Competition-oriented multi-dimensional dynamic real-time scoring method for autonomous driving simulation

Abstract

This invention discloses a multi-dimensional dynamic real-time scoring method for autonomous driving simulation competitions, belonging to the field of autonomous driving simulation testing and competition evaluation technology. It aims to solve the problems of existing scoring technologies, such as single-dimensionality, static and fixed weights, poor data synchronization, and insufficient real-time performance and traceability. This invention constructs a five-layer hierarchical multi-dimensional scoring index system, establishes a millisecond-level data acquisition channel synchronized with the simulation clock, dynamically and adaptively adjusts index weights based on real-time operating conditions and task stages, and combines weighted summation with an immediate penalty mechanism for violations to achieve real-time score fusion. It achieves synchronous output of scoring data across multiple terminals and fully encrypted and traceable storage. This invention significantly improves the accuracy, real-time performance, fairness, and scenario adaptability of autonomous driving competition scoring, and is applicable to the evaluation of various autonomous driving simulation competitions.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving simulation testing and competition evaluation technology, and in particular to a multi-dimensional dynamic real-time scoring device for autonomous driving simulation oriented towards competitions. Background Technology

[0002] The simulation scoring device for autonomous driving competitions is a core component of the autonomous driving simulation competition system. Its function is to quantitatively evaluate and score the driving performance, task completion, and rule compliance of participating autonomous driving systems in a virtual simulation test scenario. This provides a fair and objective judging basis for the autonomous driving competition and also provides quantitative data references for participating teams to optimize their autonomous driving algorithms. Existing autonomous driving competition simulation scoring devices mainly include a data acquisition module, a rule matching module, a fixed-weight scoring calculation module, and a result output module. The data acquisition module collects basic driving data of the tested vehicle during the simulation; the rule matching module pre-stores the basic scoring rules for the competition; the fixed-weight scoring calculation module calculates the score based on preset fixed weights and the collected data; and the result output module outputs the final score after the simulation test.

[0003] The existing autonomous driving competition simulation scoring devices have several shortcomings in practical applications: First, their scoring dimensions are limited, covering only basic rule compliance and task completion, lacking a hierarchical, multi-dimensional indicator system. This makes it impossible to comprehensively quantify and evaluate the core performance of autonomous driving systems, such as smooth operation and emergency response capabilities, thus failing to meet the multi-dimensional scoring requirements of high-level autonomous driving competitions. Second, they employ a static, fixed weight configuration mode, which cannot dynamically and adaptively adjust weights based on the real-time operating conditions of the simulation scenario and the stage of the competition task execution. This makes it impossible to specifically evaluate the core capabilities of autonomous driving systems in different scenarios, and the scoring results cannot truly reflect the actual performance of the algorithm. Third, the real-time performance of data acquisition and scoring calculation is insufficient; data acquisition is not synchronized with the simulation... The true clock achieves strict synchronization and alignment, but suffers from data latency and timestamp misalignment. Furthermore, most rely on offline scoring after simulation, failing to achieve millisecond-level real-time scoring calculations and multi-terminal synchronous output during competition, thus failing to meet the real-time visualization needs of judges and participants. Fourth, the traceability and tamper-proof capabilities of the scoring process are insufficient. The original data, calculation process, and weight adjustment records for the entire scoring process are not encrypted and stored in a chain, making it impossible to ensure the immutability of scoring results and full-process auditing, thus compromising the fairness and authority of the competition. Fifth, it has poor adaptability to different competition rules and subjects, unable to quickly adapt to the rule requirements of different autonomous driving competitions and the scoring focus of different competition subjects, resulting in insufficient versatility and flexibility of the device. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide a multi-dimensional dynamic real-time scoring device for autonomous driving simulation in competitions. This device solves the problems of existing autonomous driving competition simulation scoring devices, such as single scoring dimensions, static and fixed weights, poor data synchronization, insufficient real-time scoring capabilities, weak traceability, and poor scenario adaptability. It achieves multi-dimensional, dynamic, high-precision, and strong real-time scoring in the autonomous driving competition simulation process, while ensuring the traceability of the scoring process and the fairness of the competition.

[0005] To address the aforementioned technical problems, this invention provides the following technical solution: a multi-dimensional dynamic real-time scoring device for autonomous driving simulation oriented towards competitions, comprising a competition adaptation and initialization module, a scoring index system construction module, a synchronous data acquisition and index calculation module, a dynamic weight adjustment and scoring fusion module, and a scoring output and traceability storage module; the competition adaptation and initialization module is connected to the autonomous driving simulation platform and the scoring index system construction module, respectively, and is used to acquire the simulation test scenario parameters, competition rule constraints, scoring dimension thresholds, and basic weight configurations of the target autonomous driving competition, and construct a dedicated scoring rule library matching the target competition; simultaneously, it connects to the simulation clock of the autonomous driving simulation platform. The system adapts to communicate with the vehicle status data interface of the autonomous vehicle under test and the data interface of dynamic traffic participants and environmental elements in the simulation scenario, establishing a millisecond-level data acquisition channel synchronized with the simulation clock. The scoring index system construction module is connected to the competition adaptation and initialization module and the synchronous data acquisition and index calculation module, respectively. Based on the competition scoring rule library, it builds a five-layer hierarchical scoring index system including a core compliance layer, task completion layer, handling performance layer, emergency response layer, and additional reward / penalty layer. Each layer is configured with corresponding quantifiable real-time scoring indicators, index threshold ranges, and basic weight coefficients. The scoring indicators for each layer can be obtained through the millisecond-level data acquisition. The input data source is acquired in real time through the channel; the synchronous data acquisition and index calculation module is connected to the autonomous driving simulation platform, the scoring index system construction module, and the dynamic weight adjustment and scoring fusion module, respectively. It is used to synchronously acquire real-time vehicle status data, simulation scenario dynamic environment data, and competition task real-time progress data of the tested vehicle through the millisecond-level data acquisition channel, using the simulation clock as the sole time reference and following a preset fixed simulation step size; after performing timestamp alignment, outlier removal, and filtering preprocessing on the acquired synchronous data, it calculates the current quantized value of each scoring index in the five-layer hierarchical scoring index system in real time, and simultaneously determines whether each index triggers a preset violation threshold; The dynamic weight adjustment and scoring fusion module is connected to the synchronous data acquisition and index calculation module and the scoring output and traceability storage module, respectively. It is used to identify the working condition type and competition task execution stage of the current simulation scenario in real time. Based on the preset working condition-weight mapping rules and the type of violation event triggered at present, it dynamically and adaptively adjusts the basic weight coefficients of the scoring indicators at each level to generate the real-time weight matrix at the current moment. Based on the current quantized value of each indicator and the real-time weight matrix, it calculates the cumulative real-time score at the current moment by using a weighted summation method combined with the real-time deduction of violation events. At the same time, it performs the corresponding score deduction and score status locking operation for indicators that trigger hard violation thresholds.The scoring output and traceability storage module is connected to the dynamic weight adjustment and scoring fusion module, the competition referee terminal, and the contestant visualization terminal, respectively. It is used to synchronously output the cumulative real-time score, real-time scores of each level of indicators, details of violations, and weight adjustment records to the competition referee terminal and the contestant visualization interface throughout the entire simulation testing process. Simultaneously, it encrypts and stores the synchronously collected data, indicator calculation results, weight adjustment process, and scoring calculation details according to competition specifications, generating a traceable scoring audit file.

[0006] Furthermore, in the competition adaptation and initialization module, a millisecond-level data acquisition channel synchronized with the simulation clock is established. The hardware trigger signal of the simulation clock is used as the synchronization reference for data acquisition, and the data acquisition step size is set to 5-20ms. This ensures that each set of acquired data has a timestamp that is completely aligned with the simulation clock, and the data acquisition delay does not exceed one simulation step. At the same time, the acquisition channel is equipped with a dual redundancy backup mechanism. When the main channel experiences a communication interruption, it automatically switches to the backup channel to complete the data acquisition.

[0007] Furthermore, in the five-layer hierarchical scoring indicator system constructed by the scoring indicator system construction module, the scoring indicators of the core compliance layer include traffic rule compliance, number of collision violations, duration of lane line violations, and number of traffic light violations, with hard violation thresholds set; the scoring indicators of the task completion layer include competition path completion rate, deviation of arrival time of designated task nodes, completion degree of designated actions, and accuracy of destination stopping; the scoring indicators of the handling performance layer include trajectory tracking lateral deviation, longitudinal vehicle speed control deviation, absolute value of vehicle acceleration, and steering wheel angle fluctuation rate; the scoring indicators of the emergency response layer include hazardous condition identification response time, emergency braking deceleration compliance rate, obstacle avoidance path safety, and vehicle stability time after emergency response; and the scoring indicators of the additional reward and punishment layer include vehicle energy consumption economy, bonus points for standardized operation, and deduction points for malicious driving behavior.

[0008] Furthermore, in the synchronous data acquisition and indicator calculation module, a corresponding normalized quantization model is pre-configured for each scoring indicator. The collected raw data can be input into the corresponding normalized quantization model to calculate the real-time quantization value of the indicator in the range of 0-100. For positive indicators, the quantization value increases as the indicator performance improves, and for negative indicators, the quantization value decreases as the indicator performance deteriorates.

[0009] Furthermore, the dynamic weight adjustment and scoring fusion module can identify the current working condition type in real time based on the collected dynamic environment data of the simulation scenario and the real-time status data of the vehicle through a preset working condition classification model. The working condition type includes at least straight-line cruising, curve driving, following vehicle driving, intersection driving, emergency obstacle avoidance, and parking operation. At the same time, it can identify the execution stage of the current competition task based on the real-time progress data of the competition task. The execution stage includes at least the starting stage, the middle stage of task execution, the final sprint stage, and the task completion stage.

[0010] Furthermore, the dynamic weight adjustment and scoring fusion module pre-stores a weight adjustment coefficient matrix corresponding to each work condition type and task execution stage. Based on the identified current work condition type and task execution stage, it can call the corresponding weight adjustment coefficient matrix and perform matrix operations with the basic weight coefficients of each level of indicators to obtain the initial adjustment weight. When a violation event is detected at the current moment, the initial adjustment weight of the corresponding level of indicators is corrected a second time according to the type and severity of the violation event, and finally the real-time weight matrix at the current moment is generated. During the weight adjustment process, the sum of the weight coefficients of all level indicators is kept constant at 1.

[0011] Furthermore, the dynamic weight adjustment and scoring fusion module has a pre-stored list of hard violation thresholds, which includes serious collisions, running red lights, driving in the wrong direction, and driving out of the competition road boundary. When a vehicle being tested is detected to have triggered a hard violation threshold, all values ​​corresponding to that indicator are immediately deducted. At the same time, according to the competition rules, the cumulative real-time score is reduced by a corresponding proportion, or the current score is locked and subsequent score calculations are terminated. A hard violation record and trigger timestamp are generated simultaneously.

[0012] Furthermore, in the dynamic weight adjustment and scoring fusion module, the cumulative real-time score at the current moment is calculated using the following formula: Where St is the cumulative real-time score at the current time t, St−1 is the cumulative real-time score at the previous time, Wi,t is the real-time weight coefficient of the i-th scoring indicator at the current time t, Vi,t is the real-time quantified value of the i-th scoring indicator at the current time t, Pt is the instant deduction value of the violation event triggered at the current time t, and n is the total number of scoring indicators.

[0013] Furthermore, the scoring output and traceability storage module can store all data from the entire process in a chain according to timestamp order. Each data block contains the hash verification value of the previous data block, forming an immutable scoring data blockchain. At the same time, it can generate a PDF audit report containing details of the entire scoring process according to the competition requirements, with the data hash verification value attached to the report.

[0014] Furthermore, the scoring output and traceability storage module can also update the contestant ranking data in real time based on the cumulative real-time scores of all contestants in the same competition, and synchronously output the ranking data to the competition referee terminal and the public display interface.

[0015] Furthermore, the scoring indicator system construction module also configures corresponding indicator subsets and weight templates in the scoring rule library for different autonomous driving competition subjects. The competition subjects include at least urban road driving subjects, highway driving subjects, automatic parking subjects, emergency obstacle avoidance subjects, and extreme handling subjects. When switching competition subjects, the corresponding indicator subsets and weight templates are automatically called.

[0016] Furthermore, the synchronous data acquisition and index calculation module is linked with the dynamic weight adjustment and scoring fusion module. When a sudden dangerous situation is detected in the simulation scenario, the emergency response special scoring window is automatically opened. Within the preset time window, the emergency operation data of the tested vehicle is collected at high frequency, the quantitative values ​​of each index of the emergency response layer are calculated, and the weight coefficient of the emergency response layer index is temporarily increased through the dynamic weight adjustment and scoring fusion module. After the dangerous situation is resolved, the weight is restored to the normal weight configuration.

[0017] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages compared with the prior art: A. This invention establishes a five-level, multi-dimensional scoring index system that comprehensively covers multiple core dimensions such as competition compliance, task completion, control performance, and emergency response capabilities. It solves the problem of the single dimension of existing scoring devices and can provide a comprehensive and refined quantitative evaluation of the overall performance of participating autonomous driving systems, meeting the scoring requirements of high-level autonomous driving competitions.

[0018] B. This invention adopts a dynamic weight adaptive adjustment mechanism, which can automatically adjust the weight coefficients of each scoring indicator according to the real-time identified simulation working condition type and competition task execution stage. It can specifically evaluate the core capabilities of the autonomous driving system in different scenarios. For example, in emergency obstacle avoidance situations, the weight of emergency response indicators can be increased, making the scoring results more consistent with the actual performance of the algorithm and greatly improving the rationality and accuracy of the scoring.

[0019] C. This invention establishes a millisecond-level data acquisition channel synchronized with the simulation clock, achieving synchronous acquisition of multi-source data and strict timestamp alignment, avoiding scoring deviations caused by data delays and misalignments. At the same time, it realizes real-time index calculation and scoring calculation throughout the simulation process, enabling real-time output of scoring results and violation details during the competition, meeting the needs of real-time judges and contestants for visualization, and significantly improving the real-time performance and accuracy of scoring.

[0020] D. This invention achieves full traceability and tamper-proofing of the scoring process by encrypting and chaining the data throughout the entire scoring process, generating an immutable scoring audit file with a hash verification value. This provides a reliable basis for the review and public announcement of competition results, effectively ensuring the fairness and authority of the competition.

[0021] E. By constructing a configurable, dedicated scoring rule library, this invention configures corresponding subsets of indicators and weight templates for different competition subjects, enabling rapid adaptation to the rule requirements and scoring focus of different autonomous driving competitions. This significantly improves the device's versatility, flexibility, and scenario adaptability, without requiring large-scale hardware and software modifications for different competitions.

[0022] F. This invention sets up a hard violation instant processing mechanism and dual redundant data acquisition channels, which can not only process serious violations in a timely and accurate manner to ensure the strict enforcement of competition rules, but also avoid the loss of scoring data due to data acquisition interruption, thereby improving the stability and reliability of device operation.

[0023] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0024] Figure 1 This is a system architecture diagram of the multi-dimensional dynamic real-time scoring method for autonomous driving simulation in competitions according to the present invention. Figure 2 This is a timing diagram of synchronous data acquisition and real-time index calculation for the multi-dimensional dynamic real-time scoring method for autonomous driving simulation in competitions according to the present invention. Figure 3 This is a flowchart of the dynamic weight adjustment and score fusion process of the multi-dimensional dynamic real-time scoring method for autonomous driving simulation in competitions according to the present invention. Figure 4 This is a timing diagram of the scoring output and traceability storage of the multi-dimensional dynamic real-time scoring method for autonomous driving simulation in competitions according to the present invention. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] It should be noted that the terms "vertical," "horizontal," "up," "down," "left," "right," and similar expressions used in this article are for illustrative purposes only and do not represent the only possible implementation.

[0027] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains; the terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to limit the invention; the term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0028] like Figure 1-4 As shown, the present invention discloses a multi-dimensional dynamic real-time scoring device for autonomous driving simulation in competitions, including a competition adaptation and initialization module, a scoring index system construction module, a synchronous data acquisition and index calculation module, a dynamic weight adjustment and scoring fusion module, and a scoring output and traceability storage module. The modules are connected to each other via industrial Ethernet or high-speed bus. At the same time, the device interfaces with an autonomous driving simulation platform, a competition referee terminal, a contestant visualization terminal, and a public display terminal through standardized interfaces.

[0029] In this embodiment, the target competition is the National Undergraduate Autonomous Driving Innovation Competition. The simulation platform used is the Prescan / Simulink co-simulation platform, and the simulation step size is set to 10ms. The specific implementation process is as follows: The first step is pre-competition adaptation and initialization for competition scoring. The competition adaptation and initialization module first acquires the simulation test scenario parameters for this competition (a comprehensive urban road scenario, including intersections, curves, following vehicles, emergency obstacle avoidance, etc.), competition rule constraints, scoring dimension thresholds, and basic weight configurations, constructing a dedicated scoring rule library matching this competition. Simultaneously, it adapts the communication between the Prescan simulation platform's simulation clock interface, the vehicle status data interface of the tested autonomous vehicle, and the data interfaces of dynamic traffic participants and environmental elements within the simulation scenario. A millisecond-level data acquisition channel synchronized with the simulation clock is established, with a acquisition step size set to 10ms, ensuring that each set of acquired data has a timestamp perfectly aligned with the simulation clock, and the data acquisition delay does not exceed 10ms. A dual-redundancy backup mechanism is configured for the acquisition channel: the primary channel uses TCP / IP communication, and the backup channel uses UDP communication. When the primary channel experiences a communication interruption, it automatically switches to the backup channel to complete data acquisition, avoiding the loss of scoring data.

[0030] The second step is to construct a hierarchical, multi-dimensional scoring indicator system. Based on the scoring rule library of this competition, the scoring indicator system construction module builds a five-layer hierarchical scoring indicator system comprising a core compliance layer, a task completion layer, a handling performance layer, an emergency response layer, and an additional reward / penalty layer. Each layer is configured with corresponding quantifiable real-time scoring indicators, indicator threshold ranges, and basic weight coefficients. The specific configurations are as follows: Core Compliance Layer: Basic weight 0.35; scoring indicators include traffic rule compliance, number of collision violations, duration of lane line violations, and number of traffic light violations; hard violation thresholds are set, including serious collisions, running red lights, driving against traffic, and leaving the competition road boundary; Task Completion Layer: Basic weight 0.25; ... The sub-indicators include the completion rate of the competition path, the deviation of the arrival time of the specified task nodes, the completion rate of the specified actions, and the accuracy of the finish line stop; the handling performance layer has a basic weight of 0.15, and the scoring indicators include lateral deviation of trajectory tracking, longitudinal speed control deviation, absolute value of vehicle acceleration, and steering wheel angle fluctuation rate; the emergency response layer has a basic weight of 0.15, and the scoring indicators include the response time for identifying dangerous working conditions, the compliance rate of emergency braking deceleration, the safety of the obstacle avoidance path, and the vehicle stabilization time after emergency response; the additional reward and penalty layer has a basic weight of 0.10, and the scoring indicators include vehicle energy economy, bonus points for standardized operation, and deduction points for malicious driving behavior. Meanwhile, for the three subjects of this competition—urban road driving, emergency obstacle avoidance, and extreme handling—corresponding indicator subsets and weight templates are configured respectively. When switching competition subjects, the corresponding template is automatically called to complete the adaptation.

[0031] The third step is synchronous data acquisition and real-time indicator calculation. The synchronous data acquisition and indicator calculation module uses the simulation clock as the sole time reference. Following a fixed simulation step of 10ms, it synchronously acquires real-time vehicle status data (including vehicle speed, acceleration, steering wheel angle, brake pedal opening, accelerator pedal opening, etc.), dynamic environment data of the simulation scenario (including traffic light status, lane line position, position and speed of surrounding traffic participants, road boundaries, etc.), and real-time progress data of the competition task (including path completion progress, task node completion status, etc.) through a millisecond-level data acquisition channel. After timestamp alignment, outlier removal, and Kalman filtering preprocessing of the acquired synchronous data, the module uses a pre-configured normalized quantization model to calculate the current quantization value of each scoring indicator in the five-level hierarchical scoring indicator system. This yields real-time quantization values ​​for indicators within the 0-100 range. Positive indicators (such as path completion rate) increase with improved performance, while negative indicators (such as lateral deviation) decrease with worsening performance. Simultaneously, it determines whether each indicator triggers a preset violation threshold. When a sudden dangerous situation is detected in the simulation scenario (such as a vehicle braking suddenly or a pedestrian crossing the road), the emergency response scoring window is automatically opened. Within a 2-second time window, the emergency operation data of the tested vehicle is collected at a high frequency with a step size of 5ms, and the quantitative values ​​of each indicator of the emergency response layer are calculated.

[0032] The fourth step is the multi-dimensional dynamic weight adaptive adjustment and real-time scoring fusion. Based on the collected dynamic environment data of the simulation scenario and the real-time vehicle status data, the dynamic weight adjustment and scoring fusion module uses a pre-set random forest condition classification model to identify the current simulation scenario's condition type in real time, including straight-line cruising, curve driving, following other vehicles, intersection crossing, and emergency obstacle avoidance. Simultaneously, based on the real-time progress data of the competition task, it identifies the current competition task's execution stage, including the starting stage, mid-task execution, final sprint stage, and task completion stage. Based on the identified current condition type and task execution stage, it calls the pre-stored corresponding weight adjustment coefficient matrix and performs matrix operations with the basic weight coefficients of each level of indicators to obtain the initial adjustment weights. When a violation event is detected at the current moment, the initial adjustment weights of the corresponding level of indicators are further corrected according to the type and severity of the violation event, ultimately generating the real-time weight matrix for the current moment. During the weight adjustment process, the sum of the weight coefficients of all level indicators remains constant at 1. For example, when an emergency obstacle avoidance situation is identified, the weight of the emergency response layer is temporarily increased from 0.15 to 0.4, while the weights of the core compliance layer and the task completion layer are correspondingly reduced to ensure that the total weight is 1. Based on the current quantitative value of each indicator and the real-time weight matrix, a weighted summation method combined with immediate deduction of points for violations is used to calculate the current cumulative real-time score through a formula. When the tested vehicle triggers a hard violation threshold (such as a serious collision or running a red light), all values ​​corresponding to that indicator are immediately deducted. At the same time, according to the competition rules, the cumulative real-time score is reduced by a corresponding proportion, or the current score is locked and subsequent score calculations are terminated. A hard violation record and trigger timestamp are generated simultaneously.

[0033] The fifth step is real-time scoring output and full-process traceable storage. Throughout the simulation test, the scoring output and traceability storage module synchronously outputs the cumulative real-time scores, real-time scores for each level of indicators, details of violations, and weight adjustment records to the competition judge's terminal and the contestant's visual interface. Simultaneously, based on the cumulative real-time scores of all contestants in the same competition, it updates the contestant ranking data in real time and synchronously outputs the ranking data to the public display interface. Furthermore, the synchronously collected data, indicator calculation results, weight adjustment process, and scoring calculation details are encrypted and stored according to the competition specifications. All data is stored in a chain according to timestamp order, with each data block containing the SHA256 hash check value of the previous data block, forming an immutable scoring data blockchain. After the simulation test, a PDF audit report containing details of the entire scoring process is generated according to the competition requirements, with the data hash check value attached to the report for review and public announcement of the competition results.

[0034] Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Anyone skilled in the art can make various modifications and alterations without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention should be determined by the claims.

Claims

1. A multi-dimensional dynamic real-time scoring method for autonomous driving simulation in competitions, characterized in that, Includes the following steps: S1 Competition Scoring Pre-Adaptation and Initialization: Obtain the simulation test scenario parameters, competition rule constraints, scoring dimension thresholds and basic weight configurations of the target autonomous driving competition, and build a dedicated scoring rule library that matches the target competition; perform communication adaptation on the simulation clock interface of the autonomous driving simulation platform, the vehicle status data interface of the autonomous vehicle under test, and the data interface of dynamic traffic participants and environmental elements in the simulation scenario, and establish a millisecond-level data acquisition channel synchronized with the simulation clock. S2 Layered Multi-Dimensional Scoring Index System Construction: Based on the competition scoring rule library, a five-layered hierarchical scoring index system is built, including the core compliance layer, task completion layer, operational performance layer, emergency response layer, and additional reward and punishment layer. Each layer is configured with corresponding quantifiable real-time scoring indicators, indicator threshold ranges, and basic weight coefficients. The scoring indicators of each layer correspond to input data sources that can be obtained in real time through the data acquisition channel. S3 Synchronous Data Acquisition and Real-time Index Calculation: Using the simulation clock as the sole time reference, and following a preset fixed simulation step size, the system synchronously acquires real-time vehicle status data, simulation scenario dynamic environment data, and competition task real-time progress data of the tested vehicle through the millisecond-level data acquisition channel. After performing timestamp alignment, outlier removal, and filtering preprocessing on the acquired synchronous data, the system calculates the current quantized value of each scoring index in the five-layer hierarchical scoring index system in real time and simultaneously determines whether each index triggers a preset violation threshold. S4 Multi-dimensional Dynamic Weight Adaptive Adjustment and Real-time Scoring Fusion: Real-time identification of the working condition type and competition task execution stage of the current simulation scenario, based on the preset working condition-weight mapping rules and the currently triggered violation event type, dynamically and adaptively adjusts the basic weight coefficients of the scoring indicators at each level, and generates the real-time weight matrix at the current moment. Based on the current quantitative value of each indicator and the real-time weight matrix, the cumulative real-time score at the current moment is calculated by weighted summation combined with real-time deduction of points for violation events. At the same time, the corresponding score is deducted and the score status is locked for indicators that trigger hard violation thresholds. S5 Real-time Scoring Output and Full-Process Traceable Storage: Throughout the simulation test process, the cumulative real-time score, real-time scores of each level of indicators, details of violations, and weight adjustment records are synchronously output to the competition referee terminal and the contestant's visual interface. At the same time, the synchronously collected data, indicator calculation results, weight adjustment process, and scoring calculation details throughout the process are encrypted and stored in accordance with the competition specifications, generating traceable scoring audit files.

2. The method according to claim 1, characterized in that, In step S1, a millisecond-level data acquisition channel synchronized with the simulation clock is established. Specifically, the hardware trigger signal of the simulation clock is used as the synchronization reference for data acquisition, and the data acquisition step size is set to 5-20ms to ensure that each set of acquired data has a timestamp that is completely aligned with the simulation clock, and the data acquisition delay does not exceed one simulation step size. At the same time, a dual redundancy backup mechanism is set for the acquisition channel. When the main channel experiences a communication interruption, it automatically switches to the backup channel to complete the data acquisition, avoiding the loss of scoring data.

3. The method according to claim 1, characterized in that, In step S2, the specific configuration of the five-level hierarchical scoring index system is as follows: The scoring indicators of the core compliance layer include traffic rule compliance, number of collision violations, duration of lane line violations, and number of traffic light violations. These indicators are used to assess the test vehicles' compliance with competition rules and traffic regulations. Hard violation thresholds are set, and points are deducted immediately upon triggering them. The scoring indicators for the task completion layer include the competition path completion rate, the arrival time deviation of the specified task node, the completion degree of the specified action, and the finishing point stopping accuracy, which are used to assess the performance of the tested vehicle in completing the core tasks of the competition. The scoring indicators for the handling performance layer include lateral deviation of trajectory tracking, longitudinal vehicle speed control deviation, absolute value of vehicle acceleration, and steering wheel angle fluctuation rate, which are used to assess the driving smoothness and control precision of the tested vehicle. The scoring indicators of the emergency response layer include hazardous condition identification response time, emergency braking deceleration compliance rate, obstacle avoidance path safety, and vehicle stability time after emergency response, which are used to assess the emergency response capability of the tested vehicle under sudden hazardous conditions. The scoring indicators for the additional reward and punishment layer include vehicle energy economy, bonus points for standardized operation, and deduction points for malicious driving behavior, which are used to supplement the test vehicle's additional performance with rewards and punishments.

4. The method according to claim 1, characterized in that, In step S3, the current quantization value of each scoring indicator is calculated in real time. Specifically, for each scoring indicator, a corresponding normalized quantization model is pre-configured, and the collected raw data is input into the corresponding normalized quantization model to calculate the real-time quantization value of the indicator in the range of 0-100. For positive indicators, the quantization value increases as the indicator performance improves, and for negative indicators, the quantization value decreases as the indicator performance deteriorates.

5. The method according to claim 1, characterized in that, In step S4, the current simulation scenario's operating condition type and competition task execution stage are identified in real time. Specifically, based on the collected dynamic environment data of the simulation scenario and the real-time status data of the vehicle, the current operating condition type is identified in real time through a preset operating condition classification model. The operating condition type includes at least straight-line cruising, curve driving, following vehicle driving, intersection crossing, emergency obstacle avoidance, and parking operation. At the same time, based on the real-time progress data of the competition task, the current competition task execution stage is identified. The execution stage includes at least the starting stage, the middle stage of task execution, the final sprint stage, and the task completion stage.

6. The method according to claim 5, characterized in that, In step S4, the basic weight coefficients of the scoring indicators at each level are dynamically and adaptively adjusted, specifically as follows: Pre-configure the corresponding weight adjustment coefficient matrix for each working condition type and task execution stage; Based on the identified current working condition type and task execution stage, the corresponding weight adjustment coefficient matrix is ​​called, and matrix operation is performed with the basic weight coefficients of each level of indicators to obtain the initial adjustment weight. When a violation event is detected at the current moment, the initial adjustment weights of the corresponding level indicators are adjusted a second time according to the type and severity of the violation event, and finally the real-time weight matrix at the current moment is generated. During the weight adjustment process, the sum of the weight coefficients of all level indicators is kept constant at 1.

7. The method according to claim 1, characterized in that, In step S4, the corresponding score is immediately deducted and the scoring status is locked for the indicators that trigger the hard violation threshold. Specifically, a hard violation threshold list is set in advance in the core compliance layer. The hard violation threshold list includes serious collisions, running red lights, driving in the wrong direction, and driving out of the boundary of the competition road. When the tested vehicle is detected to have triggered a hard violation threshold, all the corresponding scores of the indicator are immediately deducted. At the same time, according to the competition rules, the cumulative real-time score is reduced by a corresponding proportion, or the current score is locked and the subsequent score calculation is terminated. A hard violation record and trigger timestamp are generated simultaneously.

8. The method according to claim 1, characterized in that, In step S5, a traceable scoring audit file is generated. Specifically, all data from the entire process is stored in a chain according to the timestamp order, and each data block contains the hash verification value of the previous data block, forming an immutable scoring data blockchain. At the same time, a PDF audit report containing details of the entire scoring process is generated according to the format required by the competition. The report includes the data hash verification value for the review and public announcement of the competition results.

9. The method according to claim 1, characterized in that, Step S5 further includes: updating the contestant ranking data in real time based on the cumulative real-time scores of all contestants in the same competition, and synchronously outputting the ranking data to the competition referee terminal and the public display interface to realize the real-time dynamic ranking of contestant scores during the competition.

10. The method according to claim 1, characterized in that, Step S2 further includes: configuring corresponding indicator subsets and weight templates in the scoring rule library for different autonomous driving competition subjects. The competition subjects include at least urban road driving subjects, highway driving subjects, automatic parking subjects, emergency obstacle avoidance subjects, and extreme handling subjects. When switching competition subjects, the corresponding indicator subsets and weight templates are automatically called to complete the one-click adaptation of the scoring system.

Images