Intelligent evaluation method and system for path planning of autonomous vehicle

Abstract

The invention provides an intelligent evaluation method and system for path planning of an automatic driving vehicle. The method comprises the following steps: collecting original data of path planning of the automatic driving vehicle; constructing a plurality of associated evaluation indexes of the path planning intelligence of the autonomous vehicle; for the plurality of evaluation indexes, obtaining a subjective weight set of experts; aiming at the plurality of evaluation indexes, obtaining an objective weight set; determining an optimal combination coefficient of the subjective weight set and the objective weight set according to the consistency of the subjective weight set and the objective weight set; and according to the determined subjective weight set, the objective weight set, the optimal combination coefficient of the subjective weight set and the optimal combination coefficient of the objective weight set, evaluating the intelligence of the path planning of the autonomous vehicle. By collecting the subjective weight set and the objective weight set and adjusting the optimal combination coefficient of the objective weight set and the subjective weight set according to the consistency of the subjective weight set and the objective weight set, the accuracy of intelligent evaluation of path planning is improved.

Description

Technical Field

[0001] This disclosure relates to the technical field of intelligent evaluation of path planning, and more particularly to an intelligent evaluation method and system for path planning of autonomous vehicles. Background Technology

[0002] Path planning is one of the core capabilities of autonomous vehicles. With the rapid development of sensor devices and intelligent technologies, the intelligence performance of path planning varies greatly. Therefore, how to identify and compare the intelligence performance of path planning has always been a real problem that has troubled researchers.

[0003] Intelligence assessment is a feasible method for measuring the intelligence level of intelligent systems. By comprehensively evaluating the intelligence of an intelligent system, an objective and scientific assessment result is obtained, providing an important basis for decision-making regarding the application and promotion of intelligent systems.

[0004] The intelligence of autonomous vehicle path planning is reflected in multiple dimensions. In order to comprehensively and accurately assess its intelligence level, it is necessary to build a multi-indicator comprehensive evaluation system that can adapt to the evaluation needs under complex and dynamic traffic scenarios.

[0005] In the evaluation of the intelligence of autonomous vehicle path planning, autonomous vehicles possess multiple performance indicators (C). j (j = 1, 2, ..., N), however, different dimensions lead to different selected indicators, and there may also be certain correlations between the indicators. Based on the characteristics of the relationships between the indicators, an appropriate weighting method should be selected. If the evaluation problem is relatively simple, a single subjective or objective weighting method can be chosen; if the evaluation problem is relatively complex, a combined subjective and objective weighting method can be chosen.

[0006] Given the complexity of evaluating the intelligence of autonomous vehicle path planning and the strong correlation between indicators, a combined subjective and objective weighting method is adopted. Typically, the comprehensive score for intelligence is obtained by weighting and summing multiple indicators, and this score depends on the rationality of the weight allocation for each indicator.

[0007] Existing weighting methods do not consider the correlation between indicators when determining indicator weights. When assigning weights in a combination of subjective and objective methods, they rely on human experience to set the weights. The resulting combined weights are difficult to adapt to intelligent systems with strong correlations and multiple couplings, resulting in insufficient robustness and credibility of the evaluation results. Summary of the Invention

[0008] This disclosure provides a method and system for evaluating the intelligence of autonomous vehicle path planning, in order to solve one or more technical problems in the prior art.

[0009] In a first aspect, this disclosure provides a method for evaluating the intelligence of path planning in autonomous vehicles, the method comprising the following steps:

[0010] Collect raw data for autonomous vehicle path planning;

[0011] Construct multiple related evaluation metrics for the intelligence of autonomous vehicle path planning;

[0012] For multiple evaluation indicators, obtain the expert's subjective weight set;

[0013] For multiple evaluation indicators, obtain an objective set of weights;

[0014] Based on the consistency between the subjective weight set and the objective weight set, determine the optimal combination coefficients of the subjective weight set and the objective weight set;

[0015] The intelligence of autonomous vehicle path planning is evaluated based on the determined subjective weight set, objective weight set, optimal combination coefficient of subjective weight set, and optimal combination coefficient of objective weight set.

[0016] In the above technical solution, by collecting subjective weight sets and objective weight sets, and adjusting the optimal combination coefficients of the objective weight sets and subjective weight sets based on the consistency of the subjective weight sets and objective weight sets, the accuracy of intelligent assessment of path planning is improved.

[0017] In a specific feasible implementation, it also includes:

[0018] The process of obtaining the expert's subjective weight set for multiple evaluation indicators specifically includes:

[0019] For multiple evaluation indicators, obtain expert opinion set F. m F m ={f m1 ,f m2 ,...,f mn};

[0020] Based on the aforementioned opinion set, any evaluation index C is determined using ANP. j weight set w j w j ={f 1j ,f 2j ,...,f mj}; This weight set w j The average value is used as the subjective weight value of this indicator. Then the set of subjective weights for all evaluation indicators

[0021] In a specific feasible implementation, obtaining the objective and subjective weight set for multiple evaluation indicators specifically includes:

[0022] The information content of any evaluation indicator is determined by its volatility and conflict.

[0023] Calculate the objective weight ω′ of the evaluation indicator based on the determined amount of information. j ;

[0024] Obtain the objective weight set ω for all evaluation indicators o ={ω′1,ω′2,...,ω′ j}

[0025] In one specific implementation scheme, obtaining an objective set of evaluation weights for multiple evaluation indicators further includes:

[0026] Remove the influence of the dimensions of each indicator on the weights in the original data, and standardize the original data.

[0027] In one specific implementation scheme, determining the information content of an indicator by considering its volatility and conflict; specifically including:

[0028] Calculate the volatility U of the j-th evaluation indicator. j : in,

[0029] U j X j Let X' and X' represent the standard deviation and mean of the j-th evaluation indicator, respectively. ij This represents the i-th data value of the j-th evaluation indicator; n represents the sample size.

[0030] Calculate the conflict R of the j-th evaluation indicator. j : Where, r ij The coefficient of correlation between the j-th evaluation indicator and other evaluation indicators is represented by p; p is the number of evaluation indicators.

[0031] Calculate the information content C of the j-th evaluation indicator. j :C j =U j R j .

[0032] In one specific implementation scheme, the objective weight of the evaluation index is calculated based on the determined amount of information; specifically:

[0033] According to the formula Determine the objective weight of the j-th evaluation indicator.

[0034] In a specific feasible implementation, determining the optimal combination coefficients of the subjective weight set and the objective weight set based on their consistency specifically includes:

[0035] The combination coefficients of the subjective and objective weight sets are determined by finding the Nash equilibrium and constructing the Lagrange function.

[0036] The Kendall coefficient is used to assess the consistency of subjective weights; the Kendall coefficient is specifically:

[0037]

[0038] in, f represents the sum of squared deviations from the mean of all assigned values ​​for the indicators; j It is the sum of the values ​​assigned to the j-th indicator by all experts. It is the average of the sum of all the assigned values ​​for all indicators.

[0039] Consistency in calculating objective weights using the coefficient of variation;

[0040] The optimal combination coefficients for the subjective and objective weight sets are determined by assigning higher combination coefficients to weights with higher consistency.

[0041] In a specific feasible implementation, the combination coefficients of the subjective weight set and the objective weight set are determined by finding the Nash equilibrium and constructing the Lagrange function; specifically, this includes the following steps:

[0042] (1) Finding Nash Equilibrium

[0043] 1) Introduce constraints and establish optimization;

[0044] Let α1 and α2 be the subjective weight vectors ω. s and objective weight vector ω o The combination coefficients are then used to determine the combination weight vector ω:

[0045] ω=α1ω s +α2ω o (1) Based on minimizing the sum of squared deviations between the combined weights and the subjective and objective weights, an optimization can be established:

[0046]

[0047] 2) Construct the Lagrange function;

[0048] From equation (2), we can obtain:

[0049]

[0050] Substituting equation (1) into equation (3) and further simplifying it, we get:

[0051]

[0052] 3) Take the partial derivatives of equation (4) with respect to α1, α2 and λ respectively;

[0053]

[0054] 4) The combined coefficients of the subjective and objective weights are α′1 and α′2, respectively.

[0055] In one specific implementation, the consistency W2 of the objective weights is calculated using the coefficient of variation; specifically: W2 = 1 / (1+CV)

[0056] Among them, the coefficient of variation and These are the standard deviation and mean of multiple sets of objective weights, respectively.

[0057] In a specific feasible implementation, the step of determining the combination coefficients of the subjective weight set and the objective weight set by assigning higher combination coefficients to weights with higher consistency specifically includes:

[0058] The final combination coefficients are adjusted by assigning higher coefficients to weights with higher consistency.

[0059]

[0060] in,

[0061] Calculate the optimal combination weights using the optimal combination coefficients:

[0062]

[0063] Where W1 represents the consistency of subjective weights, and W2 represents the consistency of objective weights. The optimal combination coefficients for subjective weights. The optimal combination coefficients for objective weights.

[0064] Secondly, an intelligent evaluation system for autonomous vehicle path planning is provided, the system comprising:

[0065] The data acquisition module collects raw data for autonomous vehicle path planning;

[0066] The data processing module constructs multiple related evaluation metrics for the intelligence of autonomous vehicle path planning;

[0067] For multiple evaluation indicators, obtain the subjective weight set of experts; for multiple evaluation indicators, obtain the objective weight set; based on the consistency of the subjective weight set and the objective weight set, determine the optimal combination coefficient of the subjective weight set and the objective weight set; evaluate the intelligence of autonomous vehicle path planning based on the determined subjective weight set, objective weight set, optimal combination coefficient of the subjective weight set and the optimal combination coefficient of the objective weight set.

[0068] In the above technical solution, by collecting subjective weight sets and objective weight sets, and adjusting the optimal combination coefficients of the objective weight sets and subjective weight sets based on the consistency of the subjective weight sets and objective weight sets, the accuracy of intelligent assessment of path planning is improved.

[0069] Thirdly, this disclosure provides an electronic device, including:

[0070] At least one processor; and

[0071] A memory communicatively connected to the at least one processor; wherein,

[0072] The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform any of the methods in the embodiments of this disclosure.

[0073] Fourthly, this disclosure provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to perform any of the methods in the embodiments of this disclosure.

[0074] Fifthly, this disclosure provides a computer program product, including a computer program that, when executed by a processor, implements any of the methods in the embodiments of this disclosure.

[0075] The technical solution disclosed herein ensures the accuracy of path planning evaluation by collecting subjective weight sets and objective weight sets, and adjusting the coefficients of the objective weight sets and subjective weight sets based on the consistency between the subjective weight sets and objective weight sets.

[0076] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0077] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the various drawings denote the same or similar parts or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings depict only some embodiments provided according to this disclosure and should not be construed as limiting the scope of this disclosure.

[0078] Figure 1 A flowchart illustrating the method for evaluating the intelligence of autonomous vehicle path planning provided in this embodiment of the disclosure;

[0079] Figure 2 This is a schematic diagram of a test scenario provided in an embodiment of this disclosure;

[0080] Figure 3 A schematic diagram illustrating the evaluation metrics of the intelligent evaluation method for autonomous vehicle path planning provided in this embodiment of the disclosure;

[0081] Figure 4 A schematic diagram comparing the intelligent evaluation method for autonomous vehicle path planning provided in this disclosure with other methods;

[0082] Figure 5 A block diagram of an electronic device provided in an embodiment of this disclosure. Detailed Implementation

[0083] The present disclosure will now be described in further detail with reference to the accompanying drawings. The same reference numerals in the drawings denote elements that have the same or similar functions. Although various aspects of embodiments are shown in the drawings, they are not necessarily drawn to scale unless specifically indicated otherwise.

[0084] Furthermore, to better illustrate this disclosure, numerous specific details are set forth in the following detailed description. Those skilled in the art will understand that this disclosure can be practiced without certain specific details. In some instances, methods, means, components, and circuits well known to those skilled in the art have not been described in detail in order to highlight the main points of this disclosure.

[0085] To facilitate understanding of the intelligent assessment method for autonomous vehicle path planning provided in this application (hereinafter referred to as the assessment method), its application scenario is first described. The assessment method provided in this application is used to assess the intelligence of autonomous vehicle path planning. Current assessment methods heavily rely on human experience, resulting in low accuracy. For example, in existing assessment methods, the correlation characteristics between assessment indicators are ignored when assigning weights, leading to the selected weighting method failing to effectively reflect the interaction mechanism between indicators, thus causing the weight allocation to be inconsistent with reality. Furthermore, in existing assessment methods, when combining weights, it is necessary to determine the combination coefficients of subjective and objective weights. Existing research relies on human experience and does not consider the internal consistency of subjective and objective weights, making it difficult to guarantee the scientific validity of the weighting results. Therefore, this application provides an assessment method, which will be described in detail below with reference to specific figures and implementation.

[0086] refer to Figure 1 As shown, Figure 1 A flowchart of the evaluation method provided in an embodiment of this application is shown.

[0087] The evaluation method provided in this application includes the following steps:

[0088] Step 001: Collect raw data for autonomous vehicle path planning;

[0089] Specifically, raw data for autonomous vehicle path planning can be collected through a data acquisition module. This can be achieved by uploading the vehicle's raw data to the data acquisition module via a data cable or wirelessly.

[0090] Step 002: Construct multiple related intelligent evaluation indicators for autonomous vehicle path planning;

[0091] Specifically, when constructing evaluation indicators, based on the needs and characteristics of intelligent path planning assessment for autonomous vehicles, the selection of evaluation indicators should adhere to the principles of "observability, quantifiability, and strong correlation." The evaluation indicator is designated as C. j (j = 1, 2, ..., N) where N is a positive integer.

[0092] For example, the evaluation metrics constructed in this application embodiment can be safety, efficiency, and comfort. During data collection, safety can be assessed using the obstacle collision rate (rate). oc Mean lateral acceleration a h Quantification and effectiveness can be achieved through metrics such as task completion time (t) and average speed. The total trajectory length (l), energy consumption (ec), etc., can be quantified; comfort can be assessed through average jerk. Mean curvature Therefore, after constructing the evaluation indicators, data from 3 categories and 8 indicators were collected from the raw data for subsequent data processing.

[0093] Of the three indicators mentioned above, safety is one of the core indicators in evaluating the intelligence of autonomous driving systems. It measures the autonomous driving system's ability to ensure its own driving safety and protect passengers, pedestrians, and other road users from harm in various driving scenarios. Safety not only concerns technical fault avoidance but also includes the prediction and response to potential hazards. Efficiency measures the autonomous driving system's economy and time management capabilities in completing tasks, thus reflecting the intelligent system's ability to optimize resource utilization. Comfort refers to the riding experience provided by the autonomous driving system to passengers and is an important dimension for evaluating the human-centered design of intelligent systems.

[0094] It should be understood that, in addition to the evaluation indicators shown in the examples above, other evaluation indicators can be selected based on the needs and characteristics of evaluating the intelligence of autonomous vehicle path planning. These will not be illustrated here.

[0095] Step 003: Obtain the expert's subjective weight set for multiple evaluation indicators;

[0096] Specifically, considering the interrelationships among multiple evaluation indicators, this application employs ANP (Analytic Network Process) to calculate subjective weights. ANP, or Analytic Hierarchy Process, is designed for evaluation indicator systems with complex dependencies and feedback relationships within the system. It overcomes the limitation of AHP (Analytic Hierarchy Process) in failing to account for inter-indicator relationships, resulting in more reasonable evaluation results than AHP. The specific steps for obtaining the subjective weight set include:

[0097] Step a: Obtain expert opinion set F for multiple evaluation indicators. m F m ={f m1 ,f m2 ,...,f mn};

[0098] Step b: Based on the opinion set, use ANP to determine any evaluation indicator C. j weight set w j w j ={f 1j ,f 2j ,...,f mj}; This weight set w j The average value is used as the subjective weight value of this indicator. Then the set of subjective weights for all evaluation indicators

[0099] Step 004: Obtain an objective set of weights for multiple evaluation indicators;

[0100] Specifically, the objective weights are calculated using CRITIC (Criteria Importance Though Intercrieria Correlation), a weighting method that calculates weights based on the importance criterion of inter-criteria correlation. The specific implementation includes the following steps:

[0101] Step 1: Determine the information content of any evaluation indicator by examining its volatility and conflict.

[0102] Specifically, in practical processing, the following method can be used: first, calculate the volatility U of the j-th evaluation indicator. j : Among them, U j X j Let X' and X' represent the standard deviation and mean of the j-th evaluation indicator, respectively. ij Let represent the i-th data value of the j-th evaluation indicator; n represents the sample size. Then calculate the conflict Rj of the j-th evaluation indicator. j : Where, r ij Let represent the correlation coefficient between the j-th evaluation indicator and other evaluation indicators; p is the number of evaluation indicators. Then, the information content of the evaluation indicators is determined by considering their volatility and conflict: the information content C of the j-th evaluation indicator is calculated. j :C j =U j R j .

[0103] Step 2: Calculate the objective weight ω′ of the evaluation indicator based on the determined amount of information. j ;

[0104] Specifically, according to the formula Determine the objective weight of the j-th evaluation indicator.

[0105] Step 3: Obtain the objective weight set ω for all evaluation indicators. o ={ω′1,ω′2,...,ω′ j}

[0106] Specifically, the weight set of the guest weight is determined based on each of the objective weights mentioned above.

[0107] It should be understood that the above method may include steps such as removing the influence of the dimensions of various indicators on the weights in the original data and standardizing the original data before data processing to ensure data accuracy.

[0108] As described above, CRITIC can be completed in five steps: indicator standardization, calculation of volatility, calculation of conflict, calculation of information content, and calculation of objective weights. When conducting objective evaluations using CRITIC, the relationships between indicators can be fully explored, leading to more reasonable weight settings.

[0109] Step 005: Determine the optimal combination coefficients of the subjective weight set and the objective weight set based on the consistency between them;

[0110] Specifically, it includes the following steps:

[0111] Step 01: Determine the combination coefficients of the subjective weight set and the objective weight set by finding the Nash equilibrium and constructing the Lagrange function;

[0112] Specifically, the process begins by finding the Nash equilibrium, and then the combination coefficients of the subjective and objective weight sets are determined by constructing a Lagrange function. The details are as follows:

[0113] (1) Finding Nash Equilibrium

[0114] 1) Introduce constraints and establish optimization;

[0115] Let α1 and α2 be the subjective weight vectors ω. s and objective weight vector ω o The combination coefficients are then used to determine the combination weight vector ω:

[0116] ω=α1ω s +α2ω o (1) Based on minimizing the sum of squared deviations between the combined weights and the subjective and objective weights, an optimization can be established:

[0117]

[0118] 2) Construct the Lagrange function;

[0119] From equation (2), we can obtain:

[0120]

[0121] Substituting equation (1) into equation (3) and further simplifying it, we get:

[0122]

[0123] 3) Take the partial derivatives of equation (4) with respect to α1, α2 and λ respectively;

[0124]

[0125] 4) The combined coefficients of the subjective and objective weights are α′1 and α′2, respectively.

[0126] Step 02: Use the Kendall coefficient to assess the consistency of subjective weights; the Kendall coefficient is specifically:

[0127] in, f represents the sum of squared deviations from the mean of all assigned values ​​for the indicators; j It is the sum of the values ​​assigned to the j-th indicator by all experts. It is the average of the sum of all the assigned values ​​for all indicators. In addition, m represents the number of experts, and n represents the number of evaluation indicators.

[0128] When performing the calculation, the consistency of subjective weights W1 is obtained by calculating according to the above formula.

[0129] Step 03: Calculate the consistency of objective weights using the coefficient of variation;

[0130] Specifically, the coefficient of variation is used to calculate the consistency of objective weights. To ensure the comparability of the consistency coefficients between subjective and objective weights (a larger value indicates higher reliability), a reciprocal transformation is used for standardization (because a smaller coefficient of variation value implies better consistency), thus obtaining the consistency W2 of objective weights:

[0131] W2 = 1 / (1 + CV)

[0132] Among them, the coefficient of variation and These are the standard deviation and mean of multiple sets of objective weights, respectively.

[0133] Step 04: Determine the combination coefficients of the subjective weight set and the objective weight set by assigning higher combination coefficients to weights with higher consistency.

[0134] Specifically, the final combination coefficients are adjusted by assigning higher coefficients to weights with higher consistency:

[0135]

[0136] in,

[0137] Calculate the optimal combination weights using the optimal combination coefficients:

[0138]

[0139] Where W1 represents the consistency of subjective weights, and W2 represents the consistency of objective weights. The optimal combination coefficients for subjective weights. The optimal combination coefficients for objective weights.

[0140] Step 006: Evaluate the intelligence of the autonomous vehicle's path planning based on the determined subjective weight set, objective weight set, coefficients of the subjective weight set, and optimal combination coefficients of the objective weight set.

[0141] Specifically, after determining the optimal group coefficients of the subjective weight set and the optimal combination coefficients of the objective weight set, the intelligence of the autonomous vehicle path planning is evaluated based on the determined subjective weight set and objective weight set.

[0142] As can be seen from the above description, the scheme disclosed in this application, given the strong correlation between intelligence evaluation indicators, selects a weighting method that can fully explore the correlation between indicators. Subjective weights are calculated using ANP, while objective weights are determined using the CRITIC method, and the subjective and objective weights are combined. In the combined weighting process, the Kendall coefficient is used to measure the consistency of subjective weights; the larger the value, the higher the credibility. The coefficient of variation is used to measure the consistency of objective weights; the smaller the value, the higher the credibility. Then, a game theory optimization model is used to dynamically adjust the optimal combination coefficients, thereby determining the optimal combined weights.

[0143] In implementation, firstly, ANP is used to construct expert-knowledge-oriented subjective weights, combined with CRITIC to extract objective weights derived from the discrete features of the data. Then, a game equilibrium model of subjective and objective weights is constructed, solving for the Nash equilibrium while simultaneously measuring the internal consistency of the subjective and objective weights to determine the combination coefficients, thus obtaining the combined weights of each evaluation index. Therefore, the scheme disclosed in this application can ensure the accuracy of the path planning intelligence assessment by collecting subjective and objective weight sets and adjusting the coefficients of the objective and subjective weight sets based on their consistency. Compared with existing evaluation methods, this invention has a significant advantage in ranking consistency, thus providing a theoretically rigorous and robust evaluation method for assessing the intelligence level of autonomous vehicles.

[0144] To facilitate understanding of the methods disclosed in this application, specific examples will be provided below.

[0145] First, a typical simulation scenario is constructed for case analysis. The car navigates from a stationary state from the starting point (0,0) to the destination (20,20), with obstacle avoidance as the premise and minimizing energy consumption and travel time as the task objectives. The test scenario is as follows: Figure 2 As shown, Figure 2 In this diagram, Start represents the starting position, Goal represents the target position, Static obstacle represents a static obstacle, Moving obstacle represents a dynamic obstacle, and Path represents the driving trajectory.

[0146] The specific environment for the constructed test scenario is as follows:

[0147] Terrain structure: Construct a flat rectangular area of ​​20m × 20m with a ground friction coefficient of 0.8. Eliminate the interference of terrain undulations on the kinematic model.

[0148] Obstacle distribution: All obstacles are fully visible. There are 3 static obstacles with radii of 1.5m, 1m and 0.5m respectively, and their coordinates are (15, 15), (8, 3) and (5, 5). There is 1 dynamic obstacle with a radius of 1m, which moves back and forth between (5, 10) and (15, 10) at a speed of 1m / s along the x-axis.

[0149] Vehicle parameters: Dimensions are 1m long × 0.5m wide; maximum acceleration is ±1m / s². 2 It adopts a differential drive model with a maximum steering angle of ±30°, a safety distance of 4m, and a path planning algorithm of SAC (Soft Actor-Critic).

[0150] 2. Construct an evaluation index system. When selecting indicators, adhere to the principles of "observability, quantifiability, and strong correlation." Based on the intelligent characteristics of autonomous vehicle path planning, construct an evaluation index system from three aspects: safety, efficiency, and comfort. For example... Figure 3 As shown.

[0151] (1) Security

[0152] Safety is one of the core indicators in evaluating the intelligence of autonomous driving systems. It measures the system's ability to ensure its own driving safety and protect passengers, pedestrians, and other road users from harm in various driving scenarios. Safety involves not only technical fault avoidance but also the prediction and response to potential hazards.

[0153] (2) Effectiveness

[0154] Efficiency measures the economy and time management capabilities of an autonomous driving system in completing a task, thereby reflecting the intelligent system's ability to optimize resource utilization.

[0155] (3) Comfort

[0156] Comfort is the riding experience provided by autonomous driving systems and is an important dimension for evaluating the human-centered design of intelligent systems.

[0157] 3. Collect data on each evaluation metric, including safety (obstacle collision rate). oc Mean lateral acceleration a h ), efficiency (task completion time t, average speed) Total track length (l), energy consumption (ec), comfort (average jerk) Mean curvature The data consisted of 8 indicators across 3 categories, and the differences in indicator dimensions were eliminated through min-max standardization. After normalization, the indicator data evaluation value g was obtained. i As shown in Table 1.

[0158] Table 1. Average values ​​of each evaluation indicator (after normalization)

[0159]

[0160] 4. According to the method of the present invention, the weights of each evaluation index can be calculated. As shown in Table 2.

[0161] Table 2 Weights of each evaluation indicator

[0162]

[0163] 5. Calculate the intelligent evaluation value of the vehicle's path planning in this scenario.

[0164]

[0165] 6. This invention is compared and analyzed for consistency with three existing typical evaluation methods: AHP, AHP-EWM, and ANP-CRITIC. Five cars were configured with five different path planning algorithms (A2C, DWA, PPO, RRT, SAC), and each was tested 17 times in the test scenario (testing stopped when the data variance was less than 5%). Following the above methodological steps, the Kendall coefficient was used to evaluate the consistency of each method's ranking for the five cars. Specifically, as follows... Figure 4 As shown. In Figure 4 In the diagram, the vertical axis represents ranking, and the horizontal axis represents test order. The ranking distribution represents the ranking distribution.

[0166] This method achieved the highest consistency (W = 0.924), indicating higher stability in the resulting ranking order compared to other evaluation methods. The W value obtained by the AHP-EWM combined method was slightly lower, while the purely subjective AHP method exhibited significantly lower reliability (W = 0.812). The consistency of the ANP-CRITIC method (W = 0.892) was also lower than that of this patented method and AHP-EWM. Therefore, the ranking consistency performance of the evaluation methods is ranked as follows: GTKC > AHP-EWM > ANP-CRITIC > AHP.

[0167] This application also provides an intelligent evaluation system for autonomous vehicle path planning, the system comprising:

[0168] The data acquisition module collects raw data for autonomous vehicle path planning; please refer to the detailed description in the methods described above.

[0169] The data processing module constructs multiple related evaluation metrics for autonomous vehicle path planning; for each evaluation metric, it obtains a set of subjective weights from experts; for each evaluation metric, it obtains a set of objective weights; based on the consistency of the subjective and objective weight sets, it determines the combination coefficients of the subjective and objective weight sets; and based on the determined subjective and objective weight sets, their coefficients, and the coefficients of the objective weight sets, it evaluates the intelligence of the autonomous vehicle path planning. See the detailed description in the above method for further details.

[0170] In the above technical solution, by collecting subjective weight sets and objective weight sets, and adjusting the coefficients of the objective weight sets and subjective weight sets according to the consistency of the subjective weight sets and objective weight sets, the accuracy of the intelligent assessment of path planning is guaranteed.

[0171] This application also provides an electronic device, including:

[0172] At least one processor; and

[0173] A memory that is communicatively connected to at least one processor; wherein,

[0174] The memory stores instructions that can be executed by at least one processor to enable the at least one processor to perform any of the methods in the embodiments of this disclosure.

[0175] This application also provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to perform any of the methods in the embodiments of this disclosure.

[0176] This application also provides a computer program product, including a computer program that, when executed by a processor, implements any of the methods in the embodiments of this disclosure.

[0177] The technical solution disclosed herein ensures the accuracy of the intelligent assessment of path planning by collecting subjective weight sets and objective weight sets, and adjusting the coefficients of the objective weight sets and subjective weight sets based on the consistency between the subjective weight sets and objective weight sets.

[0178] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description.

[0179] Figure 5This is a structural block diagram of an electronic device according to an embodiment of the present disclosure. Figure 5 As shown, the electronic device includes a memory 610 and a processor 620. The memory 610 stores a computer program that can run on the processor 620. There can be one or more memories 610 and processors 620. The memory 610 can store one or more computer programs, which, when executed by the electronic device, cause the electronic device to perform the methods provided in the above-described method embodiments. The electronic device may also include a communication interface 630 for communicating with external devices and performing data exchange and transmission.

[0180] If the memory 610, processor 620, and communication interface 630 are implemented independently, they can be interconnected via a bus to communicate with each other. This bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 5 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0181] Optionally, in a specific implementation, if the memory 610, processor 620, and communication interface 630 are integrated on a single chip, then the memory 610, processor 620, and communication interface 630 can communicate with each other through an internal interface.

[0182] It should be understood that the aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. General-purpose processors can be microprocessors or any conventional processor. It is worth noting that the processor can be a processor supporting Advanced Reduced Instruction Set Machines (ARM) architecture.

[0183] Further, optionally, the aforementioned memory may include read-only memory and random access memory, and may also include non-volatile random access memory. The memory may be volatile or non-volatile, or may include both. Non-volatile memory may include read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory may include random access memory (RAM), which serves as an external cache. Many forms of RAM are available by way of example, but not limitation. Examples include Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate Synchronous DRAM (DDR SDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct RAMBUS RAM (DR RAM).

[0184] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this disclosure are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line, DSL) or wireless (e.g., infrared, Bluetooth, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer, or a data storage device such as a server or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., Digital Versatile Discs (DVDs)), or semiconductor media (e.g., Solid State Disks (SSDs)). It is worth noting that the computer-readable storage media mentioned in this disclosure may be non-volatile storage media; in other words, they may be non-transient computer-readable storage media.

[0185] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0186] In the description of the embodiments of this disclosure, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this disclosure. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0187] In the description of the embodiments disclosed herein, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone.

[0188] In the description of embodiments of this disclosure, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of embodiments of this disclosure, unless otherwise stated, "a plurality of" means two or more.

[0189] The above description is merely an exemplary embodiment of this disclosure and is not intended to limit this disclosure. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this disclosure should be included within the protection scope of this disclosure.

Claims

1. A method for evaluating the intelligence of path planning in autonomous vehicles, characterized in that, Includes the following steps: Collect raw data for autonomous vehicle path planning; Construct multiple related evaluation metrics for the intelligence of autonomous vehicle path planning; For multiple evaluation indicators, obtain the expert's subjective weight set; For multiple evaluation indicators, obtain an objective set of weights; Based on the consistency between the subjective weight set and the objective weight set, determine the optimal combination coefficients of the subjective weight set and the objective weight set; The intelligence of autonomous vehicle path planning is evaluated based on the determined subjective weight set, objective weight set, optimal combination coefficient of subjective weight set, and optimal combination coefficient of objective weight set.

2. The method according to claim 1, characterized in that, Also includes: The process of obtaining the expert's subjective weight set for multiple evaluation indicators specifically includes: For multiple evaluation indicators, obtain expert opinion set F. m F m ={f m1 ,f m2 ,...,f mn }; Based on the aforementioned opinion set, any evaluation index C is determined using ANP. j weight set w j w j ={f 1j ,f 2j ,...,f mj }; This weight set w j The average value is used as the subjective weight value of this indicator. Then the set of subjective weights for all evaluation indicators 3. The method according to claim 1, characterized in that, The process of obtaining an objective weight set for multiple evaluation indicators specifically includes: The information content of any evaluation indicator is determined by its volatility and conflict. Calculate the objective weight ω′ of the evaluation indicator based on the determined amount of information. j ; Obtain the objective weight set ω for all evaluation indicators o ={ω′1,ω′2,...,ω′ j }.

4. The method according to claim 3, characterized in that, The process of obtaining an objective weight set for multiple evaluation indicators also includes: Remove the influence of the dimensions of each indicator on the weights in the original data, and standardize the original data.

5. The method according to claim 3, characterized in that, The determination of the information content of an indicator through the volatility and conflict of any evaluation indicator; specifically including: Calculate the volatility U of the j-th evaluation indicator. j : in, U j X j Let X' and X' represent the standard deviation and mean of the j-th evaluation indicator, respectively. ij This represents the i-th data value of the j-th evaluation indicator; n represents the sample size. Calculate the conflict R of the j-th evaluation indicator. j : Where, r ij This represents the correlation coefficient between the j-th evaluation indicator and other evaluation indicators; p is the number of evaluation indicators. Calculate the information content C of the j-th evaluation indicator. j :C j =U j R j .

6. The method according to claim 5, characterized in that, The objective weight of the evaluation indicator is calculated based on the determined amount of information; specifically: According to the formula Determine the objective weight of the j-th evaluation indicator.

7. The method according to any one of claims 1-6, characterized in that, The step of determining the optimal combination coefficients of the subjective weight set and the objective weight set based on their consistency specifically includes: The combination coefficients of the subjective and objective weight sets are determined by finding the Nash equilibrium and constructing the Lagrange function. The Kendall coefficient is used to assess the consistency of subjective weights; the Kendall coefficient is specifically: in, f represents the sum of squared deviations from the mean of all assigned values ​​for the indicators; j It is the sum of the values ​​assigned to the j-th indicator by all experts. It is the average of the sum of all the assigned values ​​for all indicators. Consistency in calculating objective weights using the coefficient of variation; By assigning higher coefficients to weights with higher consistency, the optimal combination coefficients of the subjective weight set and the objective weight set are determined.

8. The method according to claim 7, characterized in that, The combination coefficients of the subjective and objective weight sets are determined by finding the Nash equilibrium and constructing the Lagrange function; specifically, the following steps are included: (1) Finding Nash Equilibrium 1) Introduce constraints and establish optimization; Let α1 and α2 be the subjective weight vectors ω. s and objective weight vector ω o The combination coefficients are then used to determine the combination weight vector ω: ω=α1ω s +a2w o (1) Based on minimizing the sum of squared deviations between the combined weights and the subjective and objective weights, an optimization can be established: 2) Construct the Lagrange function; From equation (2), we can obtain: Substituting equation (1) into equation (3) and further simplifying it, we get: 3) Take the partial derivatives of equation (4) with respect to α1, α2 and λ respectively; 4) The combined coefficients of the subjective and objective weights are α′1 and α′2, respectively.

9. The method according to claim 8, characterized in that, The consistency W2 calculated using the coefficient of variation for objective weights is specifically as follows: W2 = 1 / (1 + CV) Wherein, the coefficient of variation CV = σ ω2 / μ ω2 , σ ω2 and μ ω2 These are the standard deviation and mean of multiple sets of objective weights, respectively.

10. The method according to claim 9, characterized in that, The optimal combination coefficients for the subjective and objective weight sets are determined by assigning higher combination coefficients to weights with higher consistency; specifically, this includes: The final combination coefficients are adjusted by assigning higher coefficients to weights with higher consistency. in, Calculate the optimal combination weights using the optimal combination coefficients: Where W1 represents the consistency of subjective weights, and W2 represents the consistency of objective weights. The optimal combination coefficients for subjective weights. The optimal combination coefficients for objective weights.

11. An intelligent evaluation system for path planning in autonomous vehicles, characterized in that, include: The data acquisition module collects raw data for autonomous vehicle path planning; The data processing module constructs multiple related evaluation metrics for the intelligence of autonomous vehicle path planning; For multiple evaluation indicators, obtain the subjective weight set of experts; for multiple evaluation indicators, obtain the objective weight set; based on the consistency of the subjective weight set and the objective weight set, determine the optimal combination coefficient of the subjective weight set and the objective weight set; evaluate the intelligence of autonomous vehicle path planning based on the determined subjective weight set, objective weight set, optimal combination coefficient of the subjective weight set and the optimal combination coefficient of the objective weight set.

12. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the method of any one of claims 1-10.

13. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-10.

14. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1-10.

Images