Auxiliary driving method and system based on artificial intelligence

Abstract

The invention relates to the technical field of auxiliary driving, in particular to an auxiliary driving method and system based on artificial intelligence. The method comprises the following steps: S1, generating an abnormal event image recognition result; s2, event area information is generated; s3, generating a preferable detour path; s4, generating a path experience evaluation result; s5, generating a detour scheme recommendation; and S6, generating a driving assistance prompt. The system comprises an abnormity identification module, an area delimiting module, a path analysis module, a path selection module, a path simulation module, an evaluation sorting module and a navigation prompting module. According to the method, the decision-making process of auxiliary driving is optimized through the random forest algorithm, the priority of the detouring scheme can be dynamically adjusted according to the current traffic condition through the Q learning algorithm, the driving experience and safety are optimized, the intelligent decision-making ability of auxiliary driving is improved, errors caused by manual operation are reduced, and the traffic safety and the road use efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of assisted driving, and in particular to an assisted driving method and system based on artificial intelligence. Background Art

[0002] The field of assisted driving technology focuses on the development and application of technical means to improve driving safety, convenience and efficiency. By integrating various sensors and computing platforms, it can achieve real-time perception, recognition and analysis of the surrounding environment, thereby supporting the vehicle to automatically perform operations including lane changing, speed adjustment and emergency braking in target situations, which can reduce traffic accidents and improve road use efficiency.

[0003] Among them, the assisted driving method is a method that uses technical means to analyze and process batch data collected by vehicle sensors to cope with complex traffic environments. The purpose is to improve the perception of the surrounding environment and decision-making intelligence, so that the vehicle can safely assist driving or automatically drive in a wider range of scenarios, thereby reducing human operating errors and improving traffic safety and efficiency.

[0004] For example, the Chinese patent with publication number CN116985829A discloses a car driving method, system and medium based on 5G NR wireless perception, which enables the target car to make auxiliary or automatic driving decisions based on the surrounding environment perception results or the perception data of the current on-board perception sensor, thereby improving the stability of the auxiliary or automatic driving function, and does not need to rely on expensive perception hardware, reducing costs. However, its car driving method cannot quickly adjust the driving path when dealing with emergencies, and cannot fully consider all traffic data and road conditions when selecting a detour path, thereby increasing the risk of traffic congestion and accidents. The fixed algorithm needs to be reconfigured and debugged under new traffic conditions, which is not only time-consuming but also inefficient. Summary of the invention

[0005] The technical problem to be solved by the present invention is: to overcome the deficiencies in the background technology, to provide an assisted driving method based on artificial intelligence, to optimize driving experience and safety, to enhance the intelligent decision-making ability of assisted driving, to reduce errors in human operation, and to improve traffic safety and road use efficiency.

[0006] The technical solution of the present invention is:

[0007] An artificial intelligence-based assisted driving method comprises the following steps:

[0008] S1. Generate abnormal event image recognition results: obtain real-time traffic data, perform continuous frame analysis on traffic images based on real-time traffic data collection and historical event analysis, identify abnormal moving objects in the images, and extract features for classification by comparing similar events in historical image data to generate abnormal event image recognition results;

[0009] S2. Generate event area information: According to the abnormal event image recognition result, the abnormal area in the image is segmented to determine the location and range of the accident or temporary road closure, and the boundary of the affected area is demarcated according to the spatial layout of the event to generate event area information;

[0010] S3. Generate a preferred detour route: Based on the event area information, a random forest algorithm is used to calculate alternative routes that bypass the affected area through multi-dimensional analysis of the current traffic conditions and road network, and the traffic flow and estimated travel time of each route are predicted and evaluated to generate a preferred detour route;

[0011] S4. Generate a path experience evaluation result: Based on the preferred detour path, simulate the driving performance of the differentiated detour solution under traffic conditions, evaluate the congestion level and potential risks of the path, and evaluate it in combination with the driver's historical preferences to generate a path experience evaluation result;

[0012] S5. Generate detour solution recommendation: Based on the path experience evaluation result, the Q learning algorithm is used to sort the detour solutions with reference to time efficiency, safety and driving experience, and the priority of the solutions is dynamically adjusted according to the current traffic conditions to generate detour solution recommendation;

[0013] S6. Generate driving assistance prompts: Based on the detour plan recommendation, the detour plan is intuitively displayed through the display interface of the auxiliary driving device, including a route map, an estimated arrival time and safety prompts, and information transmission is enhanced through sound prompts to generate driving assistance prompts.

[0014] Preferably, in step S1, the abnormal event image recognition results include information on vehicles that illegally change lanes, vehicles that suddenly stop, and pedestrians crossing the road; the event area information includes the coordinates of the accident site and changes in traffic flow; the preferred detour route includes the starting and ending points of the route, the estimated driving time, and key turning points on the route; the route experience evaluation results include an assessment of the degree of congestion on the detour route, an analysis of potential safety risks, and a score for matching the expected driving experience; and the driving assistance prompts include a visual display of the route map, voice navigation instructions, and real-time safety warnings.

[0015] Preferably, the step S1 performs continuous frame analysis on traffic images based on real-time traffic data collection and historical event analysis, identifies abnormal moving objects in the images, and extracts features for classification by comparing similar events in historical image data. The process of generating abnormal event image recognition results includes the following sub-steps:

[0016] S11. Based on real-time traffic data collection and historical event analysis, the real-time images of road traffic are continuously captured by the camera, and vehicles and pedestrians in each frame of the image are preliminarily marked, with their positions and moving directions, to generate a preliminary marking record of moving objects;

[0017] S12, based on the preliminary marking record of the moving objects, comparing the position of each marked object between frames, calculating the moving speed and change trajectory of the object, identifying and comparing the abnormal behavior of the normal traffic flow pattern, and generating a record of abnormal movement behavior;

[0018] S13. According to the abnormal movement behavior record, the abnormal event image stored in the history database is compared, and the event type is classified by matching features to generate an abnormal event image recognition result.

[0019] Preferably, the step S2 determines the location and range of the accident or temporary road closure by segmenting the abnormal area in the image according to the abnormal event image recognition result, and demarcates the boundary of the affected area according to the spatial layout of the event. The process of generating event area information includes the following sub-steps:

[0020] S21, according to the abnormal event image recognition result, identifying the color and texture changes of the regular road environment, marking the position and boundary of the area, and generating abnormal area contour data;

[0021] S22, locating each abnormal area according to the abnormal area contour data, drawing a preliminary boundary line of the affected area by calculating the geometric center, edge contour and overall area of ​​the area, and generating a boundary contour map;

[0022] S23. Analyze the event space layout and road network information according to the boundary contour map, adjust the impact area boundary, match the road layout, and generate event area information.

[0023] Preferably, the step S3 uses a random forest algorithm based on the event area information to calculate alternative paths bypassing the affected area through multi-dimensional analysis of the current traffic conditions and road network, and predicts and evaluates the traffic flow and estimated travel time of each path. The process of generating the preferred detour path includes the following sub-steps:

[0024] S31. Based on the event area information, a random forest algorithm is used to review the existing road network, identify potential detour paths, verify the connectivity of each path, and generate a list of alternative detour paths by referring to road types and connection status;

[0025] S32, according to the list of alternative detour routes, collect and analyze the real-time traffic flow data of the routes, predict the traffic flow change trend, estimate the expected travel time in various time periods, and generate route evaluation information with reference to the impact of weather conditions on traffic flow;

[0026] S33. Based on the path evaluation information, compare and analyze the traffic flow, estimated travel time and path length to determine the optimal path, and generate a preferred detour path with reference to the need for emergency vehicles to have priority passage.

[0027] Preferably, the random forest algorithm has the formula:

[0028]

[0029] Among them, p i is the proportion of the i-th label in the node, f i is the criticality score of feature i, l i is the local density of label i, c i is the average connection strength of category i, and α, β and γ are weight coefficients.

[0030] Preferably, the step S4 evaluates the congestion level and potential risks of the route according to the preferred detour route by simulating the driving performance of the differentiated detour scheme under traffic conditions, and evaluates the route in combination with the driver's historical preferences. The process of generating the route experience evaluation result includes the following sub-steps:

[0031] S41, according to the preferred detour route, simulating the driving conditions of the route in various time periods, capturing the driving speed and flow distribution under differentiated weather conditions, evaluating the performance of the detour solution under traffic conditions, and generating simulated driving performance information;

[0032] S42, analyzing the congestion points and potential risk areas of the path, including intersections, narrow roads, and common accident-prone areas, based on the simulated driving performance information, scoring the congestion degree of the path, and generating path risk assessment information;

[0033] S43. Evaluate each route based on the route risk assessment information and the driver's historical driving preference to generate a route experience assessment result.

[0034] Preferably, the step S5 uses a Q learning algorithm according to the path experience evaluation result, and sorts the detour solutions with reference to time efficiency, safety and driving experience, and dynamically adjusts the priority of the solutions according to the current traffic conditions. The process of generating a detour solution recommendation includes the following sub-steps:

[0035] S51. According to the path experience evaluation result, the evaluation criteria are used to quantitatively score the indicators of each detour solution, the scores are accumulated to evaluate the performance, and a preferred ranking of the solutions is generated;

[0036] S52, according to the preferred ranking of the schemes, using a Q learning algorithm, collecting road traffic data in real time, including speed, flow, accident frequency and temporary road closure information, and dynamically adjusting the priority of each detour scheme to reflect changes in real-time traffic conditions and generate a priority update list;

[0037] S53: Analyze the priority update list, select the best detour solution with reference to the current traffic conditions, driver preferences and solution scores, and generate a detour solution recommendation.

[0038] Preferably, the step S6 intuitively displays the detour plan, including the route map, estimated arrival time and safety tips, through the display interface of the auxiliary driving device according to the detour plan recommendation, and enhances the information transmission through sound prompts. The process of generating driving assistance prompts includes the following sub-steps:

[0039] S61. Displaying a detour route map, marking an estimated arrival time and safety tips for key sections according to the recommended detour plan, and generating a detour plan display interface;

[0040] S62, according to the detour solution display interface, automatically broadcast key detour solution information according to user settings and needs, and generate a customized prompt solution;

[0041] S63. Perform an integrated test based on the detour solution display interface and the customized prompt solution, evaluate the performance in various driving scenarios, adjust the priority of information display and the timing of sound prompts, and generate driving assistance prompts.

[0042] An artificial intelligence-based assisted driving system, the system comprising:

[0043] The anomaly recognition module captures real-time images of road traffic through cameras based on real-time traffic data, preliminarily marks vehicles and pedestrians, records their positions and movement directions, and identifies abnormal behaviors through position comparison and trajectory calculation, and compares them with event images in the historical database to generate an abnormal event feature set;

[0044] The area demarcation module analyzes the color and texture changes of the road environment according to the abnormal event feature set, marks the location and boundary of the abnormal area, calculates the geometric center and edge of the area, and generates the boundary of the affected area;

[0045] A path analysis module, based on the impact area boundary, uses a random forest algorithm to review the road network, identify detour paths, verify path connectivity, and predict traffic changes to generate preferred path details;

[0046] A path selection module, based on the preferred path details, simulates the performance of the detour solution under various traffic conditions, evaluates the congestion level and risk, evaluates the driving preference, and establishes a path priority sequence;

[0047] A path simulation module, which marks the arrival time and safety tips according to the path priority sequence, automatically broadcasts key information, and generates navigation voice instructions;

[0048] An evaluation and sorting module, based on the navigation voice instructions, uses a Q-learning algorithm to perform integration testing, adjust the priority of information display and the timing of voice prompts, and generate navigation optimization;

[0049] The navigation prompt module displays a detour plan based on the navigation optimization, including a route map, an estimated arrival time and safety prompts, and generates a driving assistance prompt.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] By using the random forest algorithm to optimize the decision-making process of assisted driving and analyzing multi-dimensional traffic conditions and road network data, the traffic flow and estimated driving time of alternative detour routes can be accurately calculated and predicted, providing data support for selecting the optimal detour route, so that the route selection can accurately reflect the driving conditions of the route. The Q learning algorithm can dynamically adjust the priority of the detour plan according to the current traffic conditions, optimize the driving experience and safety, enhance the intelligent decision-making ability of assisted driving, reduce human operation errors, and improve traffic safety and road use efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0053] Figure 1 It is a flow chart of the assisted driving method of the present invention.

[0054] Figure 2 It is a flow chart of the process of generating abnormal event image recognition results of the present invention.

[0055] Figure 3 It is a flow chart of the process of generating event area information of the present invention.

[0056] Figure 4 It is a flow chart of the process of generating a preferred detour path of the present invention.

[0057] Figure 5 It is a flow chart of the process of generating path experience evaluation results of the present invention.

[0058] Figure 6 It is a flow chart of the process of generating detour solution recommendation of the present invention.

[0059] Figure 7 It is a flow chart of the process of generating driving assistance prompts of the present invention.

[0060] Figure 8 It is a structural block diagram of the system of the present invention. DETAILED DESCRIPTION

[0061] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0062] Example

[0063] like Figure 1 As shown, this embodiment provides an artificial intelligence-based assisted driving method, including:

[0064] S1. Generate abnormal event image recognition results: obtain real-time traffic data, perform continuous frame analysis on traffic images based on real-time traffic data collection and historical event analysis, identify abnormal moving objects in the images, and extract features for classification by comparing similar events in historical image data to generate abnormal event image recognition results;

[0065] S2. Generate event area information: According to the abnormal event image recognition result, the abnormal area in the image is segmented to determine the location and range of the accident or temporary road closure, and the boundary of the affected area is demarcated according to the spatial layout of the event to generate event area information;

[0066] S3. Generate a preferred detour route: Based on the event area information, a random forest algorithm is used to calculate alternative routes that bypass the affected area through multi-dimensional analysis of the current traffic conditions and road network, and the traffic flow and estimated travel time of each route are predicted and evaluated to generate a preferred detour route;

[0067] S4. Generate a path experience evaluation result: Based on the preferred detour path, simulate the driving performance of the differentiated detour solution under traffic conditions, evaluate the congestion level and potential risks of the path, and evaluate it in combination with the driver's historical preferences to generate a path experience evaluation result;

[0068] S5. Generate detour solution recommendation: Based on the path experience evaluation result, the Q learning algorithm is used to sort the detour solutions with reference to time efficiency, safety and driving experience, and the priority of the solutions is dynamically adjusted according to the current traffic conditions to generate detour solution recommendation;

[0069] S6. Generate driving assistance prompts: Based on the detour plan recommendation, the detour plan is intuitively displayed through the display interface of the auxiliary driving device, including a route map, an estimated arrival time and safety prompts, and information transmission is enhanced through sound prompts to generate driving assistance prompts.

[0070] The abnormal event image recognition results include information on vehicles that change lanes illegally, vehicles that stop suddenly, and pedestrians crossing the road. The event area information includes the coordinates of the accident site and traffic flow changes. The preferred detour route includes the starting and ending points of the route, the estimated driving time, and key turning points on the route. The route experience evaluation results include an assessment of the congestion level of the detour route, an analysis of potential safety risks, and a score that matches the expected driving experience. The driving assistance prompts include a visual display of the route map, voice navigation instructions, and real-time safety warnings.

[0071] like Figure 2 As shown, based on real-time traffic data collection and historical event analysis, continuous frame analysis is performed on traffic images to identify abnormal moving objects in the images, and features are extracted for classification by comparing similar events in historical image data. The specific steps for generating abnormal event image recognition results are as follows:

[0072] S11: Based on real-time traffic data collection and historical event analysis, the real-time images of road traffic are continuously captured by the camera, and vehicles and pedestrians in each frame of the image are preliminarily marked, and the positions and moving directions are annotated. The data is recorded as a preliminary marking record of the moving object, and a preliminary marking record of the moving object is generated;

[0073] S12: Based on the preliminary marking records of the moving objects, the positions of each marked object are compared between frames, the moving speed and change trajectory of the object are calculated using the optical flow tracking technology, and the abnormal behavior is identified by comparing with the normal traffic flow pattern, thereby generating a record of abnormal movement behavior;

[0074] S13: Based on the abnormal movement behavior records, feature matching technology is used to compare the abnormal event images stored in the historical database, image similarity comparison technology is used to classify the event type, and image processing algorithms are used to analyze and match features, and finally abnormal event image recognition results are generated;

[0075] like Figure 3As shown, according to the abnormal event image recognition results, the abnormal area in the image is segmented to determine the location and range of the accident or temporary road closure, and the boundary of the affected area is delineated according to the spatial layout of the event. The specific steps for generating event area information are as follows:

[0076] S21: Based on the abnormal event image recognition results, image analysis technology is used to detect changes in the color and texture of the regular road environment. The edge detection algorithm is used to accurately mark the location and boundaries of the abnormal area. The Sobel operator is used to calculate the intensity and direction of the image gradient, identify the boundary line, and generate the abnormal area contour data.

[0077] S22: Based on the contour data of the abnormal area, the geometric center positioning algorithm is used to accurately locate each abnormal area. The total area of ​​each marked area is calculated by the area calculation function, and the contour tracking technology is applied to draw the preliminary boundary line of the affected area. The findContours and drawContours functions in the OpenCV library are used to implement contour tracking and drawing, thereby generating a boundary contour map.

[0078] S23: Based on the boundary contour map, spatial analysis tools are used, including the spatial overlay analysis function in GIS software, to integrate and analyze the event spatial layout and road network information. The boundary of the affected area is adjusted through the road layout matching algorithm. The road network is reconstructed and optimized with reference to the road type and frequency of use to generate event area information.

[0079] like Figure 4 As shown in the figure, based on the event area information, the random forest algorithm is used to calculate alternative paths to bypass the affected area through multi-dimensional analysis of the current traffic conditions and road network, and the traffic flow and estimated travel time of each path are predicted and evaluated. The specific steps for generating the preferred detour path are as follows:

[0080] S31: Based on the event area information, the random forest algorithm is used to perform road network analysis and optimization operations. By constructing multiple decision trees and outputting pattern recognition results, the process is made more accurate. The road type and connection status are used as input variables. The number of decision trees is set to 100 to ensure the generalization ability of the model and reduce the risk of overfitting. The maximum depth of each tree is set to 10. The Gini impurity is used as the criterion for judging the split quality. The various connection states in the existing road network are reviewed and evaluated, and possible detour paths based on current traffic flow and road conditions are identified. The connectivity of each potential path is verified to ensure the feasibility of each path in theory and practical applications, and a list of alternative detour paths is generated.

[0081] Random forest algorithm, according to the formula:

[0082]

[0083] Among them, p i is the proportion of the i-th label in the node, f i is the criticality score of feature i, l i is the local density of label i, c i is the average connection strength between category i and other categories, α, β and γ are weight coefficients;

[0084] Execution process: Calculate each category label p i The initial impurity is evaluated by the ratio of , which serves as the basic basis for splitting decision trees in random forests, and the importance score f of feature i is added i , the impact of the model prediction results is evaluated by feature selection and importance ranking methods, and the local density l is calculated i , represents the degree of aggregation of label i in its surrounding area, and the average connection strength c between category i and other categories i , calculate the average distance or correlation between categories, and determine the weight coefficients α, β and γ through the model optimization process, which are determined by methods such as cross-validation and are used to balance the impact of the original impurity measure and the newly introduced parameters, ensuring that the calculation more accurately reflects the actual situation of the data.

[0085] S32: Based on the list of alternative detour routes, real-time traffic flow data of the routes are collected and analyzed, and the traffic flow change trend is predicted using time series analysis methods. Python's Pandas library is used for data time indexing and statistical analysis. Traffic flow in differentiated time periods is estimated based on past data trends. At the same time, the influence of weather conditions is referred to and numerical calculations are performed using the NumPy library, so as to accurately process and generate path evaluation information.

[0086] S33: Based on the path evaluation information, a comparative analysis of traffic flow, estimated travel time and path length is conducted. The optimal path is determined by data clustering analysis. The Matplotlib library is used for data visualization. The performance comparison of multiple paths is monitored through the view. The path that best suits the need for priority passage of emergency vehicles is selected. During the operation, the weight distribution of multiple indicators is focused on, and finally the preferred detour path is generated.

[0087] See also Figure 5 According to the preferred detour route, the driving performance of the differentiated detour scheme under traffic conditions is simulated to evaluate the congestion level and potential risks of the route, and the evaluation is combined with the driver's historical preferences. The specific steps for generating the route experience evaluation results are as follows:

[0088] S41: Based on the preferred detour route, the Dijkstra algorithm is used to calculate the shortest path from the starting point to each node to simulate the path, and the driving conditions in various time periods are simulated. The queuing theory model is used to evaluate the performance of the detour solution under traffic conditions by calculating the waiting time and service time, and the flow and speed data are analyzed. The distribution of data points in each time period is analyzed to determine the flow peaks and speed bottlenecks, and generate simulated driving performance information.

[0089] S42: Based on the simulated driving performance information, spatial analysis technology is used to analyze the geographical location of congestion points and potential risk areas. The spatial analysis function of ArcGIS software is used to calculate the congestion probability of intersections, narrow roads and accident-prone areas, analyze and score the congestion level of the path, and generate path risk assessment information.

[0090] S43: Based on the path risk assessment information and combined with the driver's historical driving preferences, a multi-factor decision analysis method is adopted. The historical preference data is integrated and analyzed using Python's Pandas library. The driving experience of the path is evaluated through the weighted average method, and finally the path experience evaluation result is generated.

[0091] like Figure 6 As shown in the figure, according to the path experience evaluation results, the Q learning algorithm is used to sort the detour plans with reference to time efficiency, safety and driving experience, and the plan priority is dynamically adjusted according to the current traffic conditions. The specific steps for generating detour plan recommendations are as follows:

[0092] S51: Based on the results of the path experience evaluation, a quantitative scoring method is used to quantify the performance indicators of each detour plan. The performance of the plan is evaluated by accumulating the scores. The Python Pandas library is used to process the data and calculate the scores. By defining the weight coefficient of each indicator, the scores are weighted and synthesized, and a preferred ranking of the plans is generated.

[0093] S52: Based on the preferred ranking of the schemes, the Q-learning algorithm is used for real-time data processing, the Q-table is initialized, the learning rate is set to 0.8 and the discount factor is set to 0.9, and the optional actions under each state are defined, including speed changes, traffic increases and decreases, accident frequency increases, and road closure status changes. The values ​​in the Q-table are updated in real time through the collected road traffic data, and the optimal action is selected to adjust the priority of the detour scheme based on the current state. Data is collected through sensors, including speed sensor readings and traffic flow counters, and the expected benefits of each detour scheme under the new situation are calculated. The priority is updated to reflect changes in traffic conditions in real time and generate a priority update list.

[0094] S53: Perform analysis based on the priority update list, combine the current traffic conditions, the driver's historical driving preferences and the plan scores, adopt a multi-attribute decision analysis method, use decision support system software to integrate and analyze the data, and obtain the bypass plan with the highest score through comparative analysis. During the operation, refer to the impact of multiple indicators on the driving experience, and finally generate a bypass plan recommendation.

[0095] like Figure 7 As shown, according to the detour plan recommendation, the detour plan is intuitively displayed through the display interface of the auxiliary driving device, including the route map, estimated arrival time and safety tips, and the information transmission is enhanced through sound prompts. The specific steps for generating driving assistance prompts are as follows:

[0096] S61: Based on the recommended detour plan, design and develop an interface display module, use HTML5 and JavaScript technology to build an interactive map interface, visualize the detour route map, mark the estimated arrival time and safety tips for key sections, use CSS for style optimization, integrate front-end development tools, and generate a detour plan display interface.

[0097] S62: Based on the detour plan display interface, develop a sound prompt module, use the WebAudio API for audio processing, and automatically broadcast the key information of the detour plan according to the user's settings, including using JavaScript for trigger event programming, setting automatic and manual broadcast modes, and adjusting the playback settings of audio files to match differentiated user needs, optimize the user interaction experience, and generate a customized prompt plan.

[0098] S63: Perform integration testing based on the detour solution display interface and customized prompt solution, use Selenium for automated interface testing, evaluate performance in multiple driving scenarios, adjust the priority of information display and the timing of voice prompts, automate the test process through Python scripts, record test results and performance indicators, ensure that the function works as expected, and finally generate driver assistance prompts.

[0099] like Figure 8 As shown, an artificial intelligence-based assisted driving system includes:

[0100] The anomaly recognition module is based on real-time traffic data. It uses cameras to capture real-time images of road traffic, preliminarily mark vehicles and pedestrians, record their positions and movement directions, and identify abnormal behaviors through position comparison and trajectory calculation. It compares them with event images in the historical database to generate an abnormal event feature set.

[0101] The area delineation module analyzes the color and texture changes of the road environment based on the abnormal event feature set, marks the location and boundaries of the abnormal area, calculates the geometric center and edge of the area, and generates the boundary of the affected area;

[0102] The route analysis module uses a random forest algorithm to review the road network, identify detours, verify route connectivity, and predict traffic changes based on the impact area boundaries to generate optimal route details;

[0103] The route selection module simulates the performance of detour solutions under various traffic conditions based on the preferred route details, evaluates the congestion level and risk, and establishes a route priority sequence based on driving preferences;

[0104] The path simulation module marks the arrival time and safety tips based on the path priority sequence, automatically broadcasts key information, and generates navigation voice instructions;

[0105] The evaluation and sorting module uses the Q-learning algorithm based on navigation voice commands to conduct integrated testing, adjust the priority of information display and the timing of voice prompts, and generate navigation optimization;

[0106] The navigation prompt module displays detour plans based on navigation optimization, including route maps, estimated arrival times and safety tips, and generates driving assistance prompts.

[0107] Although the present invention has been described in detail by referring to the accompanying drawings and in combination with the preferred embodiments, the present invention is not limited thereto. Without departing from the spirit and essence of the present invention, a person of ordinary skill in the art may make various equivalent modifications or substitutions to the embodiments of the present invention, and these modifications or substitutions shall be within the scope of the present invention. Any person of ordinary skill in the art may easily think of changes or substitutions within the technical scope disclosed by the present invention, and these shall be within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. An artificial intelligence-based assisted driving method, characterized in that: The steps include: S1. Generate abnormal event image recognition results: obtain real-time traffic data, perform continuous frame analysis on traffic images based on real-time traffic data collection and historical event analysis, identify abnormal moving objects in the images, and extract features for classification by comparing similar events in historical image data to generate abnormal event image recognition results; S2. Generate event area information: According to the abnormal event image recognition result, the abnormal area in the image is segmented to determine the location and range of the accident or temporary road closure, and the boundary of the affected area is demarcated according to the spatial layout of the event to generate event area information; S3. Generate a preferred detour route: Based on the event area information, a random forest algorithm is used to calculate alternative routes that bypass the affected area through multi-dimensional analysis of the current traffic conditions and road network, and the traffic flow and estimated travel time of each route are predicted and evaluated to generate a preferred detour route; S4. Generate a path experience evaluation result: Based on the preferred detour path, simulate the driving performance of the differentiated detour solution under traffic conditions, evaluate the congestion level and potential risks of the path, and evaluate it in combination with the driver's historical preferences to generate a path experience evaluation result; S5. Generate detour solution recommendation: Based on the path experience evaluation result, the Q learning algorithm is used to sort the detour solutions with reference to time efficiency, safety and driving experience, and the priority of the solutions is dynamically adjusted according to the current traffic conditions to generate detour solution recommendation; S6. Generate driving assistance prompts: Based on the detour plan recommendation, the detour plan is intuitively displayed through the display interface of the auxiliary driving device, including a route map, an estimated arrival time and safety prompts, and information transmission is enhanced through sound prompts to generate driving assistance prompts.

2. The artificial intelligence-based assisted driving method according to claim 1, characterized in that: In step S1, the abnormal event image recognition results include information on vehicles that illegally change lanes, vehicles that suddenly stop, and pedestrians crossing the road; the event area information includes the coordinates of the accident site and traffic flow changes; the preferred detour route includes the starting and ending points of the route, the estimated driving time, and key turning points on the route; the route experience evaluation results include a detour route congestion assessment, a potential safety risk analysis, and an expected driving experience matching score; the driving assistance prompts include a route map visualization display, voice navigation instructions, and real-time safety warnings.

3. The artificial intelligence-based assisted driving method according to claim 1, characterized in that: The step S1 performs continuous frame analysis on traffic images based on real-time traffic data collection and historical event analysis, identifies abnormal moving objects in the images, and extracts features for classification by comparing similar events in historical image data. The process of generating abnormal event image recognition results includes the following sub-steps: S11. Based on real-time traffic data collection and historical event analysis, the real-time images of road traffic are continuously captured by the camera, and vehicles and pedestrians in each frame of the image are preliminarily marked, with their positions and moving directions, to generate a preliminary marking record of moving objects; S12, based on the preliminary marking record of the moving objects, comparing the position of each marked object between frames, calculating the moving speed and change trajectory of the object, identifying and comparing the abnormal behavior of the normal traffic flow pattern, and generating a record of abnormal movement behavior; S13. According to the abnormal movement behavior record, the abnormal event image stored in the history database is compared, and the event type is classified by matching features to generate an abnormal event image recognition result.

4. The artificial intelligence-based assisted driving method according to claim 1, characterized in that: The step S2 determines the location and range of the accident or temporary road closure by segmenting the abnormal area in the image according to the abnormal event image recognition result, and demarcates the boundary of the affected area according to the spatial layout of the event. The process of generating event area information includes the following sub-steps: S21, according to the abnormal event image recognition result, identifying the color and texture changes of the regular road environment, marking the location and boundary of the area, and generating abnormal area contour data; S22, locating each abnormal area according to the abnormal area contour data, drawing a preliminary boundary line of the affected area by calculating the geometric center, edge contour and overall area of ​​the area, and generating a boundary contour map; S23. Analyze the event space layout and road network information according to the boundary contour map, adjust the impact area boundary, match the road layout, and generate event area information.

5. The artificial intelligence-based assisted driving method according to claim 4, characterized in that: The step S3 uses the random forest algorithm based on the event area information to calculate alternative paths bypassing the affected area through multi-dimensional analysis of the current traffic conditions and road network, and predicts and evaluates the traffic flow and estimated travel time of each path. The process of generating the preferred bypass path includes the following sub-steps: S31. Based on the event area information, a random forest algorithm is used to review the existing road network, identify potential detour paths, verify the connectivity of each path, and generate a list of alternative detour paths by referring to road types and connection status; S32, according to the list of alternative detour routes, collect and analyze the real-time traffic flow data of the routes, predict the traffic flow change trend, estimate the expected travel time in various time periods, and generate route evaluation information with reference to the impact of weather conditions on traffic flow; S33. Based on the path evaluation information, compare and analyze the traffic flow, estimated travel time and path length to determine the optimal path, and generate a preferred detour path with reference to the need for emergency vehicles to have priority passage.

6. The artificial intelligence-based assisted driving method according to claim 5, characterized in that: The random forest algorithm, the formula is: Among them, p i is the proportion of the i-th label in the node, f i is the criticality score of feature i, l i is the local density of label i, c i is the average connection strength of category i, and α, β and γ are weight coefficients.

7. The artificial intelligence-based assisted driving method according to claim 1, characterized in that: The step S4 evaluates the congestion level and potential risks of the route according to the preferred detour route by simulating the driving performance of the differentiated detour solutions under traffic conditions, and evaluates the route in combination with the driver's historical preferences. The process of generating the route experience evaluation result includes the following sub-steps: S41, according to the preferred detour route, simulating the driving conditions of the route in various time periods, capturing the driving speed and flow distribution under differentiated weather conditions, evaluating the performance of the detour solution under traffic conditions, and generating simulated driving performance information; S42, analyzing the congestion points and potential risk areas of the path, including intersections, narrow roads, and common accident-prone areas, based on the simulated driving performance information, scoring the congestion degree of the path, and generating path risk assessment information; S43. Evaluate each route based on the route risk assessment information and the driver's historical driving preference to generate a route experience assessment result.

8. The artificial intelligence-based assisted driving method according to claim 1, characterized in that: The step S5 uses a Q learning algorithm to sort the detour solutions according to the path experience evaluation result, with reference to time efficiency, safety and driving experience, and dynamically adjusts the priority of the solutions according to the current traffic conditions. The process of generating a detour solution recommendation includes the following sub-steps: S51. According to the path experience evaluation result, the evaluation criteria are used to quantitatively score the indicators of each detour solution, the scores are accumulated to evaluate the performance, and a preferred ranking of the solutions is generated; S52, according to the preferred ranking of the schemes, using a Q learning algorithm, collecting road traffic data in real time, including speed, flow, accident frequency and temporary road closure information, and dynamically adjusting the priority of each detour scheme to reflect changes in real-time traffic conditions and generate a priority update list; S53: Analyze the priority update list, select the best detour solution with reference to the current traffic conditions, driver preferences and solution scores, and generate a detour solution recommendation.

9. The artificial intelligence-based assisted driving method according to claim 1, characterized in that: The step S6 intuitively displays the detour plan, including the route map, estimated arrival time and safety tips, through the display interface of the auxiliary driving device according to the detour plan recommendation, and enhances the information transmission through sound prompts. The process of generating driving assistance prompts includes the following sub-steps: S61. Displaying a detour route map, marking an estimated arrival time and safety tips for key sections according to the recommended detour plan, and generating a detour plan display interface; S62, according to the detour solution display interface, automatically broadcast key detour solution information according to user settings and needs, and generate a customized prompt solution; S63. Perform an integrated test based on the detour solution display interface and the customized prompt solution, evaluate the performance in various driving scenarios, adjust the priority of information display and the timing of sound prompts, and generate driving assistance prompts.

10. An artificial intelligence-based assisted driving system, applying the artificial intelligence-based assisted driving method according to any one of claims 1 to 9, characterized in that: The system comprises: The anomaly recognition module captures real-time images of road traffic through cameras based on real-time traffic data, preliminarily marks vehicles and pedestrians, records their positions and movement directions, and identifies abnormal behaviors through position comparison and trajectory calculation, and compares them with event images in the historical database to generate an abnormal event feature set; The area demarcation module analyzes the color and texture changes of the road environment according to the abnormal event feature set, marks the location and boundary of the abnormal area, calculates the geometric center and edge of the area, and generates the boundary of the affected area; A path analysis module, based on the impact area boundary, uses a random forest algorithm to review the road network, identify detour paths, verify path connectivity, and predict traffic changes to generate preferred path details; A path selection module, based on the preferred path details, simulates the performance of the detour solution under various traffic conditions, evaluates the congestion level and risk, evaluates the driving preference, and establishes a path priority sequence; A path simulation module, which marks the arrival time and safety tips according to the path priority sequence, automatically broadcasts key information, and generates navigation voice instructions; An evaluation and sorting module, based on the navigation voice instructions, uses a Q-learning algorithm to perform integration testing, adjust the priority of information display and the timing of voice prompts, and generate navigation optimization; The navigation prompt module displays a detour plan based on the navigation optimization, including a route map, an estimated arrival time and safety prompts, and generates a driving assistance prompt.

Images