Driving assistance teaching method and device, electronic equipment and storage medium

Abstract

The application relates to a driving auxiliary teaching method and device, electronic equipment and a storage medium. The method provided in the application comprises the following steps: acquiring environmental sensing data around a target vehicle and reference driving track data corresponding to a driving route of the target vehicle; determining driving reference data for indicating safe driving of the target vehicle along the reference driving track according to the data; determining driving gap data according to driving behavior data of the target vehicle and the driving reference data; and generating corresponding driving teaching video according to the driving reference data and the driving gap data. By comparing the driving behavior data and the driving reference data, the error driving operation of the user is quantified as the driving gap data, the error correction rate of the driving behavior of the student is improved, meanwhile, the driving gap data is converted into intuitive visual experience through the personalized teaching video, the intuitiveness and attraction of the teaching are greatly improved, and the teaching effect is greatly improved.

Description

Technical Field

[0001] This application relates to the field of vehicle control technology, and in particular to a driving assistance teaching method, device, electronic device and storage medium. Background Technology

[0002] Traditional driver training and assessment systems rely primarily on human instructors and related courses. Trainees train in fixed driving areas or pre-set scenarios, with instructors providing guidance based on observation and subjective judgment.

[0003] Furthermore, existing intelligent driving assistance systems are mostly used for vehicle control (such as autonomous driving) rather than for instruction and guidance. Driving behavior analysis systems are also mostly based on rule engines or simple threshold judgments, lacking the ability to fuse and analyze multi-dimensional driving data and generate personalized suggestions. Summary of the Invention

[0004] This application provides a driving assistance teaching method, device, electronic device, and storage medium to solve the technical problem that existing driving assistance teaching methods rely on manual labor, have a low level of intelligence, and thus result in poor teaching effectiveness.

[0005] Firstly, this application provides a driving assistance teaching method, the method comprising:

[0006] Acquire driving behavior data of the target vehicle and environmental perception data around the target vehicle;

[0007] Obtain reference driving trajectory data corresponding to the current driving route of the target vehicle;

[0008] Based on the reference driving trajectory data and the environmental perception data, driving reference data is determined; wherein, the driving reference data is used to indicate the standard operating parameters corresponding to the safe driving of the target vehicle along the reference driving trajectory under the driving scenario represented by the environmental perception data;

[0009] Based on the driving behavior data and the driving reference data, driving gap data is determined;

[0010] Based on the driving reference data and the driving gap data, a driving instruction video is generated; wherein, the driving instruction video is used to compare and demonstrate the differences between standard driving operations and user driving operations.

[0011] In one possible implementation, determining driving reference data based on the reference driving trajectory data and the environmental perception data includes:

[0012] Obtain the historical driving dataset of the user of the target vehicle prior to this driving session;

[0013] From the historical driving dataset, target historical driving data that matches the driving scenario represented by the environmental perception data are selected;

[0014] The reference driving trajectory data, the environmental perception data, and the target historical driving data are input into the trained intelligent driving model to obtain the driving reference data output by the intelligent driving model.

[0015] In one possible implementation, determining driving gap data based on the driving behavior data and the driving reference data includes:

[0016] Align the driving behavior data with the driving reference data on the time axis;

[0017] By comparing the aligned driving behavior data with driving reference data, initial gap data containing at least one operational parameter deviation value is obtained; wherein, the operational parameter deviation value is used to represent the deviation value between the operational parameter corresponding to the user's driving behavior and the standard operational parameter in the driving reference data;

[0018] Multi-dimensional feature analysis is performed on the initial gap data to obtain driving gap data.

[0019] In one possible implementation, multi-dimensional feature analysis is performed on the initial gap data to obtain driving gap data, including:

[0020] The driving behavior data, the environmental perception data, and the initial gap data are input into a trained multimodal feature analysis model to obtain multidimensional deviation values ​​output by the multimodal feature analysis model. The multidimensional deviation values ​​include at least one or more of trajectory deviation, operation smoothness deviation, and safety distance deviation.

[0021] Multi-dimensional deviations that exceed the set safety threshold range are identified as driving gap data.

[0022] In one possible implementation, the multimodal feature analysis model includes:

[0023] The feature extraction layer is used to extract the operation time sequence features of the driving behavior data using a convolutional network, and to construct the dynamic spatial relationship features between the target vehicle and surrounding obstacles using a graph neural network.

[0024] The spatiotemporal fusion layer is used to fuse the operation temporal features and the dynamic spatial relationship features using a recurrent neural network to obtain the correlation features between operation and space.

[0025] The gap quantization layer is used to output multi-dimensional deviation values ​​based on the associated features.

[0026] In one possible implementation, the security threshold range is set in the following manner:

[0027] Adjust the preset safety threshold range based on the driving scenario represented by the environmental perception data;

[0028] And / or,

[0029] Obtain the user's historical driving behavior data prior to this driving session; adjust the preset safety threshold range based on the historical driving behavior data.

[0030] In one possible implementation, a driving instruction video is generated based on the driving reference data and the driving gap data, including:

[0031] Based on the environmental perception data, a virtual driving scenario corresponding to the driving scenario of the target vehicle is reconstructed;

[0032] Based on the driving behavior data, the user's actual operation trajectory in the virtual driving scenario is generated;

[0033] In the virtual driving scenario, a demonstration screen of standard driving operation is generated and displayed based on the driving reference data, and the actual operation trajectory is compared and displayed with the standard driving operation trajectory using a first marker.

[0034] Based on the location and type of deviation indicated by the driving gap data, at least one of a second marker or a slow-motion replay clip is superimposed on the demonstration screen to generate a driving instruction video.

[0035] In one possible implementation, the method further includes:

[0036] According to a preset period, obtain driving difference data of the user driving the target vehicle within the corresponding period;

[0037] The obtained driving gap data within the corresponding period is input into the trained data analysis model to obtain a driving behavior summary report output by the data analysis model.

[0038] Secondly, this application provides a driving assistance teaching device, the device comprising:

[0039] An environmental perception data acquisition module is used to acquire driving behavior data of the target vehicle and environmental perception data around the target vehicle.

[0040] The driving trajectory data acquisition module is used to acquire reference driving trajectory data corresponding to the current driving route of the target vehicle;

[0041] The driving reference data determination module is used to determine driving reference data based on the reference driving trajectory data and the environmental perception data; wherein, the driving reference data is used to indicate the standard operating parameters corresponding to the safe driving of the target vehicle along the reference driving trajectory under the driving scenario represented by the environmental perception data;

[0042] The driving gap data determination module is used to determine driving gap data based on the driving behavior data and the driving reference data.

[0043] The driving instruction video generation module is used to generate driving instruction videos based on the driving reference data and the driving gap data; wherein, the driving instruction videos are used to compare and demonstrate the differences between standard driving operations and user driving operations.

[0044] In one possible implementation, the driving reference data determination module is specifically used for:

[0045] Obtain the historical driving dataset of the user of the target vehicle prior to this driving session;

[0046] From the historical driving dataset, target historical driving data that matches the driving scenario represented by the environmental perception data are selected;

[0047] The reference driving trajectory data, the environmental perception data, and the target historical driving data are input into the trained intelligent driving model to obtain the driving reference data output by the intelligent driving model.

[0048] In one possible implementation, the driving gap data determination module includes:

[0049] A data alignment unit is used to align the driving behavior data with the driving reference data on the time axis.

[0050] The data comparison unit is used to compare the aligned driving behavior data with driving reference data to obtain initial difference data containing at least one operational parameter deviation value; wherein, the operational parameter deviation value is used to represent the deviation value between the operational parameter corresponding to the user's driving behavior and the standard operational parameter in the driving reference data;

[0051] The driving gap data acquisition unit is used to perform multi-dimensional feature analysis on the initial gap data to obtain driving gap data.

[0052] In one possible implementation, the driving gap data acquisition unit includes:

[0053] The feature analysis subunit is used to input the driving behavior data, the environmental perception data, and the initial gap data into the trained multimodal feature analysis model to obtain the multidimensional deviation value output by the multimodal feature analysis model. The multidimensional deviation value includes at least one or more of trajectory deviation, operation smoothness deviation, and safety distance deviation.

[0054] The driving gap data determination subunit identifies multi-dimensional deviation values ​​that exceed the set safety threshold range as driving gap data.

[0055] In one possible implementation, the multimodal feature analysis model in the feature analysis subunit is specifically used for:

[0056] The feature extraction layer is used to extract the operation time sequence features of the driving behavior data using a convolutional network, and to construct the dynamic spatial relationship features between the target vehicle and surrounding obstacles using a graph neural network.

[0057] The spatiotemporal fusion layer is used to fuse the operation temporal features and the dynamic spatial relationship features using a recurrent neural network to obtain the correlation features between operation and space.

[0058] The gap quantization layer is used to output multi-dimensional deviation values ​​based on the associated features.

[0059] In one possible implementation, the safety threshold range of the driving gap data determination subunit is set in the following manner:

[0060] Adjust the preset safety threshold range based on the driving scenario represented by the environmental perception data;

[0061] And / or,

[0062] Obtain the user's historical driving behavior data prior to this driving session; adjust the preset safety threshold range based on the historical driving behavior data.

[0063] In one possible implementation, the device further includes:

[0064] The prompt word generation module is used to generate prompt words based on at least one of the driving gap data, the environmental perception data, and the vehicle status data of the target vehicle;

[0065] The driving suggestion text determination module is used to input the prompt words into a trained large language model to obtain natural language driving suggestion text output by the large language model; wherein, the natural language driving suggestion text includes improvement measures and / or operation guidance for the driving gap data.

[0066] In one possible implementation, the driving instruction video generation module is specifically used for:

[0067] Based on the environmental perception data, a virtual driving scenario corresponding to the driving scenario of the target vehicle is reconstructed;

[0068] Based on the driving behavior data, the user's actual operation trajectory in the virtual driving scenario is generated;

[0069] In the virtual driving scenario, a demonstration screen of standard driving operation is generated and displayed based on the driving reference data, and the actual operation trajectory is compared and displayed with the standard driving operation trajectory using a first marker.

[0070] Based on the location and type of deviation indicated by the driving gap data, at least one of a second marker or a slow-motion replay clip is superimposed on the demonstration screen to generate a driving instruction video.

[0071] In one possible implementation, the device further includes:

[0072] The driving gap data acquisition module is used to acquire driving gap data of the user driving the target vehicle within a preset period.

[0073] The driving report generation module is used to input the acquired driving gap data within the corresponding period into the trained data analysis model to obtain a driving behavior summary report output by the data analysis model.

[0074] Thirdly, this application provides an electronic device, including: a processor and a memory, wherein the processor is configured to execute a driving assistance teaching program stored in the memory to implement the driving assistance teaching method described in any one of the first aspects.

[0075] Fourthly, this application provides a storage medium storing one or more programs that can be executed by one or more processors to implement the driving assistance teaching method described in any one aspect.

[0076] Compared with the prior art, the technical solution provided in this application has the following advantages: The method provided in this application acquires environmental perception data around the target vehicle and reference driving trajectory data corresponding to the target vehicle's current driving route; determines driving reference data corresponding to instruct the target vehicle to drive safely along the reference driving trajectory based on these data; determines driving gap data based on the acquired driving behavior data and driving reference data of the target vehicle; and then generates corresponding driving instruction videos based on the driving reference data and driving gap data. By comparing driving behavior data and driving reference data, the driving gap data corresponding to the user's erroneous driving operations is determined, eliminating the reliance on manual judgment of the student's erroneous driving operations, thus improving the error correction rate of the student's driving behavior. At the same time, through personalized instruction videos, the abstract driving gap data is transformed into an intuitive visual experience, greatly enhancing the intuitiveness and attractiveness of the teaching, thereby significantly improving the teaching effect. Attached Figure Description

[0077] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0078] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0079] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0080] Figure 1 A flowchart illustrating an embodiment of a driving assistance teaching method provided in this application;

[0081] Figure 2 A flowchart illustrating another embodiment of the driving assistance teaching method provided in this application;

[0082] Figure 3 A flowchart illustrating another embodiment of the driving assistance teaching method provided in this application;

[0083] Figure 4 A structural block diagram of a driving assistance teaching device provided in this application;

[0084] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0085] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0086] The following disclosure provides numerous different embodiments or examples for implementing various structures of this application. To simplify the disclosure, specific examples and arrangements are described below. These are merely examples and are not intended to limit the scope of this application. Furthermore, reference numerals and / or letters may be repeated in different examples. Such repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed.

[0087] To address the technical problem that existing driver assistance teaching methods rely on manual labor and have low levels of intelligence, resulting in poor teaching effectiveness, this application provides a driver assistance teaching method, device, electronic equipment, and storage medium. Based on a multi-dimensional data collection and analysis mechanism of the target vehicle, it achieves comprehensive monitoring and evaluation of driving behavior. Simultaneously, it enables intelligent and standardized analysis and evaluation of driving behavior, improving teaching quality and the ability to provide targeted guidance to learners.

[0088] Figure 1 A flowchart illustrating an embodiment of a driving assistance teaching method provided in this application includes the following steps:

[0089] Step 101: Obtain the driving behavior data of the target vehicle and the environmental perception data around the target vehicle.

[0090] Driving behavior data refers to various operational parameters generated by the driver during the operation of the target vehicle, as well as the vehicle's own operating status data. It is the core data reflecting the driver's actual driving habits and actions.

[0091] Environmental perception data refers to the external environment data of the driving scenario in which the target vehicle is located, including the distance between the vehicle and surrounding obstacles and other vehicles, road conditions, traffic flow and other external environmental information. It is an important basis for judging the driving scenario and formulating driving references.

[0092] In one embodiment, vehicle-mounted sensors or additional sensors are used to collect operational data such as starting / braking acceleration, steering wheel angle, vehicle speed, accelerator / brake pedal opening, steering angle, and lane change direction; simultaneously, vehicle status data such as vehicle height, vehicle weight, and tire pressure are collected. A camera detects the distance between the vehicle and surrounding obstacles, millimeter-wave radar detects and records the distance to vehicles / obstacles in front and behind, and ultrasonic radar assists in near-range environmental perception; all collected data is transmitted in real time and temporarily stored in a designated storage module to ensure data real-time performance and accuracy.

[0093] For example, suppose the collected user driving behavior data shows that the driver is driving on an urban road with the accelerator pedal open at 30%, the steering wheel turned 10° to the left, the vehicle speed at 40 km / h, the braking acceleration at 0.2G, and the tire pressure at 2.5 bar. The environmental perception data shows that the millimeter-wave radar detected a private car 5 meters directly in front of the target vehicle and an electric vehicle 3 meters to its left rear. The camera detected a traffic light intersection ahead, and the road is a two-lane road with no obstructions.

[0094] The aforementioned data collection process forms the data foundation for the entire driver assistance teaching method. Through multi-sensor fusion collection, it comprehensively, in real time, and accurately acquires data on the driver's actual driving operations and the external environment of the vehicle. This provides real, complete, and effective raw data support for subsequent reference data development, driving gap analysis, and teaching video generation, avoiding the problems of traditional driver training that rely solely on subjective human observation without quantitative data.

[0095] Step 102: Obtain reference driving trajectory data corresponding to the target vehicle.

[0096] Reference driving trajectory data refers to the ideal driving trajectory data of a vehicle that conforms to road rules and safe driving requirements, generated based on the driving route information selected by the driver of the target vehicle, combined with map information and real-time traffic conditions. It serves as the basis for subsequent development of driving reference data.

[0097] In one embodiment, the navigation module communicates with the map server in real time to receive the driving route information selected by the driver on the in-vehicle navigation / associated terminal; and generates basic reference driving trajectory data based on the real-time positioning data of the target vehicle and the road planning of the navigation route (such as lanes, turning points, ramps, etc.).

[0098] For example, when a driver selects a route from residential area A to shopping mall B, the navigation module combines data from Gaode / Baidu map servers to generate reference driving trajectory data.

[0099] Step 103: Determine driving reference data based on reference driving trajectory data and environmental perception data.

[0100] Among them, driving reference data is used to indicate the standard operating parameters corresponding to the safe driving of the target vehicle along the reference driving trajectory in the driving scenario represented by the environmental perception data.

[0101] Driving reference data refers to standard driving operation parameters that fit the current driving scenario and reference driving trajectory data. These are quantified standard parameters for the accelerator, brake, steering, lane changing, and other operations that drivers should follow to drive safely and in accordance with the reference driving trajectory data. They are also the core reference for subsequent comparative analysis of the driver's actual operation parameters.

[0102] Driving scenarios can refer to a combination of information, such as the road environment and traffic environment, that the target vehicle is currently in, as represented by environmental perception data. Examples include: going straight at a traffic light intersection on an urban road, merging into the main road from a highway ramp, and turning in a rainy parking garage.

[0103] In one embodiment, the intelligent driving module receives reference driving trajectory data and pre-stored environmental perception data from the navigation module; based on the hierarchical Gaussian process regression algorithm, combined with massive historical driving data (safe driving data in the same scenario and route as the current driving data) and real-time traffic conditions, it automatically identifies the current driving scenario mode (such as merging from a highway ramp, turning left at an urban intersection, etc.); according to the scenario mode and reference driving trajectory data, it generates a corresponding driving strategy and converts the driving strategy into quantified driving reference data (standard vehicle speed, acceleration, steering timing / angle, lane change timing / direction, etc.); and standardizes all the obtained driving reference data to prepare for subsequent gap analysis.

[0104] For example, suppose the target vehicle's driving scenario is merging from a highway ramp into the main road; and its corresponding environmental perception data is: the ramp is a right-turn ramp, the speed of vehicles on the main road is 80 km / h, and the distance to a truck on the main road 3 meters to the left rear of the target vehicle; its corresponding reference driving trajectory data can be: drive straight along the ramp to the dotted line and then merge into the main road. The intelligent driving module generates the following corresponding driving reference data based on the above data: standard ramp driving speed 30 km / h, accelerator pedal opening 20%, drive straight to the dotted line of the ramp (50 meters from the current position) and then turn right, gradually increasing the steering angle to 20°, accelerating to 60 km / h before merging into the main road, and maintaining a distance of ≥5 meters from the vehicle behind when changing lanes.

[0105] The above embodiments combine reference trajectories and the current driving scenario, and based on a hierarchical Gaussian process regression algorithm and massive historical safe driving data, generate driving reference data that fits the current driving scenario (such as highway ramps, rainy parking lots). Furthermore, the generated driving reference data undergoes standardization processing to ensure data consistency and comparability for subsequent comparative analysis with actual driving behavior data, avoiding analytical errors caused by differences in data formats and dimensions.

[0106] Step 104: Determine the driving gap data based on driving behavior data and driving reference data.

[0107] Driving gap data refers to the quantitative difference between a driver's actual driving behavior data and standardized driving reference data. It also includes qualitative conclusions such as the type of difference and the level of risk. It is the core data reflecting the driver's non-standard and unsafe driving operations.

[0108] In one embodiment, the driver's driving behavior data is spatiotemporally aligned with standardized driving reference data (synchronized by time axis and spatial location). Based on the resulting discrepancy data, multi-layered analysis is performed, outputting quantified discrepancy data from three dimensions: trajectory consistency, operational smoothness, and safety margin. Simultaneously, warning thresholds are set (e.g., triggering a warning if trajectory consistency < 85% or if sudden acceleration / deceleration occurs more than 3 times per minute). Based on these warning thresholds, the quantified discrepancy data is analyzed to generate corresponding qualitative conclusions.

[0109] For example, suppose the driver's actual driving behavior data (actual operation) is as follows: the driver does not proceed straight to the dashed line, but turns right 15° from the current position, accelerator pedal opening is 40%, vehicle speed is 40 km / h, and the distance to the truck on the left rear is 2.8 meters. The driving reference data (standard operation) is: proceed straight to the dashed line (50 meters later), turn right, steering angle is 20°, accelerator pedal opening is 20%, vehicle speed is 30 km / h, and the distance to the vehicle behind is ≥5 meters. Comparing the two, the resulting driving discrepancy data is: trajectory matching 70% (<85% warning), accelerator pedal opening deviation 20%, vehicle speed deviation 10 km / h, and insufficient lane change safety margin (distance difference to the vehicle behind is 2.2 meters). Based on the corresponding warning safety thresholds, the qualitative conclusions of the above driving discrepancy data are: triggering risks of premature steering, ramp speeding, and excessively close following distance during lane changes.

[0110] By comparing and analyzing the actual driving data of drivers with standard reference data, the gap between the two can be quantified and driving problems can be located; compared with relying on manual methods, it can more accurately identify the driver's non-standard and unsafe driving operations.

[0111] Furthermore, this application embodiment can also generate corresponding driving suggestions in real time by combining the corresponding driving gap data with the vehicle's environmental perception data after obtaining the corresponding driving gap data. Specifically, the in-vehicle system generates a prompt word based on at least one of the driving gap data, environmental perception data, and vehicle status data of the target vehicle; the prompt word is output to the AI ​​tool to obtain natural language driving suggestion text output by the AI ​​tool; wherein, the natural language driving suggestion text includes improvement measures and / or operation guidance based on the driving gap data.

[0112] For example, suppose the driving gap data is: risk of sudden braking, braking acceleration 0.8G; environmental perception data is: turning in a parking garage in the rain, slippery road surface, wall 5 meters ahead; vehicle status data is: current speed 25km / h, vehicle load 1.2 tons. Based on the preset prompt word template, the corresponding prompt words are constructed as follows: {Problem type: sudden braking, scenario: turning in a parking garage in the rain, slippery road surface, wall 5 meters ahead, vehicle status: speed 25km / h, load 1.2 tons, requirement: provide specific and actionable driving operation improvement measures and guidance for this problem}; AI tools (such as commonly used large language model tools) output corresponding natural language suggestions based on the above prompt words: The current parking garage road surface is slippery in the rain, sudden braking is likely to cause the vehicle to skid, it is recommended to immediately lightly release the brake pedal to slow down using a point braking method, anticipate road conditions 20 meters in advance and lightly apply the brakes, maintain a speed ≤15km / h, and when turning, the steering wheel angle ≤10° to avoid sudden steering. Specifically, suggestions can be broadcast in real time via in-vehicle voice prompts or displayed as highlighted text in a pop-up window on the central control screen.

[0113] Step 105: Generate driving instruction videos based on driving reference data and driving gap data.

[0114] Among them, the driving instruction videos are used to compare and demonstrate the differences between standard driving operations and user driving operations.

[0115] Driving instruction videos can refer to dynamically generated, visual driving guidance videos. Their core function is to compare and display the differences between standard operations and actual operations, and to highlight the driver's incorrect operations. This is more intuitive than text or voice suggestions, making it easier for drivers to understand and correct their mistakes.

[0116] In one embodiment, based on driving reference data, driving gap data and their corresponding qualitative conclusions, a virtual 3D scene of the current road and real-time environmental parameters are loaded; and a demonstration video containing the differences between the standard driving operation of the virtual vehicle and the user's driving operation is generated.

[0117] For example, if a driver fails to maintain a safe distance from the vehicle behind and fails to signal in advance when preparing to change lanes on an urban expressway, the system combines driving reference data for that scenario (standard lane changes require signaling 3 seconds in advance and maintaining a 50-meter safe distance from the vehicle behind) and driving difference data (no signal signal, only 20 meters from the vehicle behind), loads the 3D virtual scene of the expressway and real-time traffic flow parameters, and generates a teaching video. In the video, the virtual vehicle demonstrates the standard operation of signaling, observing the rearview mirror, and smoothly changing lanes. At the same time, the system highlights the actual lane-changing operation nodes of the driver in red, uses colored trajectory lines to compare and show the difference between the ideal lane-changing path and the actual path, and slows down and adds textual prompts for incorrect operations such as failing to maintain a safe distance and failing to signal.

[0118] In addition, this application embodiment can also generate a teaching annotation document that accompanies the teaching video, making it easier for drivers to learn quickly.

[0119] The above embodiments, through 3D virtual scene reconstruction and dynamic traffic flow simulation, highly restore the driver's actual driving scenario, making the teaching videos highly consistent with the driver's actual driving experience, and enhancing the immersion and comprehension of the teaching content.

[0120] The method provided in this application acquires environmental perception data around the target vehicle and reference driving trajectory data corresponding to the target vehicle's current driving route; determines driving reference data based on this data to instruct the target vehicle to drive safely along the reference driving trajectory; determines driving gap data based on the acquired driving behavior data and driving reference data of the target vehicle; and then generates corresponding driving instruction videos based on the driving reference data and driving gap data. By comparing driving behavior data and driving reference data, the method determines the driving gap data corresponding to the user's erroneous driving operations, eliminating the reliance on manual judgment of the student's erroneous driving operations and improving the error correction rate of the student's driving behavior. Furthermore, through personalized instruction videos, the abstract driving gap data is transformed into an intuitive visual experience, greatly enhancing the intuitiveness and attractiveness of the instruction, thereby significantly improving the teaching effect.

[0121] Furthermore, in addition to the above Figure 1 In addition to generating corresponding instructional videos, the embodiments described herein can also analyze and summarize users' driving habits based on driving behavior data and driving gap data over a period of time, and output corresponding driving behavior summary reports. Specifically, driving gap data of users driving the target vehicle can be obtained according to a preset period; the obtained driving gap data for the corresponding period can be analyzed, and a corresponding driving behavior summary report can be output based on the data analysis results; wherein, the driving behavior summary report includes common problems and improvement suggestions of users' driving behavior in the corresponding period.

[0122] For example, assuming a preset period of one week, a user's total driving mileage this week is 300km, with a total driving time of 8 hours. The main driving scenarios are urban roads (70%), underground parking garages (20%), and highways (10%). The full driving discrepancy data includes four types of problems: sudden acceleration / deceleration, excessively close following distance when changing lanes, trajectory deviation, and excessive steering angle. Data analysis is performed on the user's driving behavior data and driving discrepancy data during this period to obtain the analysis results. Among them, the high-frequency common problems include sudden acceleration / deceleration (12 times in total, mainly at urban traffic light intersections, triggering high risk 8 times) and excessively close following distance when changing lanes (8 times in total, mainly when overtaking on urban roads); the driving trend is that the trajectory deviation problem is gradually resolved, and the consistency rate increases from 70% to 85%.

[0123] Furthermore, based on the above data analysis results, a driving summary report is generated. This report may include a summary of the week's driving, such as a total mileage of 300km, with 70% on urban roads, and an overall pass rate of 78%. High-frequency common problems include frequent sudden acceleration and deceleration at urban traffic light intersections, resulting in numerous high-risk warnings; and frequent lane changes with insufficient following distance when overtaking in urban areas, leading to inadequate safety margins. Corresponding improvement suggestions are generated: when changing lanes, check the rearview mirror in advance, maintain a distance of ≥5 meters from the vehicle behind, and wait 3 seconds after signaling before changing lanes; at traffic light intersections, lightly apply the brakes 50 meters in advance to slow down, avoiding sudden acceleration / braking, and maintaining a stable accelerator pedal opening of 20%-30%. Next week, the focus should be on speed control and lane changing operations at urban traffic light intersections. It is recommended to review real-time driving suggestions after each drive and correct operational errors promptly.

[0124] By extracting common, high-risk issues from massive amounts of driving gap data through driving summary reports, and clearly marking the scenarios, frequencies, and risk levels of these issues, drivers can quickly identify their core driving weaknesses (e.g., recognizing that "sudden acceleration and deceleration mainly occur at urban traffic light intersections, and excessively close following distances during lane changes are concentrated when overtaking in urban areas"), avoiding blindly correcting mistakes and improving the targetedness and efficiency of driving improvement. Simultaneously, it can clearly display the changing trends of drivers' core driving problems (e.g., trajectory deviation accuracy improving from 70% to 85%), allowing drivers to intuitively see their driving progress and generating positive incentives; it can also promptly detect trends of worsening problems (e.g., a gradual increase in the number of sudden accelerations and decelerations), enabling preventative measures and making driving improvement more instructive and proactive.

[0125] Furthermore, the teaching effectiveness can be quantified through driving summary reports. The quantitative indicators in the reports, such as the overall pass rate of driving gaps, frequency of problems, and trajectory consistency, serve as both an assessment of the individual driver's driving skills and a quantitative feedback on the teaching effectiveness of the entire intelligent driving training system. The system can optimize the standards of driving reference data based on periodic reports from multiple users, allowing the entire teaching system to continuously iterate and upgrade in practice, thus achieving assessable and optimizable teaching effectiveness.

[0126] Figure 2 A flowchart illustrating another embodiment of the driving assistance teaching method provided in this application is shown below. Figure 1 Based on the illustrated process, this section primarily describes how to determine driving reference data based on reference driving trajectory data and environmental perception data; and how to determine driving gap data based on driving behavior data and driving reference data; including the following steps:

[0127] Step 201: Obtain the driving behavior data of the target vehicle and the environmental perception data around the target vehicle.

[0128] Step 202: Obtain reference driving trajectory data corresponding to the target vehicle's current driving route.

[0129] For steps 201-202 above, please refer to the above. Figure 1 The relevant description of the illustrated embodiment.

[0130] Step 203: Obtain the target vehicle's user's historical driving data set prior to this driving session.

[0131] Historical driving datasets can refer to all relevant data collected and stored by the system from the driver of the target vehicle during multiple past driving experiences. This includes driving behavior data (such as accelerator, brake, steering, vehicle speed, etc.), environmental perception data (road, traffic, weather, etc.), driving scenario labels, and possible gap analysis results.

[0132] In one embodiment, at the start of the current driving session, the system automatically retrieves the user's historical driving records to form a time-series data set, which facilitates subsequent analysis of the user's driving gap data and provides a corresponding data foundation for generating corresponding instructional videos.

[0133] For example, before user Zhang San started this driving session, the system obtained information that he had completed five driving practice sessions in the past week, and recorded detailed data for each session: the first session was on city roads during the day, the second session was on the highway in the rain, the third session was on a ramp at night, and so on.

[0134] Step 204: From the historical driving dataset, select target historical driving data that matches the driving scenario represented by the environmental perception data.

[0135] Driving scenarios can refer to driving situations comprehensively described by environmental perception data, which usually include features such as road type (highway, city, ramp), weather conditions (rain, snow, sunny), lighting conditions (daytime, nighttime), and traffic density (congestion, smooth flow).

[0136] Target historical driving data can refer to historical data segments selected from historical datasets that are similar to or in the same context as the current driving scenario, and are used to assist in personalized analysis.

[0137] In one embodiment, the system first performs scene recognition on the current environmental perception data and extracts key features (such as "rainy day," "highway," "ramp," and "night"). Then, it uses clustering algorithms or rule-based matching to search for data with the same or highly similar scene labels in the historical dataset. If multiple similar data exist, the most recent or highest-quality data can be selected as the target data.

[0138] For example, suppose the current scenario is merging onto a highway ramp at night in rainy weather. The system identifies one driving record from Zhang San's historical data that was also a nighttime, rainy highway ramp merging (three days ago), and one record of driving straight on the highway in rainy weather. The system selects the one that perfectly matches the scenario as the target historical driving data.

[0139] Step 205: Input the reference driving trajectory data, environmental perception data, and target historical driving data into the trained intelligent driving model to obtain the driving reference data output by the intelligent driving model.

[0140] Among them, driving reference data is used to indicate the standard operating parameters corresponding to the safe driving of the target vehicle along the reference driving trajectory in the driving scenario represented by the environmental perception data.

[0141] Intelligent driving models can refer to large, trained end-to-end models that can generate standard operating parameters for safe and smooth driving in the current scenario based on the current driving trajectory, real-time environment, and the driver's historical behavior.

[0142] In one embodiment, reference driving trajectory data, environmental perception data, and target historical driving data are input into a trained intelligent driving model. The model fuses the three types of data and performs spatiotemporal alignment and feature extraction on the input data to identify the current driving scenario (such as "merging from an urban expressway into a main road" or "driving at night in rainy weather"). Based on the scenario recognition results, the model extracts the optimal driving strategy for similar scenarios from a large amount of historical driving data to generate a set of driving reference data.

[0143] The driving reference data output by the intelligent driving model includes, but is not limited to: the current recommended speed range; the recommended acceleration / deceleration timing and magnitude; the recommended steering timing, steering angle, and steering trajectory; the recommended lane change timing and direction; the expected driving trajectory (including time series); and safety warning information (such as the suggestion to maintain a distance of more than 3 seconds).

[0144] For example, suppose the input reference driving trajectory data is: merging into the main road 200 meters ahead, needing to merge from the acceleration lane. Environmental perception data is: millimeter-wave radar detects a vehicle approaching from the left rear at 80 km / h, approximately 25 meters away; the camera identifies the current lane as the acceleration lane, with no vehicles ahead; the weather is clear and visibility is good. Historical driving data shows that in the past, this user often hesitated when merging into the main road, missing the optimal time and resulting in sudden acceleration or braking. The intelligent driving model identifies the current scenario as a "high-speed merging into the main road"; combining historical data, the model determines that the user tends to "delay acceleration" when merging; the model outputs the following driving reference data: the current speed should be gradually increased to 70 km / h; acceleration is recommended within 50 meters; the distance to the vehicle on the left rear is 25 meters, merging is recommended before the distance shortens to 15 meters; the steering angle is recommended to be controlled within 5° to maintain stability; the predicted trajectory is: drive along the acceleration lane to the dashed line area, then smoothly merge into the main lane.

[0145] Compared to traditional intelligent driving models, the aforementioned intelligent driving model can generate driving reference data. This data is used for comparative analysis, teaching, and suggestions, and does not directly output parameters for controlling the vehicle. Specifically, the intelligent driving model can combine the driving scenario of the target vehicle to output corresponding driving reference data to assist and guide the driver, rather than directly replacing the driver's operation to complete the autonomous driving of the target vehicle.

[0146] Step 206: Align the driving behavior data with the driving reference data on the time axis.

[0147] In one embodiment, the reference trajectory is divided into location points according to mileage. Then, the user's actual trajectory is mapped to the same location points according to mileage, and the user's operation value and reference operation value corresponding to that location point are extracted. For time-related operations (such as turn signal duration), time axis alignment can also be performed.

[0148] For example, the reference trajectory requires the user to begin turning at the dashed line on the ramp (150 meters from the ramp start). In actual driving, due to slower speed, the user arrives at the 150-meter position 2 seconds later than the reference time. After alignment, the system will not show any time discrepancy when comparing at the 150-meter position: the user's actual steering angle at this point is 10°, while the reference steering angle is 0° (because the reference requires no turning at this point and that turning should occur further ahead), thus determining the corresponding operational deviation.

[0149] Step 207: Compare the aligned driving behavior data with the driving reference data to obtain initial gap data containing at least one operational parameter deviation value.

[0150] Among them, the operating parameter deviation value is used to represent the deviation between the operating parameters corresponding to the user's driving behavior and the standard operating parameters in the driving reference data.

[0151] Operating parameter deviation values ​​can refer to the numerical differences between the user's actual operating values ​​and reference standard values ​​at each alignment point or critical event, such as speed difference, steering angle difference, pedal opening difference, lane change time difference, etc.

[0152] Initial gap data can refer to preliminary results consisting of a series of raw deviation values, which form the basis for further multidimensional analysis.

[0153] In one embodiment, the actual user operation parameters (vehicle speed, steering angle, pedal opening, etc.) and reference standard parameters at the same time point / location point are extracted; the difference of each operation parameter is calculated to obtain the single parameter deviation value; and the deviation values ​​of all operation parameters are integrated to form the initial difference data.

[0154] For example, at the 150-meter mark on the ramp (after time alignment), the reference standard is a steering angle of 0°, a vehicle speed of 50km / h, and a throttle opening of 15%. The user's actual operation is a steering angle of 10°, a vehicle speed of 40km / h, and a throttle opening of 5%. The initial difference data is: steering angle deviation +10°, vehicle speed deviation -10km / h, and throttle opening deviation -10%.

[0155] Step 208: Perform multi-dimensional feature analysis on the initial gap data to obtain driving gap data.

[0156] In one embodiment, multimodal feature analysis is performed on driving behavior data, environmental perception data, and initial gap data to obtain multidimensional deviation values ​​obtained from the multimodal feature analysis. The multidimensional deviation values ​​include at least one or more of trajectory deviation, operation smoothness deviation, and safe distance deviation. Multidimensional deviation values ​​that exceed the set safety threshold range are identified as driving gap data.

[0157] The aforementioned safety threshold range is set in the following ways: adjusting the preset safety threshold range based on the driving scenario represented by the environmental perception data; and / or obtaining the historical driving behavior data of the user driving the target vehicle before this driving; and adjusting the preset safety threshold range based on the historical driving behavior data.

[0158] The aforementioned multimodal feature analysis process includes: the feature extraction layer, which uses convolutional networks to extract the temporal features of driving behavior data and uses graph neural networks to construct the dynamic spatial relationship features between the target vehicle and surrounding obstacles; the spatiotemporal fusion layer, which uses recurrent neural networks to fuse the temporal features of the operation and the dynamic spatial relationship features to obtain the correlation features between the operation and the space; and the gap quantization layer, which outputs multidimensional deviation values ​​based on the correlation features.

[0159] Specifically, multimodal feature analysis is performed on driving behavior data, environmental perception data, and initial gap data. Through feature extraction layer, spatiotemporal fusion layer, and gap quantification layer, comprehensive deviation values ​​of dimensions such as trajectory, operation smoothness, and safe distance are output. Based on the current scenario (such as nighttime rainy highway) and user's historical driving data, the preset safety thresholds of each dimension are dynamically adjusted. Multi-dimensional deviation values ​​that exceed the safety thresholds are selected, namely driving gap data.

[0160] For example, assuming the current scenario is a highway ramp on a rainy night, the dynamically adjusted safety thresholds are: trajectory deviation ≤ 0.5 meters, steering angle change rate deviation ≤ 2° / s, and lateral distance deviation from the truck ≥ -0.3 meters. Through multimodal feature analysis, multidimensional deviation values ​​are obtained: trajectory deviation 0.8 meters, steering angle change rate deviation 3° / s, and lateral distance deviation -0.5 meters. All three exceed the thresholds, and these three sets of data are finally determined as driving difference data.

[0161] Step 209: Generate a driving instruction video based on the driving reference data and driving gap data; the driving instruction video is used to compare and demonstrate the differences between standard driving operations and user driving operations.

[0162] For step 209 above, please refer to the above. Figure 1 Detailed description of the relevant embodiments.

[0163] pass Figure 2 The description of the illustrated embodiment first eliminates comparison errors caused by differences in driving rhythm by aligning the time / position axis, ensuring the authenticity of the deviation value of a single operation parameter. Then, through multimodal feature analysis, the initial deviation data is comprehensively analyzed from core dimensions such as trajectory, operation smoothness, and safe distance, rather than simply judging the deviation of a single parameter. At the same time, the safety threshold is dynamically adjusted by combining the current driving scenario and the driver's historical data, avoiding the judgment bias of fixed thresholds in different scenarios and for different drivers. Finally, the selected driving difference data can accurately locate the core driving problems that the driver needs to correct, reducing the interference of invalid deviation information.

[0164] Figure 3 A flowchart illustrating another embodiment of the driving assistance teaching method provided in this application is shown below. Figure 1 Based on the illustrated process, this section mainly describes how to generate driving instruction videos using driving reference data and driving gap data, including the following steps:

[0165] Step 301: Based on the environmental perception data, reconstruct the virtual driving scene corresponding to the target vehicle's driving scene.

[0166] Virtual driving scenarios can refer to digital three-dimensional scenes generated by computer graphics technology based on real-world environmental perception data (such as road structure, traffic participants, weather, lighting, etc.) to simulate the driving environment in which the target vehicle is located at that time.

[0167] In one embodiment, a high-precision map (including static information such as lane lines, intersections, and ramps) is obtained from the navigation module. The location, speed, and type of dynamic traffic participants are extracted from camera and radar data. Real-time weather information is obtained from vehicle sensors or local weather data is obtained from the cloud. A 3D reconstruction engine is used to generate a virtual road image that is geometrically consistent with the real scene, and virtual vehicles and pedestrians are automatically added according to the real traffic flow density, while weather effects are superimposed to form a virtual environment that is highly similar to the original driving scene.

[0168] For example, a user merges into the main road from a highway ramp on a rainy night. The system generates 3D renderings of the ramp and the main road based on the navigation map, places a virtual truck at the corresponding location in the virtual scene (2.8 meters to the left rear) based on millimeter-wave radar data, and simulates the low visibility effect of a rainy night to reconstruct a virtual driving environment consistent with the real scene.

[0169] Step 302: Generate the user's actual operation trajectory in the virtual driving scenario based on driving behavior data.

[0170] Actual operation trajectory can refer to the movement path of a vehicle in a virtual scene, which can be deduced from driving behavior data (such as steering wheel angle, vehicle speed, acceleration, lane change signals, etc.). It is usually represented as a series of continuous spatial points or three-dimensional trajectory lines.

[0171] In one embodiment, based on the collected driving behavior data (timestamps, vehicle speed, heading angle, steering angle, etc.), the vehicle's position and attitude at each moment are calculated through integration. These positions are then mapped onto the virtual scene coordinate system reconstructed in step 301 to form a visualized trajectory line. Simultaneously, key operation points (such as braking points and steering points) can be recorded and their timestamps annotated.

[0172] For example, when a user merges into the aforementioned ramp, the steering wheel is continuously turned 15° to the right while the vehicle speed is 60 km / h. Based on this data, the system calculates the vehicle's actual driving path and draws a red trajectory line in the virtual scene, indicating that the vehicle merged to the right too early and was too close to the truck behind.

[0173] Step 303: In the virtual driving scenario, generate and display a demonstration screen of standard driving operation based on driving reference data, and use the first marker to compare and display the actual operation trajectory with the standard driving operation trajectory.

[0174] Standard driving operations can refer to the ideal operations that should be performed in the current scenario, generated by the intelligent driving module, including standard trajectory, recommended speed, and steering timing.

[0175] The demonstration screen can refer to the animation of the process of performing standard operations through a virtual vehicle.

[0176] The first marker can refer to visual identifiers used to distinguish between standard and actual trajectories, such as different colors (green for standard, red for actual), line type (solid / dashed), or dynamic arrows.

[0177] In one embodiment, a virtual vehicle representing a standard driver is generated in a virtual scene, and its movement is controlled according to driving reference data to demonstrate the correct operating procedure. At the same time, the actual operating trajectory generated in step 302 is displayed side by side or overlaid with the standard trajectory. For example, a green trajectory line represents the standard path, and a red trajectory line represents the actual path. Timestamps or directional arrows are displayed at key nodes (such as lane change starting points) to visually compare the differences between the two.

[0178] For example, in the ramp merging scenario, the standard demonstration vehicle activates its turn signal, accelerates, and smoothly merges into the main road (green trajectory line) upon reaching the dashed line area. Simultaneously, the user's actual red trajectory line is overlaid on the screen, clearly showing an premature rightward cut. The two trajectories separate before the dashed line, clearly demonstrating the user's operational error.

[0179] Step 304: Based on the location and type of deviation indicated by the driving gap data, overlay at least one of the second marker or slow-motion replay clips onto the demonstration screen to generate a driving instruction video.

[0180] The aforementioned driving instruction videos are used to compare and demonstrate the differences between standard driving operations and user driving operations.

[0181] The location of the deviation can refer to the spatiotemporal point marked in the gap data, such as insufficient steering angle 50 meters from the end of the ramp. The type of deviation can refer to the gap classification, such as understeering, excessive braking, or excessively short lane change distance.

[0182] The second marker can refer to warning visual elements, such as a flashing red box, an exclamation mark icon, or a flashing steering wheel icon, used to highlight erroneous operations. Slow-motion replay clips indicate that the period of user error is replayed at a reduced speed (e.g., 0.5x), with overlaid text descriptions or analysis.

[0183] In one embodiment, based on the results output by the driving gap analysis module, a dynamic warning marker is added at the location where the deviation occurs; the time period of the user's erroneous operation is captured, a slow-motion clip is generated, and explanatory text is superimposed on the screen; at the same time, the slow-motion of the standard operation at the same moment can be compared and displayed, and finally a complete teaching video is synthesized, which supports users to pause, rewind, and switch perspectives.

[0184] For example, in response to a user's mistake of turning too early, the video overlays a flashing yellow triangle warning sign where the red trajectory line begins to deviate, and pauses to display the text "Turning too early, only 2.8 meters from the vehicle behind." It then cuts to slow-motion replay: the user's turning process is replayed at 0.5x speed, while a standard vehicle's actions at the same time are shown alongside, with arrows indicating safe distances to help the user understand the cause of the error.

[0185] pass Figure 3 The description of the illustrated embodiment describes a scenario reconstruction and trajectory drawing based on the user's actual driving environment data and personal operation data. This generates a highly personalized, immersive learning experience, allowing users to reflect on their mistakes in a realistic manner, resulting in a learning effect far exceeding that of general theoretical instruction. Simultaneously, abstract driving gap data indicators such as insufficient steering angle or late braking are transformed into visual trajectory comparisons, lowering the learning threshold for students and allowing them to clearly see the deviation between their driving path and the ideal path. Furthermore, specific problems reflected in the driving gap data are highlighted and highlighted in the video, avoiding information overload and allowing users to clearly see every detail of their mistakes, reinforcing key learning points; thus significantly improving teaching effectiveness.

[0186] Figure 4 A structural block diagram of a driving assistance teaching device provided in this application, the device comprising:

[0187] The environmental perception data acquisition module 41 is used to acquire the driving behavior data of the target vehicle and the environmental perception data around the target vehicle.

[0188] The driving trajectory data acquisition module 42 is used to acquire reference driving trajectory data corresponding to the current driving route of the target vehicle.

[0189] The driving reference data determination module 43 is used to determine driving reference data based on the reference driving trajectory data and the environmental perception data; wherein, the driving reference data is used to indicate the standard operating parameters corresponding to the safe driving of the target vehicle along the reference driving trajectory under the driving scenario represented by the environmental perception data;

[0190] The driving gap data determination module 44 is used to determine driving gap data based on the driving behavior data and the driving reference data;

[0191] The driving instruction video generation module 45 is used to generate a driving instruction video based on the driving reference data and the driving gap data; wherein the driving instruction video is used to compare and demonstrate the differences between standard driving operations and user driving operations.

[0192] In one possible implementation, the driving reference data determination module 43 is specifically used for:

[0193] Obtain the historical driving dataset of the user of the target vehicle prior to this driving session;

[0194] From the historical driving dataset, target historical driving data that matches the driving scenario represented by the environmental perception data are selected;

[0195] The reference driving trajectory data, the environmental perception data, and the target historical driving data are input into the trained intelligent driving model to obtain the driving reference data output by the intelligent driving model.

[0196] In one possible implementation, the driving gap data determination module 44 includes:

[0197] A data alignment unit is used to align the driving behavior data with the driving reference data on the time axis.

[0198] The data comparison unit is used to compare the aligned driving behavior data with driving reference data to obtain initial difference data containing at least one operational parameter deviation value; wherein, the operational parameter deviation value is used to represent the deviation value between the operational parameter corresponding to the user's driving behavior and the standard operational parameter in the driving reference data;

[0199] The driving gap data acquisition unit is used to perform multi-dimensional feature analysis on the initial gap data to obtain driving gap data.

[0200] In one possible implementation, the driving gap data acquisition unit includes:

[0201] The feature analysis subunit is used to input the driving behavior data, the environmental perception data, and the initial gap data into the trained multimodal feature analysis model to obtain the multidimensional deviation value output by the multimodal feature analysis model. The multidimensional deviation value includes at least one or more of trajectory deviation, operation smoothness deviation, and safety distance deviation.

[0202] The driving gap data determination subunit identifies multi-dimensional deviation values ​​that exceed the set safety threshold range as driving gap data.

[0203] In one possible implementation, the multimodal feature analysis model in the feature analysis subunit is specifically used for:

[0204] The feature extraction layer is used to extract the operation time sequence features of the driving behavior data using a convolutional network, and to construct the dynamic spatial relationship features between the target vehicle and surrounding obstacles using a graph neural network.

[0205] The spatiotemporal fusion layer is used to fuse the operation temporal features and the dynamic spatial relationship features using a recurrent neural network to obtain the correlation features between operation and space.

[0206] The gap quantization layer is used to output multi-dimensional deviation values ​​based on the associated features.

[0207] In one possible implementation, the safety threshold range of the driving gap data determination subunit is set in the following manner:

[0208] Adjust the preset safety threshold range based on the driving scenario represented by the environmental perception data;

[0209] And / or,

[0210] Obtain the user's historical driving behavior data prior to this driving session; adjust the preset safety threshold range based on the historical driving behavior data.

[0211] In one possible implementation, the device further includes:

[0212] The prompt word generation module is used to generate prompt words based on at least one of the driving gap data, the environmental perception data, and the vehicle status data of the target vehicle;

[0213] The driving suggestion text determination module is used to input the prompt words into a trained large language model to obtain natural language driving suggestion text output by the large language model; wherein, the natural language driving suggestion text includes improvement measures and / or operation guidance for the driving gap data.

[0214] In one possible implementation, the driving instruction video generation module 45 is specifically used for:

[0215] Based on the environmental perception data, a virtual driving scenario corresponding to the driving scenario of the target vehicle is reconstructed;

[0216] Based on the driving behavior data, the user's actual operation trajectory in the virtual driving scenario is generated;

[0217] In the virtual driving scenario, a demonstration screen of standard driving operation is generated and displayed based on the driving reference data, and the actual operation trajectory is compared and displayed with the standard driving operation trajectory using a first marker.

[0218] Based on the location and type of deviation indicated by the driving gap data, at least one of a second marker or a slow-motion replay clip is superimposed on the demonstration screen to generate a driving instruction video.

[0219] In one possible implementation, the device further includes:

[0220] The driving gap data acquisition module is used to acquire driving gap data of the user driving the target vehicle within a preset period.

[0221] The driving report generation module is used to input the acquired driving gap data within the corresponding period into the trained data analysis model to obtain a driving behavior summary report output by the data analysis model.

[0222] like Figure 5 As shown in the figure, this application provides an electronic device, including a processor 111, a communication interface 112, a memory 113, and a communication bus 114, wherein the processor 111, the communication interface 112, and the memory 113 communicate with each other through the communication bus 114.

[0223] Memory 113 is used to store computer programs;

[0224] In one embodiment of this application, when the processor 111 executes the program stored in the memory 113, it implements the driving assistance teaching method provided in any of the foregoing method embodiments, including:

[0225] Acquire driving behavior data of the target vehicle and environmental perception data around the target vehicle;

[0226] Obtain reference driving trajectory data corresponding to the current driving route of the target vehicle;

[0227] Based on the reference driving trajectory data and the environmental perception data, driving reference data is determined; wherein, the driving reference data is used to indicate the standard operating parameters corresponding to the safe driving of the target vehicle along the reference driving trajectory under the driving scenario represented by the environmental perception data;

[0228] Based on the driving behavior data and the driving reference data, driving gap data is determined;

[0229] Based on the driving reference data and the driving gap data, a driving instruction video is generated; wherein, the driving instruction video is used to compare and demonstrate the differences between standard driving operations and user driving operations.

[0230] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the driving assistance teaching method provided in any of the foregoing method embodiments.

[0231] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0232] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented using software plus a general-purpose hardware platform, or of course, using hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0233] It should be understood that the terminology used herein is for the purpose of describing particular exemplary embodiments only and is not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “described” as used herein may also mean including the plural forms. The terms “comprising,” “including,” “containing,” and “having” are inclusive and therefore indicate the presence of the stated features, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, elements, components, and / or combinations thereof. The method steps, processes, and operations described herein are not construed as requiring them to be performed in a particular order described or illustrated unless the order of performance is explicitly indicated. It should also be understood that additional or alternative steps may be used.

[0234] The above description is merely a specific embodiment of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A driving assistance teaching method, characterized in that, The method includes: Acquire driving behavior data of the target vehicle and environmental perception data around the target vehicle; Obtain reference driving trajectory data corresponding to the target vehicle; Based on the reference driving trajectory data, the environmental perception data, and the historical driving behavior data of the user of the target vehicle prior to this driving, driving reference data is determined; wherein, the driving reference data is used to indicate the standard operating parameters corresponding to the target vehicle driving along the reference driving trajectory in the driving scenario represented by the environmental perception data; Based on the driving behavior data and the driving reference data, driving gap data is determined; A natural language driving suggestion text is generated based on at least one of the driving gap data, the environmental perception data, and the vehicle status data of the target vehicle; wherein the natural language driving suggestion text includes improvement measures and / or operation guidance for the driving gap data; Based on the driving reference data and the driving gap data, a driving instruction video is generated; wherein, the driving instruction video is used to compare and demonstrate the differences between standard driving operations and user driving operations; Based on the driving behavior data and the driving reference data, driving gap data is determined, including: Align the driving behavior data with the driving reference data on the time axis; compare the aligned driving behavior data with the driving reference data to obtain the initial gap data; Initial gap data that exceeds the set safety threshold range is identified as driving gap data; The safety threshold range is set in the following manner: Based on the driving scenario represented by the environmental perception data, adjust the preset safety threshold range; and / or, obtain the historical driving behavior data of the user driving the target vehicle before this driving; adjust the preset safety threshold range based on the historical driving behavior data.

2. The method according to claim 1, characterized in that, Based on the reference driving trajectory data and the environmental perception data, driving reference data is determined, including: Obtain the historical driving dataset of the user of the target vehicle prior to this driving session; From the historical driving dataset, target historical driving data that matches the driving scenario represented by the environmental perception data are selected; The reference driving trajectory data, the environmental perception data, and the target historical driving data are input into the trained intelligent driving model to obtain the driving reference data output by the intelligent driving model.

3. The method according to claim 1, characterized in that, Based on the driving behavior data and the driving reference data, driving gap data is determined, including: Align the driving behavior data with the driving reference data on the time axis; By comparing the aligned driving behavior data with driving reference data, initial gap data containing at least one operational parameter deviation value is obtained; wherein, the operational parameter deviation value is used to represent the deviation value between the operational parameter corresponding to the user's driving behavior and the standard operational parameter in the driving reference data; Multi-dimensional feature analysis is performed on the initial gap data to obtain driving gap data.

4. The method according to claim 3, characterized in that, Multi-dimensional feature analysis is performed on the initial gap data to obtain driving gap data, including: Based on the driving behavior data, the environmental perception data, and the initial gap data, a multi-dimensional deviation value is obtained, which includes at least one or more of trajectory deviation, operation smoothness deviation, and safe distance deviation. Multi-dimensional deviations that exceed the set safety threshold range are identified as driving gap data.

5. The method according to claim 1, characterized in that, Based on the driving reference data and the driving gap data, a driving instruction video is generated, including: Based on the environmental perception data, a virtual driving scenario corresponding to the driving scenario of the target vehicle is reconstructed; Based on the driving behavior data, the user's actual operation trajectory in the virtual driving scenario is generated; In the virtual driving scenario, a demonstration screen of standard driving operation is generated and displayed based on the driving reference data, and the actual operation trajectory is compared and displayed with the standard driving operation trajectory using a first marker. Based on the location and type of deviation indicated by the driving gap data, at least one of a second marker or a slow-motion replay clip is superimposed on the demonstration screen to generate a driving instruction video.

6. The method according to claim 1, characterized in that, The method further includes: According to a preset period, obtain driving difference data of the user driving the target vehicle within the corresponding period; The obtained driving gap data within the corresponding period is input into the trained data analysis model to obtain a driving behavior summary report output by the data analysis model.

7. A driving assistance teaching device, characterized in that, The device includes: An environmental perception data acquisition module is used to acquire driving behavior data of the target vehicle and environmental perception data around the target vehicle. The driving trajectory data acquisition module is used to acquire reference driving trajectory data corresponding to the current driving route of the target vehicle; The driving reference data determination module is used to determine driving reference data based on the reference driving trajectory data, the environmental perception data, and the historical driving behavior data of the user of the target vehicle before this driving; wherein, the driving reference data is used to indicate the standard operating parameters corresponding to the safe driving of the target vehicle along the reference driving trajectory in the driving scenario represented by the environmental perception data; The driving gap data determination module is used to determine driving gap data based on the driving behavior data and the driving reference data. The suggested text generation module is used to generate natural language driving suggestion text based on at least one of the driving gap data, the environmental perception data, and the vehicle status data of the target vehicle; wherein, the natural language driving suggestion text includes improvement measures and / or operation guidance for the driving gap data; The driving instruction video generation module is used to generate driving instruction videos based on the driving reference data and the driving gap data; wherein, the driving instruction videos are used to compare and demonstrate the differences between standard driving operations and user driving operations; Based on the driving behavior data and the driving reference data, driving gap data is determined, including: Align the driving behavior data with the driving reference data on the time axis; compare the aligned driving behavior data with the driving reference data to obtain the initial gap data; Initial gap data that exceeds the set safety threshold range is identified as driving gap data; The safety threshold range is set in the following manner: Based on the driving scenario represented by the environmental perception data, adjust the preset safety threshold range; and / or, obtain the historical driving behavior data of the user driving the target vehicle before this driving; adjust the preset safety threshold range based on the historical driving behavior data.

8. An electronic device, characterized in that, include: A processor and a memory, the processor being configured to execute a driving assistance teaching program stored in the memory to implement the driving assistance teaching method according to any one of claims 1-6.

9. A storage medium, characterized in that, The storage medium stores one or more programs, which can be executed by one or more processors to implement the driving assistance teaching method according to any one of claims 1-6.

Images