Unmanned vehicle control method and system based on artificial intelligence

Abstract

The invention is suitable for the technical field of unmanned vehicle path control and experience optimization, and provides an unmanned vehicle control method and system based on artificial intelligence. And continuously collecting local driving data corresponding to the convergent change road condition of the passing path. According to the method, a score correction mechanism based on the association trend between the path structure convergence strength and the passenger experience feeling response is constructed, so that the dynamic optimization of the path selection score of the unmanned vehicle is realized. The subjective experience index decreasing amplitude of the passenger individuals in the path convergence change process is introduced for the first time to serve as the correction basis, the weighted adjustment model of the score value is constructed in combination with the change trend of the scenic spot tourist value, a sensitive structure with the more severe convergence and the poorer experience can be recognized, and the method has the advantages of being high in practicability and easy to popularize. Therefore, the scoring result is corrected in accordance with actual perception.

Description

Technical Field

[0001] The present invention belongs to the technical field of path control and experience optimization of driverless vehicles, and particularly relates to a control method and system for driverless vehicles based on artificial intelligence. Background Art

[0002] In current driverless vehicle path planning technologies, static map data, real-time traffic information, path passing efficiency, etc. are generally used as decision-making bases. By calculating indicators such as path length, time consumption, and road condition level, path selection scores are generated, and path priority sorting is carried out accordingly. This type of scoring system has certain effects in improving passing efficiency and avoiding congested paths, and is particularly suitable for driverless application scenarios such as daily commuting or logistics transportation. However, in scenarios centered on passenger experience such as scenic area tours and city sightseeing, traditional scoring mechanisms often ignore the direct impact of path structure on passengers' perception. Especially during the process where the path structure suddenly narrows from a spacious area to a narrow space, it is likely to trigger emotional reactions such as tension, oppression, or discomfort in passengers, resulting in an obvious deviation between the actual experience and the scoring result.

[0003] In addition, existing technologies usually fail to effectively perceive or model passengers' subjective feedback on path changes, and also lack a dynamic adjustment mechanism for introducing passenger experience indices into the scoring system. In specific scenarios, even if a certain path performs excellently in terms of efficiency and geographical factors, if its structure has abrupt changes, it may still cause negative experiences for passengers. However, since the scoring system cannot perceive such emotional reactions, the path priority sorting result cannot truly reflect users' feelings, restricting the promotion effect and user satisfaction of driverless systems in "experience-oriented" applications. Summary of the Invention

[0004] The purpose of the present invention is to provide a control method and system for driverless vehicles based on artificial intelligence, aiming to solve the problems raised in the background art.

[0005] The present invention is implemented as follows. A control method for a driverless vehicle based on artificial intelligence, the method includes:

[0006] During the process of a driverless vehicle carrying current passengers for a tour, continuously collect local driving data corresponding to the road conditions of the experienced path convergence changes;

[0007] After the collected local driving data reaches a preset quantity, when the driverless vehicle is about to enter a new path convergence change road condition, obtain the initial path selection score value corresponding to this path convergence change road condition, and set it as the score value to be corrected;

[0008] Based on the collected local driving data, extract several pieces of target data with gradually increasing path convergence intensities therefrom;

[0009] Parse the experience index of the current passenger in each target data before and after the path convergence change respectively, and calculate the corresponding decrease in the experience index.

[0010] Analyze whether the decrease in the experience index shows an increasing trend as the path convergence intensity increases. If so, screen out several reference data whose path convergence intensity is within the same preset range as the current path convergence change road condition.

[0011] Based on several reference data, determine the comprehensive decrease in the experience index, and obtain the corresponding comprehensive decrease in the scenic spot tour value. Adjust the score value to be corrected according to the comprehensive decrease in the experience index and the comprehensive decrease in the scenic spot tour value.

[0012] As a further limitation of the technical solution of the embodiment of the present invention, the path convergence change road condition refers to the typical road condition where the driverless vehicle drives from an area with ample space into an area with narrow space.

[0013] As a further limitation of the technical solution of the embodiment of the present invention, the step of extracting several target data with gradually increasing path convergence intensity from the collected local driving data includes:

[0014] Parse each local driving data in sequence, and determine the first passing space envelope degree parameter of the driverless vehicle before entering the path convergence change road condition and the second passing space envelope degree parameter after entering the path convergence change road condition.

[0015] Calculate the difference between the first passing space envelope degree parameter and the second passing space envelope degree parameter to obtain the path convergence intensity corresponding to the local driving data.

[0016] After calculating the path convergence intensity corresponding to each local driving data, screen several target data in the order of gradually increasing path convergence intensity.

[0017] As a further limitation of the technical solution of the embodiment of the present invention, the step of parsing the experience index of the current passenger in each target data before and after the path convergence change respectively, and calculating the corresponding decrease in the experience index includes:

[0018] Obtain the passenger face video images of the current passenger before and after entering the path convergence change road condition in each target data respectively.

[0019] Based on the passenger face video images, use the image recognition algorithm to determine the first experience index before entering the path convergence change road condition and the second experience index after entering respectively.

[0020] Calculate the decrease in the experience index according to the first experience index and the second experience index.

[0021] As a further limitation of the technical solution of the embodiment of the present invention, based on a number of reference data, determining the decline rate of the comprehensive experience index, and obtaining the corresponding decline rate of the comprehensive scenic spot tour value, the steps of jointly adjusting the score value to be corrected according to the decline rate of the comprehensive experience index and the decline rate of the comprehensive scenic spot tour value include:

[0022] Calculating the average value of the decline rates of the corresponding experience indexes in all the reference data to obtain the decline rate of the comprehensive experience index;

[0023] Obtaining the preset tour value reference table of the scenic spot where the driverless vehicle is located, respectively determining the first scenic spot tour value and the second scenic spot tour value before and after the driverless vehicle enters the path convergence and change road conditions in each reference data, and calculating the corresponding decline rate of the scenic spot tour value;

[0024] Taking the average of the decline rates of the scenic spot tour values of all the reference data to obtain the decline rate of the comprehensive scenic spot tour value;

[0025] Based on the decline rate of the comprehensive experience index and the decline rate of the comprehensive scenic spot tour value, performing weighted correction on the score value to be corrected according to the preset correction function to obtain the corrected path selection score value;

[0026] Applying the corrected path selection score value to the decision priority evaluation and path instruction generation of the current path convergence and change road conditions.

[0027] As a further limitation of the technical solution of the embodiment of the present invention, the preset correction function is:

[0028] ;

[0029] Wherein, refers to the corrected path selection score value, refers to the score value to be corrected, refers to the total number of reference data, and respectively refer to the second experience index and the first experience index after and before the driverless vehicle enters the path convergence and change road conditions in the th reference data, refers to the th reference data corresponding decline rate of the experience index, refers to the decline rate of the comprehensive experience index, and respectively refer to the second scenic spot tour value and the first scenic spot tour value after and before the driverless vehicle enters the path convergence and change road conditions in the th reference data, refers to the th reference data corresponding decline rate of the scenic spot tour value, Refers to the decrease in the comprehensive scenic spot travel value, and are the adjustment weights for the decrease in the comprehensive experience index and the decrease in the comprehensive scenic spot travel value respectively, and and are both greater than 0.

[0030] A control system for a driverless vehicle based on artificial intelligence, the system includes: a data acquisition module, a score to be corrected determination module, a target data extraction module, an experience decrease calculation module, a reference data screening module, a score to be corrected adjustment module, where:

[0031] The data acquisition module is used to continuously collect local driving data corresponding to the road conditions of the path convergence change during the process of the driverless vehicle carrying the current passenger for sightseeing; the path convergence change road conditions refer to the typical road conditions where the driverless vehicle drives from an area with ample space into an area with narrow space;

[0032] The score to be corrected determination module is used to, after the collected local driving data reaches a preset quantity, when the driverless vehicle is about to enter a new path convergence change road condition, obtain the initial path selection score value corresponding to the path convergence change road condition and set it as the score to be corrected value;

[0033] The target data extraction module is used to extract several pieces of target data with gradually increasing path convergence intensity from the collected local driving data;

[0034] The experience decrease calculation module is used to respectively analyze the experience index of the current passenger before and after the path convergence change in each target data and calculate the corresponding experience index decrease;

[0035] The reference data screening module is used to analyze whether the experience index decrease shows an increasing trend as the path convergence intensity increases. If so, screen out several reference data whose path convergence intensity is within the same preset range as the path convergence intensity of the current path convergence change road condition;

[0036] The score to be corrected adjustment module is used to determine the comprehensive experience index decrease based on several reference data, obtain the corresponding decrease in the comprehensive scenic spot travel value, and jointly adjust the score to be corrected value according to the comprehensive experience index decrease and the decrease in the comprehensive scenic spot travel value.

[0037] As a further limitation of the technical solution of the embodiment of the present invention, the target data extraction module specifically includes:

[0038] The degree of freedom parameter determination unit is used to sequentially analyze each local driving data and determine the first passing space envelope degree parameter of the driverless vehicle before entering the path convergence change road condition and the second passing space envelope degree parameter after entering the path convergence change road condition;

[0039] A path convergence intensity calculation unit, configured to calculate the difference between the first traffic space envelope degree parameter and the second traffic space envelope degree parameter to obtain the path convergence intensity corresponding to the local driving data;

[0040] A target data screening unit, configured to screen a plurality of target data in the order of increasing path convergence intensity after calculating the path convergence intensity corresponding to each local driving data.

[0041] As a further limitation of the technical solution of the embodiment of the present invention, the experience reduction calculation module specifically includes:

[0042] A facial information acquisition unit, configured to respectively acquire the passenger facial video images of the current passenger before and after entering the path convergence changing road condition in each target data;

[0043] An experience index determination unit, configured to respectively determine a first experience index before entering the path convergence changing road condition and a second experience index after entering based on the passenger facial video images by using an image recognition algorithm;

[0044] An experience reduction calculation unit, configured to calculate the experience index reduction according to the first experience index and the second experience index.

[0045] As a further limitation of the technical solution of the embodiment of the present invention, the to-be-corrected score adjustment module specifically includes:

[0046] A first comprehensive reduction calculation unit, configured to calculate the average value of the experience index reductions corresponding to all reference data to obtain a comprehensive experience index reduction;

[0047] A scenic spot tour value acquisition unit, configured to acquire a preset tour value reference table of the scenic spot where the driverless vehicle is located, respectively determine a first scenic spot tour value and a second scenic spot tour value before and after the driverless vehicle enters the path convergence changing road condition in each reference data, and calculate the corresponding scenic spot tour value reduction;

[0048] A second comprehensive reduction calculation unit, configured to take the average of the scenic spot tour value reductions of all reference data to obtain a comprehensive scenic spot tour value reduction;

[0049] A to-be-corrected score adjustment unit, configured to perform weighted correction on the to-be-corrected score value according to a preset correction function based on the comprehensive experience index reduction and the comprehensive scenic spot tour value reduction to obtain a corrected path selection score value;

[0050] A corrected score value application unit, configured to apply the corrected path selection score value to the decision priority evaluation and path instruction generation of the current path convergence changing road condition;

[0051] The preset correction function is as follows:

[0052] ;

[0053] Wherein, refers to the corrected path selection score value, refers to the score value to be corrected, refers to the total number of reference data, and respectively refer to the second experience index and the first experience index after and before the unmanned vehicle enters the path convergence change road condition in the th reference data, refers to the th reference data corresponding experience index decline, refers to the comprehensive experience index decline, and respectively refer to the second scenic spot tour value and the first scenic spot tour value after and before the unmanned vehicle enters the path convergence change road condition in the th reference data, refers to the th reference data corresponding scenic spot tour value decline, refers to the comprehensive scenic spot tour value decline, and are respectively the adjustment weights of the comprehensive experience index decline and the comprehensive scenic spot tour value decline, and and are both greater than 0.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] By constructing a scoring correction mechanism based on the correlation trend between the path structure convergence strength and the passenger experience response, the present invention realizes the dynamic optimization of the path selection score of the unmanned vehicle. Compared with the prior art that only scores based on static maps, traffic efficiency and other factors, the present invention for the first time introduces the subjective experience index decline of passengers during the path convergence change process as the correction basis, and combines the change trend of the scenic spot tour value to construct a weighted adjustment model of the score value, which can identify the sensitive structure of "the more intense the convergence, the worse the experience", so as to correct the scoring result more in line with the actual perception. This method effectively enhances the expression ability of the path planning system to balance the passenger psychological comfort and the scene value, has higher intelligent adaptability and experience optimization value, and is particularly suitable for sightseeing and experience-oriented unmanned application scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 is a flowchart of the method provided by the embodiment of the present invention;

[0057] Figure 2 It is a flowchart for obtaining several pieces of target data with gradually increasing path convergence intensity in the method provided by the embodiment of the present invention;

[0058] Figure 3 It is a flowchart for calculating the reduction rate of the experience index in the method provided by the embodiment of the present invention;

[0059] Figure 4 It is a flowchart for weighted correction of the score value to be corrected in the method provided by the embodiment of the present invention;

[0060] Figure 5 It is an application architecture diagram of the system provided by the embodiment of the present invention;

[0061] Figure 6 It is a structural block diagram of the target data extraction module in the system provided by the embodiment of the present invention;

[0062] Figure 7 It is a structural block diagram of the experience reduction rate calculation module in the system provided by the embodiment of the present invention;

[0063] Figure 8 It is a structural block diagram of the score to be corrected adjustment module in the system provided by the embodiment of the present invention. Detailed implementation manners

[0064] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0065] Figure 1 It shows a flowchart of the method provided by the embodiment of the present invention.

[0066] Specifically, a control method for an autonomous vehicle based on artificial intelligence, the method specifically includes the following steps:

[0067] Step S100, during the process of the autonomous vehicle carrying the current passenger for a tour, continuously collect local driving data corresponding to the road conditions of the path convergence change. The path convergence change road condition refers to a typical road condition where the autonomous vehicle drives from an area with ample space into an area with narrow space.

[0068] In the embodiments of the present invention, the number of current passengers is not strictly limited. In principle, it can be one or more. However, considering factors such as face recognition accuracy, collection stability of the experience index, and layout of in-vehicle space sensors, when the number of passengers is too large, it may interfere with data collection. Therefore, it is more applicable to scenarios with fewer passengers, especially relatively static and low-speed unmanned driving application scenarios such as urban tours, park shuttles, and sightseeing on limited routes.

[0069] To accurately model the impact of path convergence changes on passenger perception, the system continuously collects local driving data corresponding to the road conditions of the experienced path convergence changes during the operation of the unmanned vehicle. The local driving data refers to the driving data segment from a certain distance before the vehicle enters the path convergence change road condition to a certain distance after passing through this change road condition, which covers the entire process of driving behavior, environmental changes, and passenger reactions in the structural transition area.

[0070] The local driving data should at least include the following categories of information:

[0071] (1) Environmental perception data of the unmanned vehicle, including forward free passage width, lateral distribution of obstacles, path boundary change rate, etc.;

[0072] (2) Vehicle's own operating parameters, such as speed, acceleration, braking frequency, steering wheel angle change value, etc.;

[0073] (3) Passenger state data, including facial expression image sequence, physiological feature recognition data (such as posture micro-movement, blink frequency, etc.), and the experience index calibrated with time stamps;

[0074] (4) Information on the scenic spots or areas where it is located, and the preset travel value reference table for this travel area.

[0075] The reason for studying the road conditions of path convergence changes is that such structural mutations often cause switching of driving behavior patterns (such as switching from cruising to decelerating to avoid obstacles), contraction of visual depth of field, and drastic changes in passengers' subjective perception. In conventional path planning systems, this type of road condition is often only regarded as a geometric change. However, in practical applications, path convergence changes are likely to induce problems such as passengers' discomfort, sudden action reactions, and downgrading of subjective evaluations. The present invention attempts to construct a scoring correction mechanism centered on passenger perception, and through the trend of "the higher the path convergence degree, the greater the decline in passenger experience" repeatedly appearing in historical local data, to achieve dynamic adjustment of the current path score value, so that the path selection strategy not only considers safety and efficiency, but also takes into account the psychological comfort and tour satisfaction of users.

[0076] Furthermore, the control method of the artificial intelligence-based unmanned vehicle further includes the following steps:

[0077] Step S200: After the collected local driving data reaches a preset amount, when the unmanned vehicle is about to enter a new path convergence change road condition, an initial path selection score value corresponding to the path convergence change road condition is obtained and set as the score value to be revised.

[0078] Step S300 : extracting a plurality of target data with successively increasing path convergence strengths based on the collected local driving data.

[0079] Specifically, Figure 2 A flow chart for obtaining several sets of target data with successively increasing path convergence strength is shown.

[0080] The process of extracting a plurality of target data with increasing path convergence strength based on the collected local driving data specifically includes the following steps:

[0081] Step S301, analyzing each local driving data in sequence, and determining a first passage space envelope parameter of the unmanned vehicle before entering a road condition with a convergent path change, and a second passage space envelope parameter after entering the road condition with a convergent path change;

[0082] Step S302: performing a difference calculation on the first passage space envelope parameter and the second passage space envelope parameter to obtain a path convergence strength corresponding to the local driving data;

[0083] Step S303 : After completing the calculation of the path convergence strength corresponding to each local driving data, a plurality of target data are screened in a manner of increasing the path convergence strength.

[0084] In an embodiment of the present invention, the preset number of local driving data is typically set based on experience as the minimum number of samples required to ensure statistical trend significance, preferably no less than five. The reaching of this preset number serves as a triggering point, and its purpose is to ensure that, before officially executing the score correction, there is sufficient data to support the determination of whether the trend characteristic of "the higher the path convergence strength, the more significant the decline in user experience" exists, thereby preventing analytical bias caused by too few samples. In other words, only when the accumulated local driving data reaches a statistically reliable number will the system initiate the correction process before the new path converges and changes in road conditions occur, thereby improving the stability and credibility of the score correction mechanism.

[0085] The initial route selection score refers to the priority score assigned to the current route's convergence-changing road conditions after a preliminary evaluation of multiple candidate road conditions in the vehicle routing system. This score is typically used to prioritize subsequent route instruction generation modules. This score may be derived based on a comprehensive assessment of factors such as route length, traffic density, and scenic index. However, this score has limitations because it does not account for passenger discomfort caused by sudden changes in route structure (such as entering a narrow area from a wide area). The present invention aims to modify this score through experience-driven adjustments to better align it with current passengers' actual preferences, thereby optimizing route selection strategies.

[0086] The passage space envelope parameter represents the size of the spatial boundary measured within a certain distance ahead of an autonomous vehicle, ensuring safe passage. It reflects the vehicle's lateral spatial envelope capability before and after changes in the path structure. This envelope parameter is derived by fusing multiple data sources, including lidar point cloud modeling, visual stereo ranging, and vehicle-to-everything (V2X) perception information. It is a common extended safety assessment metric in existing intelligent perception systems.

[0087] Specifically, in step S301, the first and second envelope parameters of the passageway space before and after the path convergence change are obtained. In step S302, the difference between the two is calculated to determine the path convergence strength corresponding to the local driving data. This is used to quantify the structural compression trend of "transitioning from a wide space to a narrow space," thereby establishing a logical basis for the impact of structural change on the user experience. A larger difference indicates a more dramatic spatial mutation, a higher convergence strength, and a more pronounced potential sense of oppression.

[0088] In addition to the difference calculation method, this embodiment also uses methods such as proportional reduction rate, envelope curve gradient change rate, or convergence angle indicators constructed based on the forward depth map to characterize the path convergence strength. For example, by calculating the proportional relationship between the envelope of the first and second passage spaces, the relative severity of the convergence can be more intuitively reflected; the change in the slope of the passage space boundary fitting curve can be used to characterize the trend gradient of spatial contraction; and the convergence angle indicator generated based on the depth map can be used to determine the degree of spatial convergence by analyzing the change in the angle between the obstacle boundaries on both sides of the vehicle's field of view. These methods complement the difference calculation and are applicable to convergence strength modeling under different sensor conditions and path geometry types.

[0089] In step S303, the system sorts the path convergence strengths corresponding to each local driving data set, prioritizing a number of target data sets with increasing convergence strength, wide coverage, and balanced distribution. These serve as the primary sample set for subsequent analysis of the user experience degradation trend, ensuring the representativeness and gradient continuity of the extracted trend relationship. This screening method not only improves the accuracy of trend fitting but also helps identify the nonlinear response range between user experience feedback and path changes.

[0090] Furthermore, the control method for the AI-based driverless vehicle further includes the following steps:

[0091] Step S400: Analyze the experience index of the current passenger before and after the path convergence change in each target data respectively, and calculate the corresponding decline in the experience index.

[0092] Specifically, Figure 3 Fig. shows the flowchart for calculating the decline in the experience index.

[0093] Among them, analyzing the experience index of the current passenger before and after the path convergence change in each target data respectively, and calculating the corresponding decline in the experience index specifically includes the following steps:

[0094] Step S401: Obtain the passenger face video images of the current passenger before and after entering the path convergence change road conditions from each target data respectively;

[0095] Step S402: Based on the passenger face video images, use the image recognition algorithm to determine the first experience index before entering the path convergence change road conditions and the second experience index after entering respectively;

[0096] Step S403: Calculate the decline in the experience index according to the first experience index and the second experience index.

[0097] In the embodiment of the present invention, the passenger face video images can be obtained by real-time acquisition through a passenger forward camera device or a side camera device arranged inside the driverless vehicle. This camera device is preferably installed in front of the seat or at the center of the carriage, and can continuously capture the facial expression changes of the passenger without disturbing the normal activities of the passenger, and ensure the accuracy of image recognition in a high frame rate and high definition manner.

[0098] For the extraction of the first experience index and the second experience index, it is preferably to adopt the AI-based human face emotion recognition technology. Specifically, a convolutional neural network (CNN) model can be constructed to analyze the key expression features (such as eyebrow lifting, mouth corner drooping, frowning, pupil dilation, etc.) in the passenger face image, and combined with a preset multi-dimensional emotion weight model to quantitatively score the current emotion state, so as to obtain the first experience index before entering the path convergence change road conditions and the second experience index after entering respectively. The perception accuracy can also be further improved by fusing facial muscle tension analysis and physiological signal feedback (such as blinking frequency, forehead sweating trend, etc.).

[0099] If there are several current passengers, the system will separately collect the facial video images of each passenger before and after the path convergence change, and use image recognition algorithms to separately extract the first experience index and the second experience index corresponding to each passenger. Subsequently, the system calculates the decline in the experience index of each passenger respectively, and takes the average of the decline results of all current passengers to obtain the comprehensive decline in the experience index in this path change scenario. This method can effectively reflect the overall perception trend in the multi-passenger scenario, avoid the distortion of the scoring result caused by the reaction deviation of a single passenger, and further improve the objectivity and application breadth of the scoring correction.

[0100] As a relative change in the experience index between two time periods, the decline in the experience index can reflect the degree of change in the psychological comfort of passengers during the path convergence change process, specifically manifested as negative emotional reactions such as increased tension, increased oppression, and spatial discomfort. The more significant the change trend of this index, the greater the impact of the path space change on the passenger experience, indicating a clear correlation with behavioral responses.

[0101] Furthermore, the control method of the driverless vehicle based on artificial intelligence further includes the following steps:

[0102] Step S500, analyze whether the decline in the experience index shows an increasing trend as the path convergence intensity increases. If so, select several reference data whose path convergence intensity is within the same preset range as the current path convergence change road condition.

[0103] In the embodiment of the present invention, judging whether the decline in the experience index shows an increasing trend as the path convergence intensity increases aims to identify whether the current passengers have a clear experience sensitivity to the path space convergence change. The existence of this trend means that in the individual perception system of the current passengers, the higher the change intensity of the path space from wide to narrow, the more obvious the decline in their subjective experience, which further indicates that they have a consistent reaction characteristic to the path convergence change. The identification of this characteristic provides a judgment premise for the subsequent correction of the path selection score, ensuring that corresponding score adjustments are only made when the experience feedback has a strong correlation, thus avoiding over-correction of the score when the subjective response is unclear or the individual differences are large.

[0104] It should be noted that the judgment of this increasing trend does not require absolute monotonicity, but allows a certain degree of error or fluctuation. Specifically, the overall trend can be judged by setting a change threshold range or using statistical analysis methods such as linear fitting and curve fitting to determine whether the overall trend shows a positive slope, and then confirm whether the trend holds. This not only ensures the robustness of the trend judgment but also reduces the impact of individual abnormal data points on the overall analysis result.

[0105] After the trend is established, the system needs to further screen out a number of reference data whose path convergence strengths for the road conditions converging and changing with the current path are within the same preset range. This process can be achieved by calculating the difference between the current path convergence strength and the path convergence strengths corresponding to each local driving data, and setting a comparison tolerance range as the preset range for matching. The setting of the preset range can be based on the statistical results of historical samples, selecting the path convergence strength interval where the decline change of the experience index is the most intensive, or using a fixed threshold (such as ±10%, ±0.5m, etc.) according to the system tuning experience, so as to ensure that the selected reference data is representative and comparable in terms of convergence structure characteristics, and ensure the effectiveness and stability of the correction result.

[0106] Further, the control method for the driverless vehicle based on artificial intelligence further includes the following steps:

[0107] Step S600, based on a number of reference data, determine the comprehensive decline of the experience index, and obtain the corresponding comprehensive decline of the scenic spot tour value, and jointly adjust the score value to be corrected according to the comprehensive decline of the experience index and the comprehensive decline of the scenic spot tour value.

[0108] Specifically, Figure 4 The flowchart showing the weighted correction of the score value to be corrected is shown.

[0109] Among them, based on a number of reference data, determining the comprehensive decline of the experience index, and obtaining the corresponding comprehensive decline of the scenic spot tour value, and jointly adjusting the score value to be corrected according to the comprehensive decline of the experience index and the comprehensive decline of the scenic spot tour value specifically includes the following steps:

[0110] Step S601, calculate the average value of the decline of the experience index corresponding to all reference data to obtain the comprehensive decline of the experience index;

[0111] Step S602, obtain the preset tour value reference table of the scenic spot where the driverless vehicle is located, respectively determine the first scenic spot tour value and the second scenic spot tour value before and after the driverless vehicle enters the road condition with path convergence change for each reference data, and calculate the corresponding decline of the scenic spot tour value;

[0112] Step S603, take the average of the decline of the scenic spot tour value of all reference data to obtain the comprehensive decline of the scenic spot tour value;

[0113] Step S604, based on the comprehensive decline of the experience index and the comprehensive decline of the scenic spot tour value, perform weighted correction on the score value to be corrected according to the preset correction function to obtain the corrected path selection score value;

[0114] Step S605, apply the corrected path selection score value to the decision priority evaluation and path instruction generation for the current road condition with path convergence change.

[0115] The preset correction function is as follows:

[0116] ;

[0117] Wherein, refers to the corrected path selection score value, refers to the score value to be corrected, refers to the total number of reference data, and respectively refer to the second experience index and the first experience index after and before the unmanned vehicle enters the road condition with convergent path change in the th reference data, refers to the decrease in the experience index corresponding to the th reference data, refers to the decrease in the comprehensive experience index, and respectively refer to the second scenic spot tour value and the first scenic spot tour value after and before the unmanned vehicle enters the road condition with convergent path change in the th reference data, refers to the decrease in the scenic spot tour value corresponding to the th reference data, refers to the decrease in the comprehensive scenic spot tour value, and are respectively the adjustment weights of the decrease in the comprehensive experience index and the decrease in the comprehensive scenic spot tour value, and and are both greater than 0.

[0118] In the embodiment of the present invention, the specific implementation process of step S601 is as follows: Based on a number of reference data screened in the previous step, the system sequentially extracts the experience index of the current passenger before and after the path convergence change in each group of data, obtains the decrease in the experience index using the preset decrease calculation formula, and performs arithmetic averaging on the decrease results of all reference data to obtain the decrease in the comprehensive experience index of the passenger within the current path convergence intensity range. This average value reflects the general downward trend of the passenger's experience under similar road condition intensities.

[0119] In step S602, the preset tour value reference table is constructed based on the existing scenic spot evaluation system in the scenic spot management system or the map content platform, and it is a mature existing technology. Its sources can include tourist ratings, popularity rankings, subjective scoring systems, etc., and it is formed by combining manual settings and data modeling. Through this reference table, the system determines the scenic spot evaluation values of the unmanned vehicle before and after entering the road condition with convergent path change in each group of reference data, that is, the first scenic spot tour value and the second scenic spot tour value, calculates their relative change amplitudes, and uses them as the decrease index of the scenic spot tour value.

[0120] In step S603, the system averages the decreases in the scenic spot tour values corresponding to the above reference data to obtain the comprehensive decrease in the scenic spot tour value within the range of the path convergence strength. This indicator is used to evaluate whether this type of road condition is generally associated with a decrease in the scenic spot value, and further assist in judging whether the reason for the experience decline is dominated by the path factor.

[0121] The reason for introducing two factors to weighted-correct the score value to be corrected in step S604 is that the decrease in the comprehensive experience index reflects the subjective discomfort of passengers, while the decrease in the comprehensive scenic spot tour value reflects the objective environmental quality factors that may affect the experience. If the passenger experience declines, but at the same time the corresponding scenic spot evaluation also declines, it indicates that the poor experience may be affected by the scenic spot itself, and the system will moderately callback the score; if the scenic spot value does not decline, or even better, but the experience is worse, it indicates that the path structure has indeed produced a strong negative reaction to this passenger, and the system will increase the score penalty. Such a correction logic realizes the dynamic fusion adjustment of subjective and objective information, which helps to generate a more real and adaptable path scoring result.

[0122] In step S605, the corrected path selection score value will be directly used in processes such as priority judgment, alternative path comparison, and path instruction generation in the path planning module. For example, in multiple path branch selection nodes, the system will preferentially select the path with a higher corrected score value to improve the continuity of the passenger experience; if the current path score is significantly lowered due to experience problems, the system can trigger the path switching logic to avoid the convergent section with poor experience.

[0123] The penalty and compensation effects are reflected in the score correction function, mainly to make the path selection score more in line with the passenger experience and the actual scenic spot evaluation. For example, the decline in the experience index reflects the degree of deterioration of the passenger experience after entering a narrow space. When this index is negative, it indicates that the passenger experience has significantly decreased. Therefore, a penalty needs to be imposed on the original score to reduce the priority of this path; this is the role of the "1 + the decline in the comprehensive experience index" part, because the more the experience decreases, the greater the absolute value of this product, and ultimately the overall score decreases. On the other hand, the decline in the scenic spot visit value reflects the reduction in the attractiveness of the scenic spot caused by the path convergence change; when the scenic spot evaluation increases, it means that the overall value of the scenic spots passed after the current path convergence change has increased. If the passenger experience still shows a significant decline at this time, it often indicates that the passenger has a subjectively overly sensitive reaction to the path structure change. Even if the actual scenic spot quality has improved, negative experiences are still generated due to factors such as psychological expectations or visual narrowing. Such abnormal reactions should not directly affect the path score. Therefore, the score correction mechanism should further weaken the original score to prevent the system from mistakenly regarding individual abnormal feedback as a general trend; when the scenic spot evaluation decreases, it means that the path change has indeed led to a weakening of the scenic spot quality. At this time, the system introduces a compensation term of "1 - the decline in the comprehensive scenic spot visit value" to moderately adjust the score correction amplitude to avoid the score being overly reduced due to the objective deterioration of the environment, so as to achieve a dynamic balance between the change in the experience and the change in the scenic spot value. In short, this design enables the correction function to not only impose necessary penalties on the experience decline but also appropriately adjust the score according to the actual attractiveness of the scenic spot. Under the combined action of the two, the final score more comprehensively reflects the dual impacts of the path on the passenger subjective experience and the objective scenic spot evaluation.

[0124] It should be noted that although the preset correction function is in a form with a simple structure and intuitive logic, it is not the only optional method. In other embodiments, alternative algorithms such as piecewise non-linear weight functions, exponential weight models, and score mapping relationships based on clustering or regression models can also be introduced to achieve a more complex corresponding structure between the score value and the dual variables of the experience - scenic spot value, so as to meet the needs of different types of passengers, path environments, and scoring systems.

[0125] Furthermore, Figure 5 The application architecture diagram of the system provided by the embodiment of the present invention is shown.

[0126] Among them, in another preferred embodiment provided by the present invention, a control system for an artificial intelligence-based driverless vehicle includes:

[0127] The data acquisition module 100 is used to continuously collect local driving data corresponding to the road conditions with path convergence changes during the process of the driverless vehicle carrying the current passenger for a tour; the road conditions with path convergence changes refer to the typical road conditions where the driverless vehicle drives from an area with ample space into an area with narrow space.

[0128] Furthermore, the control system of the driverless vehicle based on artificial intelligence further includes:

[0129] The to-be-corrected score determination module 200 is used to, after the collected local driving data reaches a preset quantity, when the driverless vehicle is about to enter a new road condition with path convergence changes, obtain the initial path selection score value corresponding to the road condition with path convergence changes and set it as the to-be-corrected score value.

[0130] The target data extraction module 300 is used to extract several pieces of target data with gradually increasing path convergence intensity based on the collected local driving data.

[0131] Specifically, Figure 6 The block diagram of the target data extraction module 300 in the system provided by the embodiment of the present invention is shown.

[0132] Among them, in the preferred implementation manner provided by the present invention, the target data extraction module 300 specifically includes:

[0133] The degree-of-freedom parameter determination unit 301 is used to parse each local driving data in sequence and determine the first traffic space envelope degree parameter of the driverless vehicle before entering the road condition with path convergence changes and the second traffic space envelope degree parameter after entering the road condition with path convergence changes from it;

[0134] The path convergence intensity calculation unit 302 is used to calculate the difference between the first traffic space envelope degree parameter and the second traffic space envelope degree parameter to obtain the path convergence intensity corresponding to the local driving data;

[0135] The target data screening unit 303 is used to, after calculating the path convergence intensity corresponding to each local driving data, screen several pieces of target data in the order of gradually increasing path convergence intensity.

[0136] Furthermore, the control system of the driverless vehicle based on artificial intelligence further includes:

[0137] The experience reduction calculation module 400 is used to parse the experience index of the current passenger before and after the path convergence change in each target data respectively and calculate the corresponding experience index reduction.

[0138] Specifically, Figure 7 The block diagram of the experience reduction calculation module 400 in the system provided by the embodiment of the present invention is shown.

[0139] Among them, in the preferred embodiment provided by the present invention, the experience reduction calculation module 400 specifically includes:

[0140] A facial information acquisition unit 401, configured to respectively acquire the passenger facial video images of the current passenger before and after entering the road condition of the path convergence change in each target data;

[0141] An experience index determination unit 402, configured to respectively determine a first experience index before entering the road condition of the path convergence change and a second experience index after entering based on the passenger facial video images by using an image recognition algorithm;

[0142] An experience reduction calculation unit 403, configured to calculate the experience index reduction according to the first experience index and the second experience index.

[0143] Furthermore, the control system of the driverless vehicle based on artificial intelligence further includes:

[0144] A reference data screening module 500, configured to analyze whether the experience index reduction shows an increasing trend as the path convergence intensity increases. If so, filter out a number of reference data whose path convergence intensity is within the same preset range as the path convergence change road condition of the current path.

[0145] Furthermore, the control system of the driverless vehicle based on artificial intelligence further includes:

[0146] A to-be-corrected score adjustment module 600, configured to determine a comprehensive experience index reduction based on a number of reference data, obtain a corresponding comprehensive scenic spot tour value reduction, and jointly adjust the to-be-corrected score value according to the comprehensive experience index reduction and the comprehensive scenic spot tour value reduction.

[0147] Specifically, Figure 8 The structural block diagram of the to-be-corrected score adjustment module 600 in the system provided by the embodiment of the present invention is shown.

[0148] Among them, in the preferred embodiment provided by the present invention, the to-be-corrected score adjustment module 600 specifically includes:

[0149] A first comprehensive reduction calculation unit 601, configured to calculate the average value of the experience index reductions corresponding to all reference data to obtain a comprehensive experience index reduction;

[0150] A scenic spot tour value acquisition unit 602, configured to acquire a preset tour value reference table of the scenic spot where the driverless vehicle is located, respectively determine a first scenic spot tour value and a second scenic spot tour value of the driverless vehicle before and after entering the road condition of the path convergence change in each reference data, and calculate the corresponding scenic spot tour value reduction;

[0151] The second comprehensive decline calculation unit 603 is configured to average the declines in scenic spot visit values of all reference data to obtain the comprehensive decline in scenic spot visit values;

[0152] The to-be-corrected score adjustment unit 604 is configured to perform weighted correction on the to-be-corrected score value according to a preset correction function based on the comprehensive experience index decline and the comprehensive scenic spot visit value decline, so as to obtain the corrected path selection score value;

[0153] The corrected score value application unit 605 is configured to apply the corrected path selection score value to the decision priority evaluation and path instruction generation for the current path convergence and changing road conditions;

[0154] The preset correction function is:

[0155] ;

[0156] wherein, refers to the corrected path selection score value, refers to the to-be-corrected score value, refers to the total number of reference data, and respectively refer to the second experience index and the first experience index after and before the driverless vehicle enters the path convergence and changing road conditions in the th reference data, refers to the experience index decline corresponding to the th reference data, refers to the comprehensive experience index decline, and respectively refer to the second scenic spot visit value and the first scenic spot visit value after and before the driverless vehicle enters the path convergence and changing road conditions in the th reference data, refers to the scenic spot visit value decline corresponding to the th reference data, refers to the comprehensive scenic spot visit value decline, and are respectively the adjustment weights of the comprehensive experience index decline and the comprehensive scenic spot visit value decline, and and are both greater than 0.

[0157] It should be understood that although the steps in the flowcharts of the embodiments of the present invention are shown in sequence according to the indications of the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise clearly stated in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, at least a part of the steps in each embodiment may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same moment, but can be executed at different moments. The execution order of these sub-steps or stages is not necessarily sequential either, but can be executed alternately or in rotation with at least a part of other steps or sub-steps or stages of other steps.

[0158] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided in this application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0159] The technical features of the above-described embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0160] The above-described embodiments merely represent several implementation manners of the present invention. The description thereof is relatively specific and detailed, but it should not be construed as a limitation to the scope of the patent of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several deformations and improvements can still be made, and these all fall within the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the appended claims.

[0161] The above is only the preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A control method for an autonomous vehicle based on artificial intelligence, characterized in that, The method includes: During the process of the driverless vehicle carrying the current passenger for a tour, continuously collect local driving data corresponding to the road conditions with convergent path changes along the traveled path; After the collected local driving data reaches a preset quantity, when the driverless vehicle is about to enter a new road condition with convergent path changes, obtain the initial path selection score value corresponding to this road condition with convergent path changes, and set it as the score value to be corrected; Based on the collected local driving data, extract several pieces of target data with gradually increasing path convergence intensities from it; Analyze the experience index of the current passenger before and after the path convergence change in each target data respectively, and calculate the corresponding reduction in the experience index; Analyze whether the reduction in the experience index shows an increasing trend as the path convergence intensity increases. If so, screen out several reference data whose path convergence intensities are within the same preset range as the path convergence intensity of the current road condition with convergent path changes; Based on several reference data, determine the comprehensive reduction in the experience index, and obtain the corresponding comprehensive reduction in the scenic spot tour value. Adjust the score value to be corrected jointly according to the comprehensive reduction in the experience index and the comprehensive reduction in the scenic spot tour value.

2. The control method of the driverless vehicle based on artificial intelligence according to claim 1, characterized in that The road condition with convergent path changes refers to the typical road condition where the driverless vehicle drives from an area with ample space into an area with narrow space.

3. The control method of an autonomous vehicle based on artificial intelligence according to claim 2, wherein The step of extracting several pieces of target data with gradually increasing path convergence intensities from the collected local driving data includes: Analyze each local driving data in sequence, and determine the first passing space envelope parameter of the driverless vehicle before entering the road condition with convergent path changes and the second passing space envelope parameter after entering the road condition with convergent path changes from it; Perform a difference calculation on the first passing space envelope parameter and the second passing space envelope parameter to obtain the path convergence intensity corresponding to this local driving data; After completing the calculation of the path convergence intensity corresponding to each local driving data, screen out several pieces of target data in the order of gradually increasing path convergence intensity.

4. The control method of an autonomous vehicle based on artificial intelligence according to claim 3, characterized in that The step of analyzing the experience index of the current passenger before and after the path convergence change in each target data respectively and calculating the corresponding reduction in the experience index includes: Obtain the passenger face video images of the current passenger before and after entering the road condition with convergent path changes in each target data respectively; Based on the passenger face video images, use an image recognition algorithm to determine the first experience index before entering the road condition with convergent path changes and the second experience index after entering respectively; Calculate the reduction in the experience index according to the first experience index and the second experience index.

5. The control method of an artificial intelligence-based driverless vehicle according to claim 4, characterized in that, The step of determining the comprehensive reduction in the experience index based on several reference data, obtaining the corresponding comprehensive reduction in the scenic spot tour value, and jointly adjusting the score value to be corrected according to the comprehensive reduction in the experience index and the comprehensive reduction in the scenic spot tour value includes: Calculate the average value of the corresponding reduction in the experience index in all reference data to obtain the comprehensive reduction in the experience index; Obtain the preset tour value reference table of the scenic spot where the driverless vehicle is located, determine the first scenic spot tour value and the second scenic spot tour value of the driverless vehicle before and after entering the road condition with convergent path changes in each reference data respectively, and calculate the corresponding reduction in the scenic spot tour value; Take the average of the decreases in the scenic spot travel values for all reference data to obtain the comprehensive decrease in the scenic spot travel value; Based on the comprehensive decrease in the experience index and the comprehensive decrease in the scenic spot travel value, perform weighted correction on the score value to be corrected according to a preset correction function to obtain the corrected path selection score value; Apply the corrected path selection score value to the decision-making priority evaluation and path instruction generation for the current path convergence and changing road conditions.

6. The control method of the driverless vehicle based on artificial intelligence according to claim 5, wherein, The preset correction function is: ； Among them, refers to the corrected path selection score value, refers to the score value to be corrected, refers to the total number of reference data, and respectively refer to the second experience index and the first experience index after and before the unmanned vehicle enters the road condition with convergent path change in the th reference data, refers to the decrease in the experience index corresponding to the th reference data, refers to the decrease in the comprehensive experience index, and respectively refer to the second scenic spot tour value and the first scenic spot tour value after and before the unmanned vehicle enters the road condition with convergent path change in the th reference data, refers to the decrease in the scenic spot tour value corresponding to the th reference data, refers to the decrease in the comprehensive scenic spot tour value, and are respectively the adjustment weights of the decrease in the comprehensive experience index and the decrease in the comprehensive scenic spot tour value, and and are both greater than 0.

7. A control system for an autonomous vehicle based on artificial intelligence, characterized in that, The system includes: a data collection module, a score to be corrected determination module, a target data extraction module, an experience decrease calculation module, a reference data screening module, and a score to be corrected adjustment module, where: The data collection module is used to continuously collect local driving data corresponding to the path convergence and changing road conditions during the process of the driverless vehicle carrying the current passenger for sightseeing; the path convergence and changing road conditions refer to the typical road conditions where the driverless vehicle drives from an area with sufficient space into an area with narrow space; The score to be corrected determination module is used to, after the collected local driving data reaches a preset quantity, when the driverless vehicle is about to enter a new path convergence and changing road condition, obtain the initial path selection score value corresponding to the path convergence and changing road condition and set it as the score to be corrected; The target data extraction module is used to extract several pieces of target data with gradually increasing path convergence intensities from the collected local driving data; The experience decrease calculation module is used to respectively analyze the experience index of the current passenger before and after the path convergence change in each target data and calculate the corresponding experience index decrease; The reference data screening module is used to analyze whether the experience index decrease shows an increasing trend as the path convergence intensity increases. If so, screen out several pieces of reference data whose path convergence intensities are within the same preset range as the path convergence intensity of the current path convergence and changing road condition; The score to be corrected adjustment module is used to determine the comprehensive experience index decrease based on several pieces of reference data, obtain the corresponding comprehensive decrease in the scenic spot travel value, and jointly adjust the score to be corrected according to the comprehensive experience index decrease and the comprehensive decrease in the scenic spot travel value.

8. The control system of the driverless vehicle based on artificial intelligence according to claim 7, characterized in that, The target data extraction module specifically includes: The degree of freedom parameter determination unit is used to sequentially analyze each piece of local driving data and determine the first traffic space envelope parameter of the driverless vehicle before entering the path convergence and changing road condition and the second traffic space envelope parameter after entering the path convergence and changing road condition from it; The path convergence intensity calculation unit is used to calculate the difference between the first traffic space envelope parameter and the second traffic space envelope parameter to obtain the path convergence intensity corresponding to the local driving data; The target data screening unit is used to, after calculating the path convergence intensities corresponding to each piece of local driving data, screen several pieces of target data in the order of gradually increasing path convergence intensity.

9. The control system of the driverless vehicle based on artificial intelligence according to claim 8, wherein The experience decrease calculation module specifically includes: The facial information acquisition unit is used to respectively acquire the passenger facial video images of the current passenger before and after entering the path convergence and changing road condition in each target data; An experience index determination unit, configured to respectively determine a first experience index before entering a road condition with a convergent change in the entry path and a second experience index after entering, based on the passenger face video image, by using an image recognition algorithm; An experience index decrease calculation unit, configured to calculate a decrease in the experience index according to the first experience index and the second experience index; 10. The control system of the driverless vehicle based on artificial intelligence according to claim 9, characterized in that, The to-be-corrected score adjustment module specifically includes: A first comprehensive decrease calculation unit, configured to calculate an average value of the decreases in the experience index corresponding to all reference data, to obtain a comprehensive decrease in the experience index; A scenic spot tour value acquisition unit, configured to acquire a preset tour value reference table of the scenic spot where the driverless vehicle is located, respectively determine a first scenic spot tour value and a second scenic spot tour value before and after the driverless vehicle enters a road condition with a convergent change in the entry path for each reference data, and calculate the corresponding decrease in the scenic spot tour value; A second comprehensive decrease calculation unit, configured to take an average of the decreases in the scenic spot tour value of all reference data, to obtain a comprehensive decrease in the scenic spot tour value; A to-be-corrected score adjustment unit, configured to perform weighted correction on the to-be-corrected score value according to a preset correction function based on the comprehensive decrease in the experience index and the comprehensive decrease in the scenic spot tour value, to obtain a corrected path selection score value; A corrected score value application unit, configured to apply the corrected path selection score value to the decision-making priority evaluation and path instruction generation for the current road condition with a convergent change in the entry path; The preset correction function is: ； Among them, refers to the corrected path selection score value, refers to the score value to be corrected, refers to the total number of reference data, and respectively refer to the second experience index and the first experience index after and before the unmanned vehicle enters the road condition with convergent path change in the th reference data, refers to the decrease in the experience index corresponding to the th reference data, refers to the decrease in the comprehensive experience index, and respectively refer to the second scenic spot tour value and the first scenic spot tour value after and before the unmanned vehicle enters the road condition with convergent path change in the th reference data, refers to the decrease in the scenic spot tour value corresponding to the th reference data, refers to the decrease in the comprehensive scenic spot tour value, and are respectively the adjustment weights of the decrease in the comprehensive experience index and the decrease in the comprehensive scenic spot tour value, and and are both greater than 0.

Images