacle perception process is performed again, if the re-obtained obstacle perception influence index is greater than the preset obstacle perception influence index, an obstacle perception early warning is performed, otherwise, the radar wave transmission power adjustment is completed and data fusion synchronization analysis is performed. 5.The complex urban environment perception and collision warning system based on intelligent vehicles according to claim 4, wherein, The radar wave adjustment optimization also includes beam width optimization, and the specific steps are as follows: Based on the beam width adjustment value obtained by mapping the re-obtained obstacle perception influence index deviation at the end of the re-performed obstacle perception process in the database, the degree of reducing the beam width is quantified to reduce the signal mixed interference caused by dense obstacles; After the beam width adjustment, if the re-obtained obstacle perception influence index is not greater than the preset obstacle perception influence index in the database, the radar wave adjustment optimization is completed and data fusion synchronization analysis is performed, otherwise, an obstacle perception risk early warning is performed. 6.The complex urban environment perception and collision warning system based on intelligent vehicle of claim 1, wherein, The specific steps of determining whether to perform data capacity window optimization are as follows: The obtained obstacle synchronization identification time is compared with the preset obstacle synchronization identification time in the database, and the obstacle synchronization identification time represents the time length from the environment perception data received by the intelligent network platform to the completion of obstacle positioning; If the obtained obstacle synchronization identification time is greater than the preset obstacle synchronization identification time in the database, it is recorded as unqualified synchronization identification and data compression optimization is performed, and the data compression optimization means that the data compression ratio is adjusted to improve the identification response of the intelligent network platform to the obstacle position; If the obtained obstacle synchronization identification time is not greater than the preset obstacle synchronization identification time in the database, it is recorded as qualified synchronization identification and collision risk assessment analysis is performed. 7.The complex urban environment perception and collision warning system based on intelligent vehicles according to claim 6, wherein, The data compression optimization includes the following specific process: Based on the data compression ratio adjustment value obtained by mapping the obtained obstacle synchronization identification time deviation in the database, the identification deviation caused by environment perception data ambiguity is reduced; After the data compression ratio adjustment, if the re-obtained obstacle synchronization identification time is greater than the preset obstacle synchronization identification time in the database, the capacity window optimization is performed, otherwise, the data capacity window optimization is completed and the collision risk assessment analysis is performed. 8.The complex urban environment perception and collision warning system based on intelligent vehicles according to claim 7, wherein, The capacity window optimization includes the following specific process: Based on the buffer capacity adjustment value obtained by mapping the obstacle synchronization identification time deviation obtained after the data compression optimization in the database, the degree of increasing the buffer capacity is quantified to improve the integrity of the identification data of the intelligent network platform. After the buffer capacity adjustment is completed, if the re-acquired obstacle synchronization identification duration is not greater than the preset obstacle synchronization identification duration in the database, data capacity window optimization is completed, otherwise, based on the obstacle synchronization identification duration deviation, a corresponding parallel processing window adjustment value is obtained by mapping in the database, which is used to quantify the degree of increasing the parallel processing window size, to reduce the task queuing delay while increasing the processing rate of multiple parallel threads; After the parallel processing window is adjusted, if the re-acquired obstacle synchronization identification duration is greater than the preset obstacle synchronization identification duration in the database, data fusion warning is performed, otherwise, data capacity window optimization is completed and collision risk assessment analysis is performed. 9.The complex urban environment perception and collision warning system based on intelligent vehicle of claim 1, wherein, The collision risk assessment analysis of the intelligent network platform obstacle motion trajectory according to the obtained collision risk prediction data is specifically: The obtained trajectory prediction delay duration is compared with the preset trajectory prediction delay duration in the database to obtain a trajectory prediction delay duration score, and the obtained ranging delay duration is compared with the preset ranging delay duration in the database to obtain a ranging delay duration score; The ranging delay duration score and the trajectory prediction delay duration score are coupled to obtain a collision risk prediction index, which is used to quantify the influence degree of the collision risk prediction data on the intelligent network platform risk warning capability; The step of judging whether to perform filter cache optimization is: If the obtained collision risk prediction index is greater than the preset collision risk prediction index in the database, it is determined that there is a collision risk and a collision risk warning is performed, otherwise, mean value filtering strength optimization is performed. 10.The complex urban environment perception and collision warning system based on intelligent vehicles according to claim 9, wherein, The mean value filtering strength optimization is specifically: Based on the obtained collision risk prediction index deviation, a mean value filtering strength adjustment value is obtained by mapping in the database, which is used to quantify the degree of increasing the Kalman filter strength, to reduce the target prediction deviation amplitude caused by noise interference; After the filter strength adjustment, if the re-acquired collision risk prediction index is greater than the preset collision risk prediction index in the database, a collision risk warning is performed, otherwise, cache length adjustment is performed; The cache length adjustment is specifically: based on the collision risk prediction index deviation re-acquired after the filter strength adjustment, a read historical data cache length adjustment value is obtained by mapping in the database, which is used to quantify the current required increase in historical cache length, to reduce the deviation misjudgment caused by short-term data fluctuations; After the cache length adjustment, if the re-acquired collision risk prediction index is greater than the preset collision risk prediction index in the database, a collision risk warning is performed, otherwise, a currently no collision risk instruction is sent.

Images