Driving training path generation and evaluation system and method based on garage location data modeling

Abstract

The invention discloses a driving training path generation and evaluation system and method based on garage location data modeling, belongs to the technical field of intelligent driving training, and is used for solving the problems of inconsistent driving training standards and insufficient subject path planning and evaluation accuracy. The system comprises a data acquisition module, a storage location modeling module, a spatial index module, a track generation module, a track evaluation module and a teaching guidance module. The method is implemented by adopting the system through the steps of information acquisition, storage location modeling, index construction, standard trajectory generation, actual trajectory deviation calculation and teaching guidance generation. According to the invention, through parking space accurate modeling and module cooperation, training standards are unified, the path evaluation precision and the teaching guidance pertinence are improved, and the driving training quality and efficiency are improved.

Description

Technical Field

[0001] This invention belongs to the field of intelligent driving training technology, specifically involving a driving training path generation and evaluation system and method based on parking space data modeling. It is applicable to path planning, driving behavior evaluation and intelligent teaching guidance for various training subjects such as reversing into a parking space, driving on a curved road, starting on a hill, turning at a right angle and parallel parking in driving training. Background Technology

[0002] In driver training, traditional teaching methods rely heavily on instructors' subjective experience for instruction and assessment. Due to a lack of standardized objective criteria, different instructors may have varying definitions of the "standard path" and "standard operation" for the same training subjects, such as reversing into a parking space, driving on a curved road, and starting on a hill. This leads to inconsistent training standards received by students, causing confusion, poor stability in the skills learning process, and overall low training efficiency.

[0003] Existing intelligent driving training systems mostly focus on simulating single scenarios or simply monitoring basic vehicle parameters (such as speed and position). They fail to design dedicated path generation and evaluation mechanisms for the spatial characteristics of different training subjects (such as the boundary lines of curved driving and the parking space boundaries for reversing into a parking space), and lack the ability to provide targeted guidance for diverse subjects. Furthermore, due to the lack of a complete data model of training site locations, standardized path planning based on the precise geometric features of the site cannot be achieved, and driving behavior evaluation also lacks spatial accuracy support, thus hindering the improvement of training quality.

[0004] Among existing technologies, there is a driving assessment system based on GPS trajectory recording. This system records the vehicle's driving trajectory using an onboard GPS device, compares the actual trajectory with a preset route, and generates a deviation report to support post-trip driving behavior analysis. However, because this system does not perform precise digital modeling of the training ground, the determination of the "standard path" still relies on the instructor's subjective experience. The teaching standards of different instructors vary significantly, resulting in inconsistent training benchmarks for trainees. Furthermore, it cannot achieve accurate assessment and real-time guidance for the spatial characteristics of different subjects, making it difficult to meet the needs of standardized and efficient driving training.

[0005] In summary, existing technologies have significant limitations in standardized teaching, precise path planning, and evaluation of driver training. There is an urgent need for a driver training path generation and evaluation system and method based on parking space data modeling to solve the above problems and improve the quality and efficiency of driver training. Summary of the Invention

[0006] In view of this, the present invention aims to overcome the shortcomings of the prior art and provide a driving training path generation and evaluation system and method based on parking space data modeling. This addresses the problems in the prior art, such as inconsistent driving training standards, lack of precise path planning and evaluation for different subjects, and insufficient targeted guidance. The system achieves accurate digital modeling of driving training grounds and parking spaces for various subjects, generates standardized training trajectories for each training subject based on the parking space data model, calculates and accurately evaluates the deviation between the student's actual driving trajectory and the standard trajectory in real time, and then generates targeted intelligent teaching guidance based on the evaluation results, thereby improving the quality and efficiency of driving training.

[0007] To achieve the above objectives, the present invention provides a first aspect of a driving training path generation and evaluation system based on parking space data modeling, comprising:

[0008] The data acquisition module is used to collect real-time information about the vehicle.

[0009] The storage location modeling module is used to digitize at least one training subject storage location in the driving training ground into a model containing the spatial features of the corresponding subject.

[0010] The spatial index module uses R-tree indexing technology to establish an index structure for storage location data and enable rapid matching of real-time vehicle location with corresponding storage location.

[0011] The trajectory generation module is used to generate standardized training trajectories for corresponding training subjects based on the model and driving specifications output by the parking location modeling module.

[0012] The trajectory evaluation module is used to calculate the deviation between the vehicle's actual driving trajectory and the standardized training trajectory.

[0013] The teaching guidance module is used to generate teaching guidance for the corresponding training subjects based on the deviation results of the trajectory evaluation module.

[0014] Furthermore, the real-time information of the vehicle includes its position, attitude, and speed.

[0015] Furthermore, the training subjects include at least one of reversing into a parking space, driving on a curved road, starting on a hill, making a right-angle turn, and parallel parking; the model output by the parking space modeling module is a polygonal model, and the spatial features include the boundary lines, guide lines, or key points of the corresponding subjects.

[0016] Furthermore, the trajectory evaluation module calculates the deviation between the vehicle's actual driving trajectory and the standardized training trajectory in real time, and the deviation includes at least one of lateral position deviation, heading deviation, and predicted trajectory deviation.

[0017] Furthermore, the teaching guidance generated by the teaching guidance module includes at least one of real-time guidance, predictive guidance, phased guidance, and subject-specific guidance.

[0018] A second aspect of this invention provides a method for generating and evaluating driving training paths based on parking space data modeling. This method employs the aforementioned system and includes the following steps:

[0019] S1: Collects real-time vehicle information;

[0020] S2: Digitize at least one training subject location in the driving training ground into a model containing the spatial features of the corresponding subject;

[0021] S3: Use R-tree indexing technology to establish an index structure for storage location data, enabling rapid matching of real-time vehicle location with corresponding storage location;

[0022] S4: Based on the model and driving specifications obtained in step S2, generate standardized training trajectories for the corresponding training subjects;

[0023] S5: Calculate the deviation between the vehicle's actual driving trajectory and the standardized training trajectory;

[0024] S6: Based on the deviation results from step S5, generate teaching guidance for the corresponding training subjects.

[0025] Furthermore, in step S1, the real-time information of the vehicle includes the vehicle's position, attitude, and speed; the actual driving trajectory of the vehicle is derived from the position and attitude information.

[0026] Furthermore, in step S2, the training subjects include at least one of reversing into a parking space, driving on a curved road, starting on a hill, making a right-angle turn, and parallel parking; the model is a polygonal model, and the spatial features include the boundary lines, guide lines, or key points of the corresponding subjects; in step S4, the generation of the standardized training trajectory also depends on the real-time vehicle information collected in step S1 and the R-tree index service established in step S3.

[0027] Furthermore, in step S5, the deviation between the vehicle's actual driving trajectory and the standardized training trajectory is calculated in real time. The deviation includes at least one of lateral position deviation, heading deviation, and predicted trajectory deviation. The R-tree index structure established in step S3 is used to help improve the spatial matching accuracy of the deviation calculation.

[0028] Furthermore, in step S6, the teaching guidance includes at least one of real-time guidance, predictive guidance, phased guidance, and subject-specific guidance; the generation logic of the teaching guidance is as follows: for lateral position deviation, priority is given to outputting direction correction guidance; for heading deviation, priority is given to outputting angle correction guidance; for predicted trajectory deviation, priority is given to outputting predictive operation guidance.

[0029] The present invention, by adopting the above technical solution, has at least the following beneficial effects:

[0030] In this invention, the parking space modeling module digitizes the parking spaces for driving training subjects such as reversing into parking spaces and driving on curves into polygonal models containing boundary lines, guide lines, and key points. Combined with the trajectory generation module, standardized training trajectories are generated based on the geometric features of the parking spaces, replacing the traditional "standard path" that relies on the subjective experience of the instructor. This unifies the training benchmarks for multiple subjects, solves the problem of student confusion caused by differences in teaching standards among different instructors, and ensures the stability of the skills learning process.

[0031] In this invention, the spatial indexing module uses R-tree indexing technology to achieve rapid matching of vehicle location and parking space data, and the trajectory evaluation module calculates multi-dimensional deviations such as lateral position, heading, and predicted trajectory in real time. It not only relies on the accurate parking space model to improve the spatial adaptability of standardized trajectories (such as conforming to the driving boundary of curves), but also strengthens the accuracy and timeliness of driving behavior evaluation through real-time, multi-dimensional calculation, breaking through the limitations of the rough and lagging evaluation of existing technologies.

[0032] In this invention, the teaching guidance module generates targeted guidance based on the type of deviation—suggestions for correcting the output direction of lateral position deviation, adjustment schemes for the output angle of heading deviation, and predictive operation prompts for the output of trajectory deviation. Combined with subject-specific guidance logic, it achieves intelligent intervention of "real-time error correction + advance prediction." Compared with the traditional single feedback mode, it significantly improves the adaptability and effectiveness of teaching guidance and accelerates the efficiency of students' skill mastery. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 This is a schematic diagram of the driving training path generation and evaluation system of the present invention;

[0035] Figure 2 The driving training path generation and evaluation method of this invention is a route Figure 1 ;

[0036] Figure 3 The driving training path generation and evaluation method of this invention is a route Figure 2 . Detailed Implementation

[0037] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention as detailed in the appended claims.

[0038] Example 1

[0039] like Figure 1 As shown, this embodiment provides a driving training path generation and evaluation system based on parking space data modeling, including:

[0040] The data acquisition module is used to collect real-time information about the vehicle.

[0041] The storage location modeling module is used to digitize at least one training subject storage location in the driving training ground into a model containing the spatial features of the corresponding subject.

[0042] The spatial index module uses R-tree indexing technology to establish an index structure for storage location data and enable rapid matching of real-time vehicle location with corresponding storage location.

[0043] The trajectory generation module is used to generate standardized training trajectories for corresponding training subjects based on the model and driving specifications output by the parking location modeling module.

[0044] The trajectory evaluation module is used to calculate the deviation between the vehicle's actual driving trajectory and the standardized training trajectory.

[0045] The teaching guidance module is used to generate teaching guidance for the corresponding training subjects based on the deviation results of the trajectory evaluation module.

[0046] As one implementation method, the real-time information of the vehicle in this embodiment includes the vehicle's position, attitude, and speed. The position information is obtained by fusing onboard GPS and inertial navigation, the attitude information is collected by a gyroscope, and the speed information is obtained by the onboard OBD interface, ensuring the real-time performance and accuracy of the data acquisition.

[0047] As one implementation method, the training subjects in this embodiment include at least one of reversing into a parking space, driving on a curved road, starting on a hill, making a right-angle turn, and parallel parking. The model output by the parking space modeling module is a polygon model, and the spatial features include the boundary lines, guide lines, or key points of the corresponding subjects. Specifically, the curved road parking space model includes a set of polygon coordinates of the left and right boundary lines and the center line; the reversing into a parking space and parallel parking models include the vertex coordinates of the parking space boundary lines and the entry guide lines; and the hill and right-angle turn models mark the path boundary lines and turning key points (such as the coordinates of the inner corner of a right-angle turn).

[0048] As one implementation method, the spatial indexing module described in this embodiment uses R-tree indexing technology to construct a multi-level spatial indexing structure from the polygon vertex coordinates of the parking space model. This supports millisecond-level matching between vehicle position and parking space, ensuring efficient positioning in multi-parking space scenarios (such as training grounds that simultaneously include reversing into parking spaces and curved driving areas), and providing real-time spatial correlation data for subsequent trajectory generation and evaluation.

[0049] As one implementation method, when the trajectory generation module in this embodiment generates a standardized training trajectory based on the parking space model, it combines the operation specifications of different subjects: for example, the standard trajectory for curved driving smoothly transitions along the center line to ensure that the vehicle body is equidistant from the two side boundaries; the reversing trajectory includes a continuous path of the starting point of entering the parking space, the first turning point, the second turning point, and the parking point, and the coordinates of each point are accurately calculated based on the parking space boundary.

[0050] As one implementation method, the trajectory evaluation module in this embodiment calculates the deviation between the actual driving trajectory of the vehicle and the standardized training trajectory in real time. The deviation includes at least one of lateral position deviation, heading deviation and predicted trajectory deviation. The lateral position deviation is the vertical distance between the actual trajectory and the standard trajectory, the heading deviation is the angle between the real-time heading angle of the vehicle and the tangent direction of the standard trajectory, and the predicted trajectory deviation is the degree of deviation between the path and the standard trajectory within 5 seconds based on the vehicle's current speed and attitude.

[0051] As one implementation method, the teaching guidance module in this embodiment generates teaching guidance including at least one of real-time guidance, predictive guidance, phased guidance, and subject-specific guidance. For example, when a lateral position deviation of more than 30cm is detected, a real-time prompt is given: "Turn the steering wheel 10 degrees to the left to adjust the distance between the vehicle body and the edge line." For heading deviation, the prompt is given: "The current front of the vehicle is 5 degrees to the right. Straighten the steering wheel by 2 degrees to correct." For predicted trajectory deviation, a warning is given 2 seconds in advance: "Straighten the steering wheel 5 meters ahead to avoid crossing the line." For the reversing parking subject, the subject-specific guidance includes subject-specific suggestions such as "The steering timing is too early. Delay the steering wheel by 0.5 seconds next time."

[0052] Example 2

[0053] like Figure 2 As shown, this embodiment provides a method for generating and evaluating driving training paths based on parking space data modeling. This method uses the above-mentioned system and includes the following steps:

[0054] S1: Collects real-time vehicle information;

[0055] S2: Digitize at least one training subject location in the driving training ground into a model containing the spatial features of the corresponding subject;

[0056] S3: Use R-tree indexing technology to establish an index structure for storage location data, enabling rapid matching of real-time vehicle location with corresponding storage location;

[0057] S4: Based on the model and driving specifications obtained in step S2, generate standardized training trajectories for the corresponding training subjects;

[0058] S5: Calculate the deviation between the vehicle's actual driving trajectory and the standardized training trajectory;

[0059] S6: Based on the deviation results from step S5, generate teaching guidance for the corresponding training subjects.

[0060] As one implementation method, in step S1 of this embodiment, the real-time information of the vehicle includes the vehicle's position, attitude, and speed; the actual driving trajectory of the vehicle is derived from the position and attitude information; wherein the position information is collected by fusing the vehicle-mounted GPS (positioning accuracy ±1 meter) and the inertial navigation module, with a sampling frequency of 10Hz, the attitude information (heading angle, pitch angle) is obtained by the vehicle-mounted gyroscope, and the speed information is read in real time through the OBD interface (update frequency 5Hz) to ensure the continuity of trajectory derivation.

[0061] As one implementation method, in step S2 of this embodiment, the training subjects include at least one of reversing into a parking space, driving on a curved road, starting on a hill, turning at a right angle, and parallel parking; the model is a polygonal model, and the spatial features include the boundary lines, guide lines, or key points of the corresponding subjects; specifically, in modeling, the curved driving parking space is scanned by LiDAR to generate left and right boundary lines and a center polygon composed of 100+ vertex coordinates; the reversing parking space is marked with the coordinates of the four corners of the parking space (forming a rectangular boundary), the entry guide line (a straight line from the starting point to the entrance of the parking space), and the coordinates of 3 turning reference points; the right-angle turning parking space specifies the coordinate parameters of the inner corner vertex, the entrance boundary line, and the turning trigger line; in step S4, the generation of the standardized training trajectory also relies on the real-time vehicle information collected in step S1 and the R-tree index service established in step S3.

[0062] As one implementation method, in step S3 of this embodiment, when using R-tree indexing technology to establish the index structure of the storage location data, the smallest bounding rectangle of the storage location model polygon is used as the index unit to construct a three-level index structure (root node → sub-region node → storage location node). By judging the spatial intersection between the real-time vehicle position coordinates and the index unit, the storage location matching response time is ≤50ms, ensuring that the current training subject can be quickly located in a multi-subject mixed training field (such as a reverse parking and curved driving area at the same time).

[0063] As one implementation method, in step S4 of this embodiment, when generating the standardized training trajectory, the subject operation specifications are combined: for example, the standard trajectory for curved driving is generated along the centerline, and a trajectory point is taken every 0.5 meters to ensure that the distance difference between the two sides of the vehicle body and the boundary line is ≤10cm; the reversing trajectory includes a continuous path of "starting point → first turning point (3 meters from the left front corner of the parking space) → second turning point (when the vehicle body is parallel to the edge line of the parking space) → parking point (1.5 meters from the rear boundary of the parking space)", and the coordinates of each point are accurately calculated based on the parking space model in step S2.

[0064] As one implementation method, in step S5 of this embodiment, the deviation between the actual driving trajectory of the vehicle and the standardized training trajectory is calculated in real time. The deviation includes at least one of lateral position deviation, heading deviation, and predicted trajectory deviation. The lateral position deviation is in meters (accuracy ±5cm), the heading deviation is in degrees (range -15° to +15°), and the predicted trajectory deviation is fitted by combining the current speed (predicting the path within 3 seconds when ≤5km / h, and predicting the path within 5 seconds when 5-10km / h) with the attitude angle. The R-tree index structure established in step S3 is used to quickly associate the vehicle position with the parking space boundary, which helps to improve the spatial matching accuracy of the deviation calculation (error ≤3cm).

[0065] As one implementation method, in step S6 of this embodiment, the teaching guidance includes at least one of real-time guidance, predictive guidance, phased guidance, and subject-specific guidance; the generation logic of the teaching guidance is as follows: when the lateral position deviation is >30cm, the real-time output is "Turn the steering wheel left / right by X degrees, adjust the distance between the vehicle body and the edge line to within 30cm"; when the heading deviation is >5°, the output is "The current front of the vehicle is deviated to the X side, straighten the steering wheel by Y degrees"; when the predicted trajectory deviation shows that it is about to cross the line, a warning is given 1.5 seconds in advance: "The vehicle will cross the line 2 meters ahead, it is recommended to straighten the steering wheel in advance"; for the parallel parking subject, the subject-specific guidance includes subject-specific operation prompts such as "Observe the right rearview mirror when reversing, and when the left rear corner of the parking space appears, turn the steering wheel fully to the right".

[0066] Example 3

[0067] like Figure 3 As shown, the system in this embodiment mainly includes the following modules:

[0068] 1. Data Acquisition Module

[0069] Core responsibility: To collect real-time information such as vehicle position, attitude, and speed;

[0070] Data output:

[0071] Transmit "real-time vehicle information as a reference" to the trajectory generation module (such as real-time speed to dynamically adjust the density of trajectory points, and to encrypt trajectory points in low-speed scenarios to ensure detail accuracy);

[0072] The trajectory evaluation module is directly provided with real-time information such as vehicle position, attitude, and speed (the actual driving trajectory is derived through position sequence fitting and attitude data correction, eliminating the drift error of a single sensor).

[0073] 2. Warehouse location modeling module

[0074] Core responsibility: To digitize the driving training ground's subject parking spaces (reverse parking, curved driving, hill start, etc.) into polygonal models;

[0075] Data output:

[0076] Provide the spatial indexing module with "digital polygon model data" (containing the coordinate set of subject boundary lines, guide lines, and key points, which serves as the geometric basis for building the R-tree index);

[0077] Transmit "basic data" (i.e., spatial characteristic constraints of the parking space, such as the boundary of the parking space for reversing into the parking space, the center line of the curved driving, and the spatial range of the standard trajectory) to the trajectory generation module.

[0078] 3. Spatial Index Module

[0079] Core responsibility: Based on the polygonal model of storage locations, build an R-tree index to support fast query and matching of vehicle locations;

[0080] Data output:

[0081] Provide the trajectory generation module with "index and location query function" (to help the standard trajectory fit the current storage location boundary, and achieve millisecond-level storage location matching in multi-subject scenarios, such as distinguishing the reversing area from the curved driving channel);

[0082] Provide the trajectory evaluation module with "auxiliary trajectory evaluation positioning" (enhancing the spatial correlation accuracy between the actual trajectory and the storage location, with a deviation of ≤3cm, ensuring the spatial accuracy of the evaluation results).

[0083] 4. Trajectory Generation Module

[0084] Core responsibility: To integrate multi-source data and generate standardized training trajectories for corresponding subjects;

[0085] Data input: Synchronously receive real-time information from the data acquisition module (dynamic adaptation), basic data from the storage location modeling module (spatial constraints), and index services from the spatial indexing module (location support);

[0086] Data output: Transmit "standard training trajectory data" (such as the continuous path of "starting point → turning point → parking point" for reversing into a parking space, and the smooth centerline trajectory for curved driving, as a benchmark template for deviation comparison) to the trajectory evaluation module.

[0087] 5. Trajectory Evaluation Module

[0088] Core responsibility: Real-time calculation of the deviation between the vehicle's actual driving trajectory and the standardized training trajectory;

[0089] Data input: integrates the actual trajectory derived from the data acquisition module, the standard trajectory from the trajectory generation module, and the positioning assistance from the spatial indexing module;

[0090] Data output: Provides "deviation assessment results data" to the teaching guidance module (covering lateral position deviation, heading deviation, and predicted trajectory deviation, with accuracies of ±5cm, ±0.5°, and 3-5 seconds respectively for path fitting prediction).

[0091] 6. Teaching Guidance Module

[0092] Core responsibility: To generate targeted teaching guidance based on deviation results;

[0093] Data input: Receive deviation data from the trajectory evaluation module;

[0094] Data output: Output "teaching guidance content" to relevant personnel (students, coaches) (including real-time correction, predictive warning, phased difficulty adaptation, subject-specific strategies, such as prompting directional correction when lateral deviation exceeds the threshold).

[0095] Through a hierarchical data flow of "basic data input → spatial positioning support → two-way trajectory comparison → intelligent guidance output," the system achieves a closed-loop logic of "digital modeling of storage locations → standardized trajectory generation → real-time deviation assessment → personalized teaching guidance," and is compatible with attached... Figure 3 The module interactions are fully aligned.

[0096] Regarding parking space data modeling: Different modeling methods are used for different subjects. For example, curved driving includes left and right boundary lines and the center line; reversing into a parking space and parallel parking include parking space boundaries and guide lines; ramps and right angles include path boundaries and key points. Regarding spatial indexing: With the help of spatial indexing, the system can quickly determine the parking space where the vehicle is located or is about to enter, thus laying the foundation for path generation and evaluation. Regarding standard trajectory generation: The system generates the optimal driving path based on the geometric characteristics of the parking space and driving regulations, which serves as a reference path for trainees. Regarding real-time trajectory deviation calculation: The deviations calculated by the system include lateral position deviation (i.e., the vertical distance between the vehicle and the standard trajectory), heading deviation (i.e., the angle difference between the vehicle's orientation and the tangent direction of the standard trajectory), and predicted trajectory deviation (i.e., the intersection of the vehicle's future path with the standard trajectory). In terms of intelligent teaching guidance generation: the system's teaching guidance features phased guidance, providing guidance of varying difficulty according to the student's training stage; it is real-time, generating guidance based on the real-time status of the vehicle; it is predictive, anticipating potential problems and providing guidance in advance; it is targeted, providing professional guidance tailored to the characteristics of different subjects; and it supports multiple subjects, with each subject having its own dedicated path generation and evaluation algorithm to ensure the professionalism and relevance of the teaching.

[0097] This embodiment focuses on the data interaction logic and functional collaboration mechanism between system modules, combined with the appendix. Figure 3 The data flow relationships are explained in detail to illustrate the system's operating principles.

[0098] I. Overview of Module Collaborative Architecture

[0099] The system consists of a data acquisition module, a storage location modeling module, a spatial indexing module, a trajectory generation module, a trajectory evaluation module, and a teaching guidance module. It forms a closed-loop collaboration through "data input → function processing → result output."

[0100] The storage location modeling module provides digital polygon model data to the spatial indexing module and provides the subject spatial feature basis to the trajectory generation module;

[0101] The data acquisition module provides the trajectory generation module with real-time vehicle information (position, attitude, speed) for reference, and provides the trajectory evaluation module with the data source for actual trajectory derivation;

[0102] The spatial indexing module provides R-tree indexing and location query functions to the trajectory generation module, and provides spatial positioning assistance to the trajectory evaluation module;

[0103] The trajectory generation module outputs standardized training trajectory data to the trajectory evaluation module;

[0104] The trajectory assessment module outputs deviation assessment results data to the teaching guidance module;

[0105] The teaching guidance module ultimately outputs intelligent teaching guidance content to students, coaches, and other relevant personnel.

[0106] II. Module Functions and Data Interaction Details

[0107] 1. Data Acquisition Module: Dynamic Information Capture

[0108] It is responsible for collecting real-time information such as vehicle position (GPS + inertial navigation fusion, positioning accuracy ±1m, sampling frequency 10Hz), attitude (gyroscope, heading angle accuracy ±0.5°), and speed (OBD interface, accuracy ±0.1km / h, update frequency 5Hz).

[0109] Transmit the "real-time vehicle status" to the trajectory generation module to assist in the dynamic adaptation and correction of the standard trajectory (e.g., when driving at low speed, the trajectory point density is increased to 0.2m / point to ensure accurate details);

[0110] The vehicle motion data is transmitted to the trajectory evaluation module, and the actual driving trajectory is derived by fitting the position sequence and correcting the attitude data (eliminating the drift error of single GPS positioning).

[0111] 2. Storage Location Modeling Module: Digitalization of Account Space

[0112] For maneuvers such as reversing into a parking space, driving on a curved road, starting on a hill, making a right-angle turn, and parallel parking, the parking space is converted into a polygon model:

[0113] Curved driving: Mark the left and right boundary lines (each composed of ≥50 vertex coordinates, with a vertex interval of 0.2m) and the center line (the vertex sequence is fitted to a smooth curve) to form a "channel-type" polygon;

[0114] Reverse parking / parallel parking: Define the parking space boundary (rectangle vertex coordinates, accuracy ±2cm) and the parking guide line (straight line parameters from the starting point coordinates to the parking space entrance, length error ≤5cm);

[0115] Ramp / Right-Angle Turn: Mark the path boundary lines (such as the coordinates of the vertices of the two sides of the ramp) and key trigger points (such as the vertex of the inner corner of the right-angle turn, with a coordinate error ≤ 3cm).

[0116] The output polygon model data is simultaneously supplied to the spatial indexing module (used to build the R-tree index) and the trajectory generation module (as spatial constraints for trajectory planning).

[0117] 3. Spatial Indexing Module: Fast Localization Driven by R Tree

[0118] Using R-tree indexing technology, a three-level index structure of "root node → sub-region node → storage location node" is constructed, with the minimum bounding rectangle (MBR) of the storage location polygon as the index unit.

[0119] When a vehicle enters the training ground, the spatial intersection of its real-time location coordinates and the index unit is used to determine the current / about-to-enter parking space within 50ms (e.g., distinguishing between the reversing parking area and the curved driving lane).

[0120] The system provides an "index query service" to the trajectory generation module (to help the standard trajectory fit the current storage location boundary and avoid the trajectory going out of bounds), and provides a "spatial positioning assistance" to the trajectory evaluation module (to improve the correlation accuracy between the actual trajectory and the storage location, with an error of ≤3cm).

[0121] 4. Trajectory Generation Module: Standard Trajectory Planning

[0122] Based on the polygonal model of the parking space (spatial constraints), driving regulations (operational standards), and real-time vehicle information (dynamic adaptation), a standardized training trajectory is generated:

[0123] Curved driving: Generate a "smooth transition trajectory" along the centerline, taking one trajectory point every 0.5m to ensure that the difference between the vehicle body and the two side boundaries is ≤10cm (verified by the storage location boundary data of the spatial index module);

[0124] Reversing into a parking space: Plan a continuous path from "starting point → first turning point (3m from the left front corner of the parking space, error ≤ 5cm) → second turning point (triggered by attitude data when the vehicle body is parallel to the side line) → parking point (1.5m from the rear boundary of the parking space, error ≤ 10cm)". The coordinates of each point are accurately calculated based on the boundary of the parking space.

[0125] The output standard trajectory data serves as the "ideal path benchmark" for the trajectory evaluation module.

[0126] 5. Trajectory Evaluation Module: Multi-dimensional Deviation Calculation

[0127] The actual driving trajectory (derived by the data acquisition module) is compared with the standard training trajectory (output by the trajectory generation module) in real time, and three types of deviations are calculated:

[0128] Lateral position deviation: The vertical distance between the actual trajectory and the standard trajectory (accuracy ±5cm, warning triggered if it exceeds 30cm);

[0129] Heading deviation: The angle between the vehicle's real-time heading angle and the tangent direction of the standard trajectory (range -15° to +15°, exceeding 5° triggers a correction prompt);

[0130] Predicted trajectory deviation: Based on the current speed and attitude, the path is fitted for the next 3 to 5 seconds (3 seconds for low speed ≤ 5 km / h, 5 seconds for medium speed 5 to 10 km / h) to determine the risk of crossing the standard trajectory (early warning of potential problems such as crossing the line or going out of bounds).

[0131] The output deviation assessment results serve as the "intervention basis" for the teaching guidance module.

[0132] 6. Teaching Guidance Module: Intelligent Intervention Output

[0133] Based on the type and degree of deviation, and the characteristics of the subject, four types of guidance content are generated:

[0134] Real-time guidance: such as "lateral deviation 40cm → turn the steering wheel 10 degrees to the left", with a second-level response based on the vehicle's real-time status;

[0135] Foresight guidance: such as "predicting that the line will be crossed in 5 seconds → correcting the direction by 2 degrees in advance", relying on the predicted trajectory deviation to achieve "prevention before the event";

[0136] Phased guidance: Adjust the difficulty of guidance according to the training phase (focus on "basic operation steps" in the initial practice, and strengthen "detail precision correction" in the proficiency stage);

[0137] Subject-specific guidance: For example, for parallel parking, "observe the right rearview mirror, and turn the steering wheel fully to the right when the left rear corner of the parking space appears," which is in line with the specific operational logic of the subject.

[0138] This embodiment achieves a complete training logic of "digitalization of storage locations → standardization of trajectories → precision of assessment → intelligent guidance" through data closure and functional collaboration between modules. Relying on technologies such as R-tree indexing, polygon modeling, and multi-dimensional deviation calculation, it breaks through the "experience dependence" of traditional teaching and supports the standardization, efficiency, and personalization of driver training.

[0139] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A driving training path generation and evaluation system based on parking space data modeling, characterized in that: include: The data acquisition module is used to collect real-time information about the vehicle. The storage location modeling module is used to digitize at least one training subject storage location in the driving training ground into a model containing the spatial features of the corresponding subject. The spatial index module uses R-tree indexing technology to establish an index structure for storage location data and enable rapid matching of real-time vehicle location with corresponding storage location. The trajectory generation module is used to generate standardized training trajectories for corresponding training subjects based on the model and driving specifications output by the parking location modeling module. The trajectory evaluation module is used to calculate the deviation between the vehicle's actual driving trajectory and the standardized training trajectory. The teaching guidance module is used to generate teaching guidance for the corresponding training subjects based on the deviation results of the trajectory evaluation module.

2. The system according to claim 1, characterized in that: The real-time information of the vehicle includes its position, attitude, and speed.

3. The system according to claim 1, characterized in that: The training subjects include at least one of reversing into a parking space, driving on a curved road, starting on a hill, making a right-angle turn, and parallel parking; the model output by the parking space modeling module is a polygonal model, and the spatial features include the boundary lines, guide lines, or key points of the corresponding subjects.

4. The system according to claim 1, characterized in that: The trajectory evaluation module calculates the deviation between the vehicle's actual driving trajectory and the standardized training trajectory in real time. The deviation includes at least one of lateral position deviation, heading deviation, and predicted trajectory deviation.

5. The system according to claim 1, characterized in that: The teaching guidance generated by the teaching guidance module includes at least one of the following: real-time guidance, predictive guidance, phased guidance, and subject-specific guidance.

6. A method for generating and evaluating driving training paths based on parking space data modeling, characterized in that: This method employs the system described in any one of claims 1 to 5. Includes the following steps: S1: Collects real-time vehicle information; S2: Digitize at least one training subject location in the driving training ground into a model containing the spatial features of the corresponding subject; S3: Use R-tree indexing technology to establish an index structure for storage location data, enabling rapid matching of real-time vehicle location with corresponding storage location; S4: Based on the model and driving specifications obtained in step S2, generate standardized training trajectories for the corresponding training subjects; S5: Calculate the deviation between the vehicle's actual driving trajectory and the standardized training trajectory; S6: Based on the deviation results from step S5, generate teaching guidance for the corresponding training subjects.

7. The method according to claim 6, characterized in that: In step S1, the real-time information of the vehicle includes the vehicle's position, attitude, and speed; the actual driving trajectory of the vehicle is derived from the position and attitude information.

8. The method according to claim 6, characterized in that: In step S2, the training subjects include at least one of reversing into a parking space, driving on a curved road, starting on a hill, turning at a right angle, and parallel parking; the model is a polygonal model, and the spatial features include the boundary lines, guide lines, or key points of the corresponding subjects; in step S4, the generation of the standardized training trajectory also depends on the real-time vehicle information collected in step S1 and the R-tree index service established in step S3.

9. The method according to claim 6, characterized in that: In step S5, the deviation between the vehicle's actual driving trajectory and the standardized training trajectory is calculated in real time. The deviation includes at least one of lateral position deviation, heading deviation, and predicted trajectory deviation. The R-tree index structure established in step S3 is used to help improve the spatial matching accuracy of the deviation calculation.

10. The method according to claim 6, characterized in that: In step S6, the teaching guidance includes at least one of real-time guidance, predictive guidance, phased guidance, and subject-specific guidance; the generation logic of the teaching guidance is as follows: for lateral position deviation, priority is given to outputting direction correction guidance; for heading deviation, priority is given to outputting angle correction guidance; for predicted trajectory deviation, priority is given to outputting predictive operation guidance.

Images