AI robot responsibility authentication method based on digital certificate

Abstract

The invention relates to the technical field of safety certification, in particular to an AI robot responsibility certification method based on a digital certificate, which comprises a CA mechanism, an AI robot and participants related to the whole life cycle of the AI robot, according to different stages, the participants comprise a data set participant, a model participant, an integrated participant and an application participant, the AI robot and the participants thereof obtain the digital certificates through the CA mechanism, and the participants in all stages sign data of corresponding structures by means of the digital certificates to form traceability vouchers and undertake corresponding responsibilities. According to the invention, the digital certificate is combined with the security authentication technology, the hierarchical traceability voucher is added in the whole life cycle of the AI robot, and the responsibility authentication of the AI robot is carried out through the hierarchical traceability voucher, so that the authenticity of the entity identity related to the AI robot, the integrity of data and the non-repudiation of behaviors are ensured, and the reliability of the AI robot is improved. And a legal basis is provided for related responsibility attribution problems of the AI robot.

Description

Technical Field

[0001] The present invention relates to the field of security authentication technology, and in particular to an AI robot responsibility authentication method based on digital certificates. Background Art

[0002] AI robots, leveraging artificial intelligence (AI), can learn patterns from massive amounts of data, enabling autonomous learning and decision-making. This transcends the technical limitations of traditional robots, transforming them from tools that execute pre-set instructions to intelligent agents that learn from data. Advances in natural language processing and computer vision, along with the development of high-precision components and the Internet of Things (IoT), are enabling AI robots to autonomously analyze environmental information based on data, formulate action plans, and control execution modules to complete specific tasks.

[0003] While AI robots offer services in real-world applications thanks to their ability to autonomously learn, make decisions, and adapt to complex environments, they also present risks of malicious use and loss of control. For example, malicious data, training biases, or deepfakes can lead to AI robots making inaccurate decisions, potentially posing a direct threat to human safety. When AI robots cause accidents, current technology makes it difficult to definitively determine the specific process at fault. Consequently, determining who bears responsibility for AI robot applications presents challenges. Summary of the Invention

[0004] In response to the above problems, the present invention provides an AI robot responsibility authentication method based on digital certificates. By combining digital certificates with security authentication technology, the authenticity of the entity identity related to the AI ​​robot, the integrity of the data, and the non-repudiation of the behavior are ensured, and a legal basis is provided for the attribution of AI robot-related responsibilities through hierarchical traceability certificates.

[0005] To achieve the above objectives, the technical solution of the present invention is: a digital certificate-based AI robot responsibility authentication method, comprising a CA organization, an AI robot and its participants. The participants are involved in the entire life cycle of the AI ​​robot, including dataset participants, model participants, integration participants, and application participants. The AI ​​robot and its participants obtain digital certificates from the CA organization, sign their corresponding information using the digital certificates, and assume corresponding responsibilities, including the following steps: S1. Dataset participants generate and share robot datasets, apply for digital certificates from CA organizations, sign and be responsible for the robot dataset information including the DatasetMetadata metadata structure; S2. The model participant verifies the signature of the robot dataset provided by the dataset participant, uses the robot dataset to train and deploy the AI ​​model, and applies for a digital certificate from the CA organization. The participant signs and is responsible for the generated AI model information containing the ModelMetadata metadata structure. S3. The integration participant verifies the signature of the AI ​​model information provided by the model participant, integrates the traditional robot with the AI ​​model into an AI robot, applies for a digital certificate from the CA organization, signs the AI ​​robot AIRobotInfo information, applies for a device certificate for the AI ​​robot and is responsible for it; S4. Application participants verify the validity of the AI ​​robot device certificate, apply for a digital certificate from the CA, sign and be responsible for the data generated during the management or use of the AI ​​robot.

[0006] As a further solution of the present invention: in step S1, the robot dataset is derived from actual operations in the physical world, including sensor data, motion trajectories, and environmental interactions. The generation of the robot dataset includes the following steps: S11. Analyze the skills required by the robot based on the scenario and determine the corresponding tasks; S12, split the task into multiple subtasks; S13, decompose each subtask into multiple consecutive sequential steps; S14. Record observations, actions, rewards, and other relevant data related to each step to form trajectory data; S15, standardizing the generated trajectory data according to a common format; S16. Storing any metadata related to the dataset in a standardized format to generate a robot dataset; the metadata related to the dataset includes dataset description information, specifically dataset name, version information, usage scenario, robot type, license, data collector, task code, number of trajectories, and dataset generation time; S17. Apply for a digital certificate; S18. Sign the robot dataset; S19. Shared robotics dataset.

[0007] As a further solution of the present invention, the signature of the robot dataset information by the dataset participant and the signature verification of the robot dataset information by the model participant include the following steps: a11. Dataset participants add metadata to the dataset; a12. Shard the data set; a13. Calculate the summary value for each shard; a14. The data set participant signs the shard summary value using the private key; a15. Share robot dataset and signature; a16. Model participants download the dataset and sign it when using it. a17. Extract dataset metadata; a18. Parse the dataset signature; a19. Verify the validity of the dataset participant's certificate. If valid, use the dataset participant's public key to verify the shard signature. If verification passes, the dataset can be used. If invalid or the verification fails, the model participant bears the risk of using the dataset. Among them, since the data volume of the robot dataset is usually large, the robot dataset can be divided into several sub-datasets, and the dataset can be processed in a sharding manner. The dataset can be split into multiple shards, and an independent signature can be generated for each shard to facilitate incremental updates and verification.

[0008] As a further solution of the present invention: in step S2, the steps of the model participant using the robot dataset to train the AI ​​model are as follows: S21, select the robot dataset to obtain data; S22. Verify the signature of the robot dataset information. If the signature verification passes, the dataset can be used. If the signature verification fails, the model participant bears the risks associated with using the dataset. S23. Preprocessing the obtained robot dataset, wherein the preprocessing includes dividing the dataset into a training set, a validation set, and a test set, and performing data cleaning, word segmentation, embedding, padding, normalization, feature extraction, and enhancement. S24, selecting a model architecture suitable for the robot task and configuring model parameters for model training; S25. Optimize the model through fine-tuning to suit specific needs; S26. Select a suitable deployment platform based on actual needs to deploy the trained model and generate an AI model; S27. Apply for a digital certificate; S28. Sign the AI ​​model information; Among them, the metadata related to the AI ​​model includes model name, registration number, version number, parameter scale, software framework, hardware configuration, data-related parameters, model architecture parameters, training strategy parameters and evaluation optimization parameters; the data-related parameters include dataset name, dataset size, training set ratio, data cleaning filter, preprocessing method, and extension part; the model architecture parameters include basic architecture, number of model layers and hidden layer dimensions, activation function, normalization method, number of attention heads and dimensions of each head, and extension part; the training strategy parameters include optimizer, learning rate, number of training rounds, batch size, evaluation interval, and extension part.

[0009] As a further solution of the present invention, the model participants signing the AI ​​model information and the integration participants verifying the signature of the AI ​​model include the following steps: a21. Record the AI ​​model. The AI ​​model must submit a record application to the provincial-level Cyberspace Administration of China. After preliminary review, it will be reviewed and publicly announced by the National Cyberspace Administration of China. a22. Edit AI model metadata files; a23. Calculate a summary value for the metadata file; a24. The model participant signs the file digest value using the private key; a25. Provide metadata files and signatures; a26. Integration participants obtain metadata files and signatures; a27. Verify the AI ​​model registration number. If the AI ​​model registration number is accurate, verify the validity of the model participant's certificate. If valid, the integration participant uses the model participant's public key to verify the document signature. If the verification passes, the AI ​​model can be used. If the AI ​​model registration number is inaccurate, the certificate is invalid, or the verification fails, the integration participant bears the risks of using the AI ​​model.

[0010] As a further solution of the present invention: When an AI robot causes certain consequences in actual applications and accountability is required, the model participants can be required to allow viewing of the AI ​​model source code and verify whether the corresponding parameter settings in the source code are consistent with those described in the signed metadata file. If they are inconsistent, the model participants shall bear the corresponding responsibilities; if they are completely consistent, the test set shall be used to perform an accuracy test. If the accuracy is low, the model participants shall bear certain responsibilities. If the accuracy is high, it can be attributed to the defects of the AI ​​technology itself, reducing or exempting the model participants from responsibility.

[0011] As a further solution of the present invention: in step S3, the steps of integrating the AI ​​robot by the integration participant are as follows: S31. Conduct demand analysis based on actual application scenarios; S32. Design the overall system architecture; S33. Comprehensively consider the requirements and costs to select and build hardware; S34, configure the software environment; S35. Verify the signature of the AI ​​model information used. If the signature verification passes, integrate and optimize the AI ​​model. S36. Debug and test the software and hardware systems to generate AI robots; S37. Apply for a digital certificate; S38. Sign the AI ​​robot information and apply for a device certificate for the AI ​​robot; S39. Provide AI robot products and services; Among them, the integration participant assigns a unique code to each AI robot. When the integration participant applies for a device certificate for the AI ​​robot, he or she needs to submit information related to the AI ​​robot's identity to the CA organization. The AI ​​robot information consists of basic information, hardware information, software information, model information and communication method.

[0012] As a further solution of the present invention: the steps of signing and verifying the AI ​​robot information are as follows: a31. Apply for an integration participant digital certificate; a32. Integrate participants to edit AI robot information files; a33. Calculate the summary value of the AI ​​robot information file; a34. The integration participant uses the private key to sign the summary value of the AI ​​robot information; a35. Submit AI robot information file and signature; a36. The CA agency obtains the AI ​​robot information file and signature; a37. Extract the integration participant certificate and AI robot unique code; a38. Verify the validity of the integration participant certificate. If valid, use the integration participant's public key to verify the signature of the information file. If the signature verification passes, issue the AI ​​robot device certificate. If invalid or the signature verification fails, do not issue the AI ​​robot device certificate. The unique codes of the AI ​​robots can be listed in batches to facilitate integration participants to apply for device certificates for mass-produced AI robots. The AI ​​robot information signature can be placed in the extended field of the AI ​​robot device certificate. If the AI ​​robots in the same batch have information discrepancies or a large proportion of failures, it will be deemed that the integrator should bear the corresponding responsibility; if only a very few AI robots in the same batch have failures, the AI ​​robot information in the extended field can be compared with the actual robot. If the information is exactly the same, the integrator will bear the responsibility; if the information is inconsistent, it means that the AI ​​robot has been modified by its manager or user, and the responsibility needs to be determined based on actual usage.

[0013] As a further solution of the present invention: the application participant uses the AI ​​robot including the following steps: S41. First, verify whether the device certificate of the AI ​​robot is valid, and check whether the actual information of the AI ​​robot is consistent with the information stored in the device certificate extension field. If valid and consistent, continue with the purchase; if invalid or inconsistent, abandon the purchase; S42. The application participant applies for a digital certificate; S43, manage AI robots; S44. Use AI robots; S45. Generate usage data of the AI ​​robot; S46. The application participant signs the usage data and decides whether to share the usage data; The usage data can be shared with dataset participants for use. When using the data set, the dataset participants need to verify the usage data. If the verification is passed, the usage data can be standardized and subsequently processed. If the verification fails and the data is still used, the dataset participants bear the risks brought by generating and sharing the dataset.

[0014] As a further solution of the present invention, the steps for the CA to issue a digital certificate are as follows: S51. CA institutions accept applications for digital certificates for AI robots and related entities involved in different stages; S52. Determine the application type. If the applicant is an individual, verify the identity information. If the identity information is correct, issue a personal certificate. If the applicant is an organization, verify the legality of the organization information. If legal, issue an organization certificate. If the applicant is a device, verify the signature of the AI ​​robot information. If the signature verification passes, issue a device certificate. S53. Manage the life cycle of the digital certificate.

[0015] Compared with the prior art, the present invention has the following beneficial effects: This invention combines digital certificates with security authentication technology to attach hierarchical traceability credentials throughout the life cycle of the AI ​​robot, ensuring the authenticity of the entity identity related to the AI ​​robot, the integrity of the data, and the non-repudiation of the behavior, providing a legal basis for the attribution of responsibility related to the AI ​​robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is an overall flow chart of the AI ​​robot responsibility authentication method based on digital certificates of the present invention; Figure 2 This is a flow chart of signature verification for data set participants in the AI ​​robot responsibility authentication method based on digital certificates of the present invention; Figure 3 This is a flow chart of signature verification of model participants in the AI ​​robot responsibility authentication method based on digital certificates of the present invention; Figure 4 This is a flow chart of signature verification for integrated participants in the AI ​​robot responsibility authentication method based on digital certificates of the present invention. DETAILED DESCRIPTION

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0018] Example 1 refer to Figures 1 to 4 A digital certificate-based AI robot responsibility authentication method includes a CA organization, an AI robot and its participants. The participants are involved in the entire life cycle of the AI ​​robot, including dataset participants, model participants, integration participants, and application participants. The AI ​​robot and its participants obtain digital certificates from the CA organization. The AI ​​robot and its participants use the digital certificates to sign their corresponding information and assume corresponding responsibilities, including the following steps: S1. Dataset participants generate and share robot datasets, apply for digital certificates from CA organizations, sign and be responsible for the robot dataset information including the DatasetMetadata metadata structure; S2. The model participant verifies the signature of the robot dataset provided by the dataset participant, uses the robot dataset to train and deploy the AI ​​model, and applies for a digital certificate from the CA organization. The participant signs and is responsible for the generated AI model information containing the ModelMetadata metadata structure. S3. The integration participant verifies the signature of the AI ​​model information provided by the model participant, integrates the traditional robot with the AI ​​model into an AI robot, applies for a digital certificate from the CA organization, signs the AI ​​robot AIRobotInfo information, applies for a device certificate for the AI ​​robot and is responsible for it; S4. Application participants verify the validity of the AI ​​robot device certificate, apply for a digital certificate from the CA, sign and be responsible for the data generated during the management or use of the AI ​​robot.

[0019] In step S1, the robot dataset is derived from actual operations in the physical world, including sensor data, motion trajectories, and environmental interactions. The generation of the robot dataset includes the following steps: S11. Analyze the skills required by the robot based on the scenario and determine the corresponding tasks; S12, split the task into multiple subtasks; S13, decompose each subtask into multiple consecutive sequential steps; S14. Record observations, actions, rewards, and other relevant data related to each step to form trajectory data; S15, standardizing the generated trajectory data according to a common format; S16. Store any metadata related to the dataset in a standardized format to generate a robot dataset; the metadata related to the dataset includes dataset description information, specifically the dataset name, version information, usage scenario, robot type, license, data collector, task code, number of trajectories, and dataset generation time; S17. Apply for a digital certificate; S18. Sign the robot dataset; S19. Shared robotics dataset.

[0020] The signature of the robot dataset information by the dataset participants and the signature verification of the robot dataset information by the model participants include the following steps: a11. Dataset participants add metadata to the dataset; a12. Shard the data set; a13. Calculate the summary value for each shard; a14. The data set participant signs the shard summary value using the private key; a15. Share robot dataset and signature; a16. Model participants download the dataset and sign it when using it. a17. Extract dataset metadata; a18. Parse the dataset signature; a19. Verify the validity of the dataset participant's certificate. If valid, use the dataset participant's public key to verify the shard signature. If verification passes, the dataset can be used. If invalid or the verification fails, the model participant bears the risk of using the dataset. Among them, since the data volume of the robot dataset is usually large, the robot dataset can be divided into several sub-datasets, and the dataset can be processed in a sharding manner. The dataset can be split into multiple shards, and an independent signature can be generated for each shard to facilitate incremental updates and verification.

[0021] In step S2, the model participant uses the robot dataset to train the AI ​​model as follows: S21, select the robot dataset to obtain data; S22. Verify the signature of the robot dataset information. If the signature verification passes, the dataset can be used. If the signature verification fails, the model participant bears the risks associated with using the dataset. S23. Preprocess the obtained robot dataset, including dividing it into training set, validation set and test set, and performing data cleaning, word segmentation, embedding, padding, normalization, feature extraction and enhancement; S24, selecting a model architecture suitable for the robot task and configuring model parameters for model training; S25. Optimize the model through fine-tuning to suit specific needs; S26. Select a suitable deployment platform based on actual needs to deploy the trained model and generate an AI model. S27. Apply for a digital certificate; S28. Sign the AI ​​model information; Among them, metadata related to AI models include model name, registration number, version number, parameter scale, software framework, hardware configuration, data-related parameters, model architecture parameters, training strategy parameters and evaluation optimization parameters; data-related parameters include dataset name, dataset size, training set ratio, data cleaning filter, preprocessing method, and extension part; model architecture parameters include basic architecture, number of model layers and hidden layer dimensions, activation function, normalization method, number of attention heads and dimensions of each head, and extension part; training strategy parameters include optimizer, learning rate, number of training rounds, batch size, evaluation interval, and extension part.

[0022] The signing of AI model information by model participants and verification of AI model signatures by integration participants include the following steps: a21. AI models must be filed. AI models must submit a filing application to the provincial-level Cyberspace Administration of China. After preliminary review, the National Cyberspace Administration of China will review and make it public. a22. Edit AI model metadata files; a23. Calculate a summary value for the metadata file; a24. The model participant signs the file digest value using the private key; a25. Provide metadata files and signatures; a26. Integration participants obtain metadata files and signatures; a27. Verify the AI ​​model registration number. If the AI ​​model registration number is accurate, verify the validity of the model participant's certificate. If valid, the integration participant uses the model participant's public key to verify the document signature. If the verification passes, the AI ​​model can be used. If the AI ​​model registration number is inaccurate, the certificate is invalid, or the verification fails, the integration participant bears the risks of using the AI ​​model.

[0023] When an AI robot causes certain consequences in actual applications and accountability is required, model participants can be required to allow viewing of the AI ​​model source code and verify whether the corresponding parameter settings in the source code are consistent with those described in the signed metadata file. If they are inconsistent, the model participants shall bear the corresponding responsibility; if they are completely consistent, the test set shall be used for accuracy testing. If the accuracy is low, the model participants shall bear certain responsibilities. If the accuracy is high, it can be attributed to the defects of the AI ​​technology itself, reducing or exempting the model participants from responsibility.

[0024] In step S3, the steps for the integration participant to integrate the AI ​​robot are as follows: S31. Conduct demand analysis based on actual application scenarios; S32. Design the overall system architecture; S33. Comprehensively consider the requirements and costs to select and build hardware; S34, configure the software environment; S35. Verify the signature of the AI ​​model information used. If the signature verification passes, integrate and optimize the AI ​​model. S36. Debug and test the software and hardware systems to generate AI robots; S37. Apply for a digital certificate; S38. Sign the AI ​​robot information and apply for a device certificate for the AI ​​robot; S39. Provide AI robot products and services; Among them, the integration participant assigns a unique code to each AI robot. When the integration participant applies for a device certificate for the AI ​​robot, he or she needs to submit information related to the AI ​​robot's identity to the CA organization. The AI ​​robot information consists of basic information, hardware information, software information, model information and communication methods.

[0025] The steps for AI robot message signing and verification are as follows: a31. Apply for an integration participant digital certificate; a32. Integrate participants to edit AI robot information files; a33. Calculate the summary value of the AI ​​robot information file; a34. The integration participant uses the private key to sign the summary value of the AI ​​robot information; a35. Submit AI robot information file and signature; a36. The CA agency obtains the AI ​​robot information file and signature; a37. Extract the integration participant certificate and AI robot unique code; a38. Verify the validity of the integration participant certificate. If valid, use the integration participant's public key to verify the signature of the information file. If the signature verification passes, issue the AI ​​robot device certificate. If invalid or the signature verification fails, do not issue the AI ​​robot device certificate. The unique codes of AI robots can be listed in batches, which makes it convenient for integration participants to apply for equipment certificates for mass-produced AI robots. The AI ​​robot information signature can be placed in the extended field of the AI ​​robot equipment certificate. If the information of AI robots in the same batch is inconsistent or the proportion of failures is large, it is deemed that the integrator should bear the corresponding responsibility; if only a few AI robots in the same batch have failures, the AI ​​robot information in the extended field can be compared with the actual robot. If the information is exactly the same, the integrator bears the responsibility; if the information is inconsistent, it means that the AI ​​robot has been modified by its manager or user, and the responsibility needs to be determined based on actual usage.

[0026] The following steps are involved in applying AI robots to participants: S41. First, verify whether the device certificate of the AI ​​robot is valid, and check whether the actual information of the AI ​​robot is consistent with the information stored in the device certificate extension field. If valid and consistent, continue with the purchase; if invalid or inconsistent, abandon the purchase; S42. The application participant applies for a digital certificate; S43, manage AI robots; S44. Use AI robots; S45. Generate usage data of the AI ​​robot; S46. The application participant signs the usage data and decides whether to share the usage data; Usage data can be shared with dataset participants for use. When using the data, the dataset participants need to verify the usage data. If the verification is passed, the usage data can be standardized and subsequently processed. If the verification fails and the data is still used, the dataset participants bear the risks brought by generating and sharing the dataset.

[0027] The steps for a CA to issue a digital certificate are as follows: S51. CA institutions accept applications for digital certificates for AI robots and related entities involved in different stages; S52. Determine the application type. If the applicant is an individual, verify the identity information. If the identity information is correct, issue a personal certificate. If the applicant is an organization, verify the legality of the organization information. If legal, issue an organization certificate. If the applicant is a device, verify the signature of the AI ​​robot information. If the signature verification passes, issue a device certificate. S53. Manage the life cycle of the digital certificate.

[0028] Example 2 Based on Example 1, the metadata of the robot dataset in step S1 is described using ASN.1 as follows: DatasetMetadata ::= SEQUENCE { datasetName UTF8String, -- dataset name version UTF8String, -- version number usageScenario UTF8String, -- usage scenario robotType UTF8String, -- robot type license UTF8String, -- license dataCollector Certificate, -- Data Collector Digital Certificate taskId UTF8String, -- task encoding trajectoryCount INTEGER, -- the number of trajectories generationTime GeneralizedTime, --dataset generation time extensions SEQUENCE OF Extension OPTIONAL-- extension field } Extension ::= SEQUENCE { key UTF8String, value OCTET STRING } The data quality of robot datasets is a core factor affecting the effectiveness of AI model training. Adding malicious data to a dataset can lead to serious consequences. Therefore, shared datasets require dataset participants to sign the dataset information and assume corresponding responsibilities. The signing and verification of robot datasets are intended to ensure the integrity of the data and the authenticity of the source, preventing unauthorized tampering or forgery.

[0029] The AI ​​model metadata in step S2 is described using ASN.1 as follows: ModelMetadata ::= SEQUENCE { modelName UTF8String, -- model name recordNumber UTF8String, -- record number version UTF8String, -- version number parameterScale INTEGER, -- parameter scale (unit: billions) softwareFramework UTF8String,-- software framework hardwareConfig HardwareConfiguration, -- hardware configuration dataParameters DataParameters, -- data related parameters modelArchParams ModelArchitectureParams, -- Model architecture parameters trainingStrategy TrainingStrategyParams -- Training strategy parameters extensions SEQUENCE OF Extension OPTIONAL-- extension field } HardwareConfiguration ::= SEQUENCE { gpuType UTF8String, -- GPU model gpuCount INTEGER, -- number of GPUs memoryPerGPU INTEGER, -- single card video memory (unit: GB) interconnectType UTF8String -- interconnection method } DataParameters ::= SEQUENCE { datasetName UTF8String, -- dataset name datasetSize INTEGER, -- dataset size (unit: number of samples) trainValSplit INTEGER, -- training set ratio (e.g. 80 means 80%) dataFilters SEQUENCE OF UTF8String, -- data cleaning filter list preprocessing SEQUENCE OF UTF8String, -- preprocessing method list extensions SEQUENCE OF Extension OPTIONAL-- extension field } ModelArchitectureParams ::= SEQUENCE { baseArchitecture UTF8String, -- Basic model architecture numLayers INTEGER, -- number of model layers hiddenDim INTEGER, -- hidden layer dimension activation UTF8String, -- activation function normalization UTF8String, -- Normalization method attentionHeads INTEGER, -- number of attention heads headDim INTEGER, -- the dimension of each head extensions SEQUENCE OF Extension OPTIONAL-- extension field } TrainingStrategyParams ::= SEQUENCE { optimizer UTF8String, -- optimizer type learningRate REAL, -- learning rate (such as 3e-5) epochs INTEGER, -- number of training rounds batchSize INTEGER, -- batch size evalInterval INTEGER, -- evaluation interval (unit: epoch) extensions SEQUENCE OF Extension OPTIONAL-- extension field } According to my country's current policies, large models must submit a filing application to the provincial-level Cyberspace Administration of China, which will be reviewed and announced by the National Cyberspace Administration of China after preliminary review. If only the model hyperparameters (such as learning rate, batch size) or non-core layers are fine-tuned and the overall architecture or core logic of the model is not changed, then re-filing is usually not required, but it must be recorded in the annual report. If the parameter update involves major modifications to the model architecture, such as the number of neural network layers, and the adjustment of key parameters leads to significant changes in the output results and behavior patterns, it may be regarded as a major change and requires re-filing. Therefore, model participants must sign the large model data metafile for model users to verify the signature.

[0030] In step S3, the structure of describing the AI ​​robot information using ASN.1 is as follows: AIRobotInfo ::= SEQUENCE { baseInfo BaseInfo, -- basic information hardwareInfo HardwareInfo, -- hardware information softwareInfo SoftwareInfo, -- software information modelInfo ModelInfo, -- model information commProtocols CommunicationInfo, -- Communication method extensions SEQUENCE OF Extension OPTIONAL--Extension field } BaseInfo ::= SEQUENCE { manufacturer UTF8String, -- Integrator modelNumbers UTF8String, -- AI robot model uniqueId SEQUENCE OF UTF8String, -- AI robot unique code list productionDate GeneralizedTime -- production date } HardwareInfo ::= SEQUENCE { sensors SEQUENCE OF UTF8String, -- sensor list actuators SEQUENCE OF UTF8String, -- actuator list controlUnit ControlChip, -- control chip degreesOfFreedom INTEGER -- degrees of freedom } ControlChip ::= SEQUENCE { vendor UTF8String, -- chip manufacturer chipModel UTF8String -- chip model } SoftwareInfo ::= SEQUENCE { operatingSystem UTF8String, -- operating system middleware SEQUENCE OF UTF8String, -- middleware list applications SEQUENCE OF UTF8String ,-- Application software list } ModelInfo ::= SEQUENCE { modelName UTF8String, -- model name registrationId UTF8String, -- Model registration number deployment DeploymentMode, -- Model deployment mode } DeploymentMode ::= ENUMERATED { -- Deployment mode enumeration cloud (0), -- Cloud deployment edge (1), -- edge deployment hybrid (2) -- Hybrid deployment } CommunicationInfo ::= SEQUENCE { protocols SEQUENCE OF UTF8String, -- Communication protocols interfaces SEQUENCE OF UTF8String, -- physical interface } In the practical application of AI robots, digital certificates can provide security assurances such as identity authentication, data encryption, and integrity protection. In the Internet of Things, AI robots, as device nodes, require digital certificates to establish trusted identities and ensure secure inter-device communication. For example, collaborative robots in smart manufacturing must verify each other's certificates to prevent malicious devices from accessing the production network. Public key encryption is used to transmit data and protect the confidentiality of communications between AI robots and users or systems. For example, medical AI systems use certificates to encrypt patient data to ensure privacy compliance. Digital signature technology ensures that data is not tampered with during transmission. Digital certificates support auditing and logging, helping to track the operation history of AI robots and provide evidence for security incident investigations. By verifying the signature of AI robot information, a third-party CA organization issues a device certificate for the AI ​​robot.

[0031] In step S4, the AI ​​robot processes input information received from the user or the external environment to drive its decisions, responses, or actions. This input may include multimodal information such as text, voice, images, video, sensor data, structured data from databases or APIs, and real-time data from IoT devices. In some cases, the data generated during the use of the AI ​​robot can become a resource for datasets. While data of different types and sources requires targeted processing based on hardware, algorithms, and scenario requirements, it must be verified through digital certificates and signatures to ensure the authenticity, integrity, and non-repudiation of the data.

[0032] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A digital certificate-based AI robot responsibility authentication method, characterized by: This includes CA organizations, AI robots and their participants. The participants are involved in the entire life cycle of the AI ​​robot, including dataset participants, model participants, integration participants, and application participants. The AI ​​robot and its participants obtain digital certificates from the CA organization. The AI ​​robot and its participants use digital certificates to sign their corresponding information and assume corresponding responsibilities, including the following steps: S1. Dataset participants generate and share robot datasets, apply for digital certificates from CA organizations, sign and be responsible for the robot dataset information including the DatasetMetadata metadata structure; S2. The model participant verifies the signature of the robot dataset provided by the dataset participant, uses the robot dataset to train and deploy the AI ​​model, and applies for a digital certificate from the CA organization. The participant signs and is responsible for the generated AI model information containing the ModelMetadata metadata structure. S3. The integration participant verifies the signature of the AI ​​model information provided by the model participant, integrates the traditional robot with the AI ​​model into an AI robot, applies for a digital certificate from the CA organization, signs the AI ​​robot AIRobotInfo information, applies for a device certificate for the AI ​​robot and is responsible for it; S4. Application participants verify the validity of the AI ​​robot device certificate, apply for a digital certificate from the CA, sign and be responsible for the data generated during the management or use of the AI ​​robot.

2. The AI ​​robot responsibility authentication method based on digital certificates according to claim 1, characterized in that: In step S1, the robot dataset is derived from actual operations in the physical world, including sensor data, motion trajectories, and environmental interactions. The generation of the robot dataset includes the following steps: S11. Analyze the skills required by the robot based on the scenario and determine the corresponding tasks; S12, split the task into multiple subtasks; S13, decompose each subtask into multiple consecutive sequential steps; S14. Record observations, actions, rewards, and other relevant data related to each step to form trajectory data; S15, standardizing the generated trajectory data according to a common format; S16. Storing any metadata related to the dataset in a standardized format to generate a robot dataset; the metadata related to the dataset includes dataset description information, specifically dataset name, version information, usage scenario, robot type, license, data collector, task code, number of trajectories, and dataset generation time; S17. Apply for a digital certificate; S18. Sign the robot dataset; S19. Shared robotics dataset.

3. The AI ​​robot responsibility authentication method based on digital certificates according to claim 2, characterized in that: The signature of the robot dataset information by the dataset participant and the signature verification of the robot dataset information by the model participant include the following steps: a11. Dataset participants add metadata to the dataset; a12. Shard the data set; a13. Calculate the summary value for each shard; a14. The data set participant signs the shard summary value using the private key; a15. Share robot dataset and signature; a16. Model participants download the dataset and sign it when using it. a17. Extract dataset metadata; a18. Parse the dataset signature; a19. Verify the validity of the dataset participant's certificate. If valid, use the dataset participant's public key to verify the shard signature. If verification passes, the dataset can be used. If invalid or the verification fails, the model participant bears the risk of using the dataset. Among them, since the data volume of the robot dataset is usually large, the robot dataset can be divided into several sub-datasets, and the dataset can be processed in a sharding manner. The dataset can be split into multiple shards, and an independent signature can be generated for each shard to facilitate incremental updates and verification.

4. The AI ​​robot responsibility authentication method based on digital certificates according to claim 3 is characterized by: In step S2, the model participant uses the robot dataset to train the AI ​​model as follows: S21, select the robot dataset to obtain data; S22. Verify the signature of the robot dataset information. If the signature verification passes, the dataset can be used. If the signature verification fails, the model participant bears the risks associated with using the dataset. S23. Preprocessing the obtained robot dataset, wherein the preprocessing includes dividing the dataset into a training set, a validation set, and a test set, and performing data cleaning, word segmentation, embedding, padding, normalization, feature extraction, and enhancement. S24, selecting a model architecture suitable for the robot task and configuring model parameters for model training; S25. Optimize the model through fine-tuning to suit specific needs; S26. Select a suitable deployment platform based on actual needs to deploy the trained model and generate an AI model; S27. Apply for a digital certificate; S28. Sign the AI ​​model information; Among them, the metadata related to the AI ​​model includes model name, registration number, version number, parameter scale, software framework, hardware configuration, data-related parameters, model architecture parameters, training strategy parameters and evaluation optimization parameters; the data-related parameters include dataset name, dataset size, training set ratio, data cleaning filter, preprocessing method, and extension part; the model architecture parameters include basic architecture, number of model layers and hidden layer dimensions, activation function, normalization method, number of attention heads and dimensions of each head, and extension part; the training strategy parameters include optimizer, learning rate, number of training rounds, batch size, evaluation interval, and extension part.

5. The AI ​​robot responsibility authentication method based on digital certificates according to claim 4 is characterized in that: The model participants sign the AI ​​model information and the integration participants verify the AI ​​model signature, including the following steps: a21. Record the AI ​​model. The AI ​​model must submit a record application to the provincial-level Cyberspace Administration of China. After preliminary review, it will be reviewed and publicly announced by the National Cyberspace Administration of China. a22. Edit AI model metadata files; a23. Calculate a summary value for the metadata file; a24. The model participant signs the file digest value using the private key; a25. Provide metadata files and signatures; a26. Integration participants obtain metadata files and signatures; a27. Verify the AI ​​model registration number. If the AI ​​model registration number is accurate, verify the validity of the model participant's certificate. If valid, the integration participant uses the model participant's public key to verify the document signature. If the verification passes, the AI ​​model can be used. If the AI ​​model registration number is inaccurate, the certificate is invalid, or the verification fails, the integration participant bears the risks of using the AI ​​model.

6. The AI ​​robot responsibility authentication method based on digital certificates according to claim 5, characterized in that: When an AI robot causes certain consequences in actual applications and accountability is required, model participants can be required to allow viewing of the AI ​​model source code and verify whether the corresponding parameter settings in the source code are consistent with those described in the signed metadata file. If they are inconsistent, the model participants shall bear the corresponding responsibility; if they are completely consistent, the test set shall be used for accuracy testing. If the accuracy is low, the model participants shall bear certain responsibilities. If the accuracy is high, it can be attributed to the defects of the AI ​​technology itself, reducing or exempting the model participants from responsibility.

7. The AI ​​robot responsibility authentication method based on digital certificates according to claim 6, characterized in that: In step S3, the steps for the integration participant to integrate the AI ​​robot are as follows: S31. Conduct demand analysis based on actual application scenarios; S32. Design the overall system architecture; S33. Comprehensively consider the requirements and costs to select and build hardware; S34, configure the software environment; S35. Verify the signature of the AI ​​model information used. If the signature verification passes, integrate and optimize the AI ​​model. S36. Debug and test the software and hardware systems to generate AI robots; S37. Apply for a digital certificate; S38. Sign the AI ​​robot information and apply for a device certificate for the AI ​​robot; S39. Provide AI robot products and services; Among them, the integration participant assigns a unique code to each AI robot. When the integration participant applies for a device certificate for the AI ​​robot, he or she needs to submit information related to the AI ​​robot's identity to the CA organization. The AI ​​robot information consists of basic information, hardware information, software information, model information and communication method.

8. The AI ​​robot responsibility authentication method based on digital certificates according to claim 7, characterized in that: The steps for signing and verifying the AI ​​robot information are as follows: a31. Apply for an integration participant digital certificate; a32. Integrate participants to edit AI robot information files; a33. Calculate the summary value of the AI ​​robot information file; a34. The integration participant uses the private key to sign the summary value of the AI ​​robot information; a35. Submit AI robot information file and signature; a36. The CA agency obtains the AI ​​robot information file and signature; a37. Extract the integration participant certificate and AI robot unique code; a38. Verify the validity of the integration participant certificate. If valid, use the integration participant's public key to verify the signature of the information file. If the signature verification passes, issue the AI ​​robot device certificate. If invalid or the signature verification fails, do not issue the AI ​​robot device certificate. The unique codes of the AI ​​robots can be listed in batches to facilitate integration participants to apply for device certificates for mass-produced AI robots. The AI ​​robot information signature can be placed in the extended field of the AI ​​robot device certificate. If the AI ​​robots in the same batch have information discrepancies or a large proportion of failures, it will be deemed that the integrator should bear the corresponding responsibility; if only a very few AI robots in the same batch have failures, the AI ​​robot information in the extended field can be compared with the actual robot. If the information is exactly the same, the integrator will bear the responsibility; if the information is inconsistent, it means that the AI ​​robot has been modified by its manager or user, and the responsibility needs to be determined based on actual usage.

9. The AI ​​robot responsibility authentication method based on digital certificates according to claim 8, characterized in that: The application participants use the AI ​​robot including the following steps: S41. First, verify whether the device certificate of the AI ​​robot is valid, and check whether the actual information of the AI ​​robot is consistent with the information stored in the device certificate extension field. If valid and consistent, continue with the purchase; if invalid or inconsistent, abandon the purchase; S42. The application participant applies for a digital certificate; S43, manage AI robots; S44. Use AI robots; S45. Generate usage data of the AI ​​robot; S46. The application participant signs the usage data and decides whether to share the usage data; The usage data can be shared with dataset participants for use. When using the data set, the dataset participants need to verify the usage data. If the verification is passed, the usage data can be standardized and subsequently processed. If the verification fails and the data is still used, the dataset participants bear the risks brought by generating and sharing the dataset.

10. The AI ​​robot responsibility authentication method based on digital certificates according to claim 9, characterized in that: The steps for the CA to issue a digital certificate are as follows: S51. CA institutions accept applications for digital certificates for AI robots and related entities involved in different stages; S52. Determine the application type. If the applicant is an individual, verify the identity information. If the identity information is correct, issue a personal certificate. If the applicant is an organization, verify whether the organization information is legal. If legal, issue an organization certificate; If the applicant is a device, the AI ​​robot information will be verified and a device certificate will be issued if the verification passes. S53. Manage the life cycle of the digital certificate.

Images