Intelligent software robot standardization implementation method

Abstract

The invention relates to a standardized implementation method for an intelligent software robot, and belongs to the technical field of process automation, and the method comprises the steps: carrying out the man-machine interaction, man-machine cooperation and artificial cognition based on a software robot system framework, a user and an external environment, assisting the software robot to fully understand a user instruction and external environment information, and improving the user experience. The software robot is controlled to carry out automatic standardized implementation of a complete business process; wherein the software robot system framework comprises a man-machine interaction subsystem, a man-machine cooperation subsystem and an artificial cognition subsystem. By adopting the method, the software robot can be controlled based on a software robot system framework to carry out automatic standardized implementation of a complete business process, the standardization, intelligence and collaboration levels of the software robot are improved, effective collaboration of people and machines in the business process execution process is realized, complex and changeable business challenges are effectively coped, and the business efficiency is improved. And the automation level and the execution efficiency of the business process are improved.

Description

Technical Field

[0001] The present application relates to the field of process automation technology, and in particular to a standardized implementation method for intelligent software robots. Background Art

[0002] In recent years, with the continuous development of automation technology, Robotic Process Automation (RPA) has become an innovative solution for the digital transformation of various industries. By deploying intelligent software robots, it simulates humans to efficiently execute various highly repetitive and rule-defined business processes, and has become a key infrastructure for optimizing enterprise operations and improving work efficiency.

[0003] However, as business processes and requirements become increasingly complex, the design of robotic process automation solutions requires consideration of multiple requirements such as intelligent decision-making, flexible collaboration, feedback loops, and security compliance. Intelligent software robots integrate various advanced artificial intelligence technologies, which are very complex in design and development, and require the selection of technologies at each stage. Existing software robot systems often ignore the effective collaboration between humans and machines, cannot give full play to human advantages in task processes, have certain limitations in handling complex scenarios, and have insufficient descriptions of technical implementations and vague implementation details, which are not conducive to developers' accurate understanding. Therefore, there is an urgent need for a clear and systematic software robot system framework to provide clear guiding principles for the development and implementation of the entire project business process, ensure that the selection of each technical component is based on evidence, and thus promote the standardized, intelligent, and collaborative development of software robots, achieve the optimal combination of technology and business value, and effectively respond to complex and changing business challenges. Summary of the invention

[0004] Based on this, it is necessary to provide a standardized implementation method for intelligent software robots to address the above technical issues.

[0005] A method for implementing standardization of an intelligent software robot, the method comprising:

[0006] Based on the software robot system framework, the robot can interact with users and the external environment through human-machine interaction, human-machine collaboration and artificial cognition, assist the software robot to fully understand user instructions and external environment information, and control the software robot to automatically and standardizedly implement the complete business process; the software robot system framework includes human-machine interaction subsystem, human-machine collaboration subsystem and artificial cognition subsystem;

[0007] The human-computer interaction subsystem is used to perform human-computer interaction through input and feedback loops between the software robot and the user and the external environment, and send the interaction data to the human-computer collaboration subsystem;

[0008] The human-machine collaboration subsystem includes a perception collaboration layer, a decision collaboration layer, and an execution collaboration layer, which are used to perform perception collaboration, decision collaboration, and execution collaboration based on interaction data, and assist the artificial cognition subsystem in human-machine collaborative cognition;

[0009] The artificial cognitive subsystem includes a perception layer, a decision-making layer, and an execution layer. It is used to convert the collected unstructured data and semi-structured data into structured data based on user instructions with the assistance of the human-machine collaborative subsystem, make decisions by analyzing and processing structured data, and control software robots to automatically and standardizedly implement the complete business process based on the decision results.

[0010] Furthermore, the human-computer interaction subsystem includes: a sensor module, a brain-computer interface module, a multimodal interaction module, an interface interaction module, a visual interaction module and a language interaction module.

[0011] Furthermore, the sensor module is used to realize human-machine interaction between the software robot and the user and the external environment through sensors, and to analyze and process environmental changes, user behavior and user identity based on data collected by the sensors;

[0012] The brain-computer interface module is used to realize human-computer interaction between the software robot and the user through the brain-computer interface, and convert the information captured by the brain-computer interface into command output;

[0013] The multimodal interaction module is used to realize human-computer interaction between the software robot and the user through the multimodal interaction device, and integrate the collected multimodal interaction data;

[0014] The interface interaction module is used to realize the human-computer interaction between the software robot and the user through direct manipulation of the interface;

[0015] The visual interaction module is used to realize the human-computer interaction between the software robot and the user through image processing;

[0016] The language interaction module is used to realize human-computer interaction between the software robot and the user through natural language processing.

[0017] Furthermore, the perception and collaboration layer includes a supervisory control module, an information fusion module, and a remote collaboration module, which are used to assist the perception layer in the artificial cognitive subsystem to achieve human-machine collaborative perception during business execution by sharing information and collaborating with the software robot;

[0018] The supervisory control module is used to mark data that is difficult for the software robot to recognize, correct the erroneous recognition results of the software robot, and limit the recognition range of the software robot through manual supervisory control;

[0019] The information fusion module is used to describe the current environment background and realize situational awareness through artificial information fusion, and to realize feature alignment by artificially preprocessing multiple extracted key feature data;

[0020] The remote collaboration module is used to achieve remote perception collaboration through virtual reality, augmented reality and mixed reality.

[0021] Furthermore, the decision-making collaboration layer is used to assist the decision-making layer in the artificial cognitive subsystem to achieve collaborative decision-making between humans and machines during business execution by combining human decision-making with machine decision-making;

[0022] Among them, when making decisions manually, users provide software robots with contextual supplements, experience guidance, error correction, and preference constraint information;

[0023] When making machine decisions, the network data, historical interaction data, manually annotated data, and industry data collected by the software robot are used as training data sets, and the training data sets are input into a large model built through natural language processing technology for training. An explanatory text description is generated based on the trained large model, and the reasoning path and supporting basis for the machine decision are provided based on the explanatory text description.

[0024] Furthermore, after building a large model through natural language processing technology, it also includes:

[0025] The large model is used to extract entities and relationships in the problem, and the extraction results are input into the knowledge graph constructed by ontology, entities, relationships and attributes for query and reasoning. The query and reasoning results based on the knowledge graph constrain the hallucination problem of the large model.

[0026] Furthermore, the execution collaboration layer includes an exception handling module, a communication and collaboration module, and an evaluation and monitoring module, which are used to assist the execution layer in the artificial cognitive subsystem to achieve human-machine collaborative execution during the business execution process by allowing users and software robots to jointly participate in the operation execution;

[0027] Among them, the exception handling module is used to identify abnormal situations during the execution of software robot business, and send the abnormal situations to relevant personnel through email for collaborative processing;

[0028] The communication and collaboration module is used to manually handle demand changes, department collaboration, and scheduling management needs during the execution of software robot business, and provide docking services to customers through manual customer service;

[0029] The evaluation and monitoring module is used to evaluate and monitor the business execution process and results of the software robot, and optimize the execution performance of the software robot based on the evaluation and monitoring results.

[0030] Furthermore, the perception layer is used to convert the collected unstructured data and semi-structured data into structured data based on user instructions with the assistance of the perception collaboration layer, and output the structured data to the decision layer for analysis and processing; wherein the unstructured data and semi-structured data include images, voice, video, files, web pages and components; data conversion methods include optical character recognition, intelligent document processing, document object model, natural language processing, automatic speech recognition and direct component labeling.

[0031] Furthermore, the decision layer is used to obtain the structured data output by the perception layer and analyze and process it with the assistance of the decision collaboration layer, and make decisions using three decision-making methods: rule-based decision-making, human-machine collaborative decision-making, and cognitive intelligent decision-making, and output the decision results to the execution layer for processing;

[0032] Among them, rule-based decision-making is to convert the elements on the screen into an operable interface through screen capture technology, and perform business process operations in sequence based on the constructed flowchart, and compare the operating status data with the pre-built decision table to determine the actions to be taken under the current conditions, and use the rules preset by the rule engine to process complex business logic;

[0033] Human-machine collaborative decision-making is to identify the user's intention through human-machine interaction, and collaborate with software robots to make human-machine decisions based on the intention recognition results;

[0034] Cognitive intelligent decision-making is to automatically identify the specific business that needs to be executed at the moment through perception of the environment, mine the business processes from the interaction logs, understand the various operations that need to be performed to complete the business, schedule and divide the work according to business characteristics, workload and system resource status, and dynamically allocate the business to the most suitable software robot.

[0035] Furthermore, the execution layer is used to control the software robot to automatically and standardizedly implement the complete business process based on the decision results output by the decision layer with the assistance of the execution collaboration layer, and store interaction logs during the implementation process, and monitor and track the status of the business process based on the interaction logs.

[0036] The above-mentioned intelligent software robot standardization implementation method has the following technical effects compared with the existing technology:

[0037] 1. The software robot system framework constructed in this application divides the software robot into three independent but interrelated subsystems: human-computer interaction, human-computer collaboration, and artificial cognition, realizing the standardization and modularization of software robot technical components, and establishing a solid foundation for the research and development and application of software robots. Developers can flexibly select and replace various subsystems to adapt to different application scenarios.

[0038] 2. The human-computer interaction subsystem within the framework performs human-computer interaction through input and feedback loops between the software robot and the user and the external environment, and can achieve high-quality human-computer collaboration; the human-computer collaboration subsystem within the framework assists the artificial cognitive subsystem in perception collaboration, decision-making collaboration, and execution collaboration through the interaction data output by the human-computer interaction subsystem, so that the software robot can fully understand user instructions and external environmental information, and achieve seamless connection and efficient cooperation between man and machine; with the assistance of the human-computer collaboration subsystem, the artificial cognitive subsystem can further accurately simulate the user's cognitive behavior, and achieve understanding, analysis, and reasoning decisions on complex information, thereby completing the automatic standardization implementation of business processes more efficiently and accurately. This application controls software robots to automatically standardize the implementation of complete business processes based on the software robot system framework, improves the standardization, intelligence, and collaboration level of software robots, realizes effective collaboration between humans and machines in the execution of business processes, effectively responds to complex and changing business challenges, and improves the automation level and execution efficiency of business processes. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A schematic diagram of the structure of a software robot system framework in one embodiment;

[0040] Figure 2 A schematic diagram of the relationship between the degree of digitization of a business process and the intelligence level of a demand target, and the risks and benefits of implementing software robot business process automation in an embodiment;

[0041] Figure 3 A schematic diagram of cloud deployment of a software robot in one embodiment. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0043] In one embodiment, a method for implementing intelligent software robot standardization is provided, comprising the following steps:

[0044] Based on the software robot system framework, it conducts human-machine interaction, human-machine collaboration and artificial cognition with users and the external environment, assists the software robot to fully understand user instructions and external environment information, and controls the software robot to automatically and standardizedly implement the complete business process; among them, the software robot system framework is as follows Figure 1 As shown, it includes human-computer interaction subsystem, human-computer collaboration subsystem and artificial cognition subsystem.

[0045] 1. The human-computer interaction subsystem is used to perform human-computer interaction through input and feedback loops between the software robot and the user and the external environment, and send the interaction data to the human-computer collaboration subsystem.

[0046] Furthermore, the human-computer interaction subsystem includes: a sensor module, a brain-computer interface module, a multimodal interaction module, an interface interaction module, a visual interaction module and a language interaction module.

[0047] Among them, the sensor module is used to realize the human-computer interaction between the software robot and the user and the external environment through sensors, and analyze and process environmental changes, user behavior and user identity based on the data collected by the sensors. Specifically, the sensor module includes gyroscopes, pressure sensors, heart rate sensors and biometrics, etc. The gyroscope can detect the rotation angle, thereby sensing the motion state of the device and triggering interactive events; the pressure sensor can detect the pressure changes in the environment and analyze the user's action behavior; the heart rate sensor can monitor the human heart rate and is mostly used for automation in the medical and health field; biometrics include fingerprints, irises, etc., which can be used for user identity verification.

[0048] The brain-computer interface module is used to realize human-computer interaction between the software robot and the user through the brain-computer interface, and convert the information captured by the brain-computer interface into command output. Specifically, the brain-computer interface module can collect brain wave signals through cortical electroencephalogram, brain electrodes, functional magnetic resonance imaging, and magnetoencephalogram, and perceive the activity information of the user's brain through changes in brain waves.

[0049] The multimodal interaction module is used to realize human-machine interaction between the software robot and the user through multimodal interaction devices, and integrate the collected multimodal interaction data. Specifically, the multimodal interaction module enables the software robot to understand complex human instructions more comprehensively through electronic skin tactile simulation, photoelectric fusion perception, multimodal biosensing and multi-parameter sensors.

[0050] The interface interaction module is used to realize the human-computer interaction between the software robot and the user through direct manipulation of the interface. Specifically, the interface interaction module directly manipulates the interface through a mouse, keyboard, touch, etc., and uses NFC (near field communication technology) to exchange information when touching, thereby realizing the most basic human-computer interaction.

[0051] The visual interaction module is used to achieve human-computer interaction between the software robot and the user through image processing. Specifically, the visual interaction module can capture human movements through posture recognition, which is used for interaction in action scenes such as games. It can determine the user's focus through eye tracking in expression recognition and achieve line of sight control. It can identify specific gestures through gesture recognition and send corresponding instructions to the machine. It can also identify users through face comparison and divide operation permissions.

[0052] The language interaction module is used to realize human-computer interaction between the software robot and the user through natural language processing. Specifically, the language interaction module can use speech recognition technology to convert voice input into text. Natural language text can be parsed through NLP (natural language processing) to determine its meaning. Intention recognition can be performed through sentiment analysis. Continuous inquiries, confirmations and feedback can also be performed in the form of human-computer dialogue to fully understand user instructions.

[0053] 2. The human-machine collaborative subsystem includes a perception collaboration layer, a decision collaboration layer, and an execution collaboration layer, which are used to perform perception collaboration, decision collaboration, and execution collaboration based on interaction data, and to assist the artificial cognition subsystem in performing human-machine collaborative cognition.

[0054] Furthermore, the perception and collaboration layer includes a supervisory control module, an information fusion module, and a remote collaboration module, which is used to assist the perception layer in the artificial cognitive subsystem in realizing human-machine collaborative perception during business execution by sharing information and collaborating with the software robot. This layer can enhance the software robot's perception of the external environment or data, and overcome the limitations of traditional automation systems in processing unstructured data and responding to unexpected situations.

[0055] Among them, the supervisory control module is used to mark the data that is difficult for the software robot to recognize through manual supervisory control, correct the software robot's erroneous recognition results, and limit the software robot's recognition range. The supervisory control module improves the accuracy of the software robot's perception of the external environment or data through manual supervisory control. The information fusion module is used to describe the current environmental background and realize situational awareness through manual information fusion, and to realize feature alignment by manually preprocessing multiple extracted key feature data. The information fusion module overcomes the defect that the software robot lacks the ability to perceive all-round information. Through information fusion and feature alignment, it is convenient to compare and analyze various types of data. The remote collaboration module is used to realize remote perception collaboration through virtual reality, augmented reality and mixed reality.

[0056] Furthermore, the decision-making collaboration layer is used to assist the decision-making layer in the artificial cognitive subsystem to achieve collaborative decision-making between humans and machines during business execution by combining manual and machine decision-making. This layer integrates intelligent decision-making capabilities into the automation process through manual intervention, enabling software robots to achieve collaborative decision-making between humans and machines when performing tasks, so as to adapt to more complex and changing business scenarios and improve flexibility and intelligence.

[0057] Among them, when making decisions manually, users provide software robots with scenario supplements, experience guidance, error correction, and preference constraint information. Machine decisions are based on a large amount of data, providing users with information and solutions, and performing interpretation and risk assessment. Specifically, when making decisions, the network data, historical interaction data, manually annotated data, and industry data collected by the software robot are used as training data sets, and NLP is used to understand and analyze the information in the training data sets, and the training data sets are input into the large model built by natural language processing technology for training, and explanatory text descriptions are generated based on the trained large model, and the reasoning path and support basis of the machine decision are provided according to the explanatory text description, thereby comprehensively improving the accuracy of the decision. Furthermore, after building a large model through natural language processing technology, the large model is also used to extract entities and relationships in the problem, and the extraction results are input into the knowledge graph constructed by ontology, entity, relationship, and attribute for query and reasoning, and the query and reasoning results based on the knowledge graph constrain the hallucination problem of the large model, and improve the accuracy and interpretability of the answer. The large model is combined with the knowledge graph to realize the functions of reasoning, prediction, recommendation, evaluation, question and answer, and screening, so as to assist humans in making decisions more accurately.

[0058] Furthermore, the execution collaboration layer includes an exception handling module, a communication and collaboration module, and an evaluation and monitoring module, which are used to assist the execution layer in the artificial cognitive subsystem to achieve human-machine collaborative execution during the business execution process by allowing users and software robots to jointly participate in the operation execution. This layer ensures the accuracy of software robot operations and maintains the compliance and continuity of the process by allowing humans and machines to jointly participate in the operation execution.

[0059] Among them, the exception handling module is used to identify abnormal situations during the execution of software robot business, and send the abnormal situations to relevant personnel via email for collaborative processing. Specifically, when the system is abnormal, an email is automatically sent to notify relevant personnel to check and handle issues such as identity authentication, data input format correction, network connection, hardware equipment repair, interface changes, system maintenance, and security audits. The communication and collaboration module is used to manually handle demand changes, department collaboration, and scheduling management needs during the execution of software robot business, and provide docking services to customers through manual customer service. The evaluation and monitoring module is used to evaluate and monitor the business execution process and results of the software robot, and optimize the execution performance of the software robot based on the evaluation and monitoring results, prevent errors caused by attacks, program vulnerabilities, etc., and continuously improve the software robot according to actual conditions.

[0060] 3. The artificial cognitive subsystem includes a perception layer, a decision-making layer, and an execution layer. It is used to convert the collected unstructured data and semi-structured data into structured data based on user instructions with the assistance of the human-machine collaborative subsystem, make decisions by analyzing and processing structured data, and control the software robot to automatically and standardizedly implement the complete business process based on the decision results.

[0061] Furthermore, the perception layer is the interface for the software robot to interact with the external environment. It mainly collects and inputs data, identifies input information through various technical means, and converts it into structured data that the robot can process. Specifically, with the assistance of the perception collaboration layer, based on user instructions, it converts the collected unstructured data and semi-structured data into structured data, and outputs the structured data to the decision layer for analysis and processing; among them, unstructured data and semi-structured data include data in the form of images, voices, videos, files, web pages, and components; data conversion methods include optical character recognition, intelligent document processing, document object model, natural language processing, automatic speech recognition, and direct component labeling. Specifically, OCR (optical character recognition) is used to extract text from images; IDP (intelligent document processing) converts complex documents into structured data in a standard format by identifying and parsing document content; DOM (document object model) parses the elements in a web page into a tree structure so that they can be accurately located; NLP is used to understand and process natural language text and perform semantic recognition; ASR (automatic speech recognition) converts input speech into text; DPM (Deformable Parts Models) is used for target detection in videos and images.

[0062] Furthermore, the decision-making layer is the core of robotic process automation. It is used to obtain and analyze the structured data output by the perception layer with the assistance of the decision-making collaboration layer, and make decisions using three decision-making methods: rule-based decision-making, human-machine collaborative decision-making, and cognitive intelligent decision-making. The decision results are then output to the execution layer for processing.

[0063] Among them, rule-based decision-making is to convert the elements on the screen into an operable interface through screen capture technology, and to perform business process operations in sequence based on the constructed flowchart, and to determine the actions to be taken under the current conditions by comparing the running status data with the pre-built decision table, and to use the rules preset by the rule engine to process complex business logic. Human-machine collaborative decision-making is to identify the user's intention through human-machine interaction, and to coordinate software robots to make human-machine decisions based on the intention recognition results. That is, human-machine collaborative decision-making enables humans to easily communicate information and issue instructions with robots through human-machine interaction, so as to further understand the needs and provide more accurate intelligent solutions, thereby assisting human decision-making. Cognitive intelligent decision-making is to automatically identify the specific business that needs to be executed at present through the perception of the environment, mine the business process from the interaction log, understand the various operations required to complete the business, and schedule and divide the work according to the business characteristics, workload and system resource status, and dynamically assign the business to the most suitable software robot.

[0064] Furthermore, the execution layer is the part where the software robot interacts with the information system. It is used to control the software robot to automatically and standardizedly implement the complete business process based on the decision results output by the decision layer with the assistance of the execution coordination layer, and store the interaction log during the implementation process, and monitor and track the status of the business process based on the interaction log. Specifically, the execution layer simulates the human operation on the direct control interface such as the mouse and keyboard according to the output of the decision layer, or directly calls the API (application programming interface) to interact with the information system, and realizes the operation of the legacy system database, such as Redis (remote dictionary service database), OSS (object storage service database), MySQL (relational database management system) and SQLServer (relational database), etc., and uses the various services provided by the system, such as application services, file downloads, message sending and information writing, etc., to automatically complete the standardized implementation of the complete business process. At the same time, the interaction log is stored to make the entire operation process transparent and facilitate the tracking of problems.

[0065] In another embodiment, the present application also provides a specific implementation process of a standardized implementation method for an intelligent software robot, which includes the following seven steps: analysis, design, development, testing, deployment, monitoring, and evaluation.

[0066] Step 1: Analysis:

[0067] The purpose of the analysis phase is to clarify requirements, select processes, evaluate feasibility and set goals. By analyzing daily work tasks, we can gain an in-depth understanding of each step of the business process, identify automation opportunities, and select appropriate processes for automation. When selecting a task process, you need to consider the following factors:

[0068] Feasibility: Ensure that the process implementation is based on a feasible idea.

[0069] Process: Provide a comprehensive and accurate description of the process.

[0070] Data: Both input and output are structured data, which reduces the difficulty of implementation.

[0071] Platform: The number of cross-platform operations required in the task. Select the process that requires more cross-platform processing.

[0072] Security: Choose a process that has fewer security issues.

[0073] Process rules: Select a process based on certain rules.

[0074] Value: Choose processes that increase efficiency, save costs, and generate high returns through automated processes.

[0075] Stability: Choose a process that is stable and does not change.

[0076] Complexity: Select a process that is cumbersome to operate.

[0077] Cost: Choose to complete the process manually which is costly.

[0078] Exceptions: Choose a process that is less prone to errors and exceptions.

[0079] Time-consuming: Select processes that take a long time to execute manually.

[0080] Frequency: Select processes that are performed frequently and repeatedly.

[0081] Data scale: Choose processes that require large amounts of data to be processed and leverage the strengths of software robots.

[0082] Standards: Ensure that processes are developed in compliance with laws, regulations and other standards.

[0083] Digital maturity: Choose processes where information is highly digitized and can be easily integrated and processed by software robots.

[0084] Optical Character Recognition: OCR technology is used in the process.

[0085] Number of tasks: Select a process with a large number of tasks per execution.

[0086] Result: Select a process with predictable execution results to facilitate verification of the correctness of the software robot execution.

[0087] Reusability: Select reusable processes to maximize resource utilization.

[0088] Number of robots: The number of software robots allowed in the same instance.

[0089] Intelligence level: whether cognitive ability is required, and choose processes that do not require cognitive ability in the early stages.

[0090] User base: The number of users who use the process automation to assess the scope of impact.

[0091] The relationship between the current degree of digitalization of business processes and the level of intelligence of demand targets and the risks and benefits of implementing software robot business process automation is shown in the following figure: Figure 2 As shown in the figure, “+” indicates recommended implementation and “-” indicates not recommended implementation.

[0092] Step 2: Design:

[0093] The design phase of robotic process automation implementation includes process optimization, tool selection, framework design, security management, etc. According to the specific goals that need to be achieved, appropriate development tools and technologies are selected to ensure the efficiency and security of the robotic process automation solution.

[0094] (1) Process optimization

[0095] The purpose of process optimization is to identify and improve existing business processes, reduce implementation costs and the complexity of subsequent maintenance, and is mainly divided into the following steps:

[0096] Simplify: Remove useless and redundant steps and merge similar tasks.

[0097] Selection: Optimize the selected process and choose the low-cost, high-efficiency and less error-prone solution from multiple equivalent task completion solutions.

[0098] Integration: Since robotic process automation requires end-to-end automation, after determining the processes of each part, they need to be integrated into an executable standardized routine.

[0099] (2) Tool selection

[0100] Choosing the most suitable RPA tool can ensure the smooth implementation of the RPA project. The following factors should be considered when choosing the tool:

[0101] Compatibility: Ensure that the selected RPA tool can be integrated with existing information systems and can be easily connected to various platforms.

[0102] Extensibility: Make sure the RPA tool is extensible and can be integrated with other automation plugins or AI services.

[0103] Ease of use: Choose a simple robotic process automation tool to reduce development and maintenance costs and shorten development cycles.

[0104] (3) Safety management

[0105] Build a safe and reliable automated process to avoid security issues during project implementation, including data leakage, unauthorized access, robot abuse, etc., mainly following the following principles:

[0106] Identity credentials: Encrypted storage of identity credentials ensures the security of access credentials used and prevents leakage.

[0107] Data permissions: Control the robot's access to data and grant only the minimum permissions required to complete the task.

[0108] Data backup: Back up data regularly to prevent data loss caused by operational errors or system failures.

[0109] Faults and vulnerabilities: Regularly and automatically check the system, automatically notify relevant personnel when problems are found, and repair them in time.

[0110] Logging: Record software robot execution logs to facilitate troubleshooting and problem tracking.

[0111] (4) Framework design

[0112] The framework design phase requires building a scalable and maintainable robotic process automation solution architecture to ensure that the automated processes can be flexibly managed and used.

[0113] The software robot system framework constructed in this embodiment is as follows Figure 1 As shown, it includes human-computer interaction subsystem, human-computer collaboration subsystem and artificial cognition subsystem.

[0114] Among them, the human-computer interaction subsystem is used to perform human-computer interaction through the input and feedback loop between the software robot and the user and the external environment, and send the interaction data to the human-computer collaboration subsystem. The human-computer collaboration subsystem includes a perception collaboration layer, a decision collaboration layer, and an execution collaboration layer, which are used to perform perception collaboration, decision collaboration, and execution collaboration based on the interaction data, and assist the artificial cognition subsystem in human-computer collaborative cognition. The artificial cognition subsystem includes a perception layer, a decision layer, and an execution layer, which are used to convert the collected unstructured data and semi-structured data into structured data based on user instructions with the assistance of the human-computer collaboration subsystem, make decisions by analyzing and processing structured data, and control the software robot to automatically and standardizedly implement the complete business process based on the decision results.

[0115] Step 3: Development:

[0116] The development phase requires writing software robot scripts, configuring workflows, and transforming designs into actual executable automation solutions, mainly following the following principles:

[0117] Standardization: Follow a unified coding style and naming conventions to enhance code readability and consistency.

[0118] Security: Ensure code security, avoid leakage of sensitive information, and use encryption technology to process data.

[0119] Performance: Optimize code logic and data processing flow to reduce resource consumption.

[0120] Compliance: Ensure that development complies with relevant laws and industry standards.

[0121] Recording: Write comments and development documents to facilitate understanding and subsequent modifications.

[0122] Step 4: Test:

[0123] After development is completed, testing is required to ensure the reliability and security of the automation process. The following aspects should be noted during the testing process:

[0124] Environmental isolation: Use a virtual shadow system for testing to avoid affecting real system data after testing while protecting sensitive information.

[0125] Exception testing: Triggering abnormal situations to verify whether the error handling mechanism of robotic process automation is effective.

[0126] Performance testing: Evaluate the execution speed, response time, resource usage, etc. of the robotic process automation to ensure that the development performance goals can be achieved.

[0127] Compatibility testing: Test compatibility in different operating systems, browsers, and other environments to ensure that the software robot can run in different environments.

[0128] Documentation: Record the test process and results for easy tracking and subsequent analysis.

[0129] Step 5. Deployment:

[0130] Deploy the tested software robot to the production environment and start executing the specified automated tasks in the actual business. The following principles should be followed during deployment:

[0131] Consistency: Ensure that the deployment environment is consistent with the production environment to reduce operational problems caused by environmental differences.

[0132] Backup: Back up data regularly to avoid data loss due to system failure.

[0133] Cloud deployment: Deploy software robots to the cloud to reduce the control of software robots on local computers. It is highly flexible. Figure 3 shown.

[0134] Step 6: Monitoring:

[0135] During the monitoring phase, it is necessary to continuously monitor the running software robots to ensure the stable operation of the automation process and to promptly discover and solve problems. The following methods can be used:

[0136] Early warning: Set thresholds for monitoring data. When it deviates from the normal range, emails are automatically sent to notify relevant personnel to facilitate quick repair of the problem.

[0137] Visualization: Visualize monitoring data to quickly understand the overall situation of robotic process automation for easy adjustment and management.

[0138] Step 7: Evaluation:

[0139] Evaluate the actual effect of the automation project and for continuous improvement, mainly evaluate from the following aspects:

[0140] Accuracy: Assess the frequency and types of errors that current software robots make when handling tasks.

[0141] Efficiency: Compare the task execution time before and after using the software robot to analyze the efficiency of the software robot.

[0142] Reasonableness: Evaluate the logic of the software robot's execution process and analyze the differences with actual business needs.

[0143] Value: Evaluate the value of the software robot in terms of cost, benefit, etc.

[0144] Calculate the man-hours saved by using software robots to perform various tasks within a certain period, using the cumulative calculation method. The formula is as follows:

[0145]

[0146] In the formula, Total Saved Hours is the total saved working hours; i refers to different task categories, m is the total number of task categories; T human,i Refers to the average time required for manual operation when the task category is i; T RPA,i Refers to the average time required for the software robot to operate when the task category is i; N i Represents the total number of tasks of task category i in the evaluation period.

[0147] The efficiency improvement of using software robots to perform a single type of task is calculated based on the time it takes for humans to perform the task. The formula is as follows:

[0148]

[0149] Where, Efficiency (%) is the percentage of efficiency improvement of using software robots to process a single type of task; T humanT is the average time required for manual execution of a task of this type; RPA The average time it takes for a software robot to perform a task of this type.

[0150] Calculate the efficiency improvement of all tasks using software robots. Use the cumulative method to calculate the time required for manual and software robot execution of tasks respectively, and then use the time required for manual execution of tasks as the standard for calculation. The formula is as follows:

[0151]

[0152] In the formula, Total Efficiency (%) is the percentage of improvement in the overall multi-tasking efficiency of the software robot; i represents different task categories, m is the total number of task categories; N i represents the total number of tasks of task category i in the evaluation period; T human,i Refers to the average time required for manual operation when the task category is i; T RPA,i Refers to the average time required for the software robot to operate when the task category is i.

[0153] The above implementation method decomposes the standardized implementation process of intelligent software robots in a structured manner and proposes relevant standards at each stage, which can promote the adherence to unified specifications during project development, enhance security and compliance, accelerate the development cycle of software robots, and improve the transparency and controllability of project management. In addition, by constructing a software robot system framework, the software robot is divided into three independent but interrelated subsystems: human-computer interaction, human-computer collaboration, and artificial cognition, which realizes the standardization and modularization of software robot technical components, and lays a solid foundation for the research and development and application of software robots. Developers can flexibly select and replace each subsystem to adapt to different application scenarios. The human-computer interaction subsystem within the framework conducts human-computer interaction through input and feedback loops between the software robot and the user and the external environment, and can achieve high-quality collaborative cooperation between humans and machines; the human-computer collaboration subsystem within the framework assists the artificial cognitive subsystem in perception collaboration, decision collaboration, and execution collaboration through the interactive data output by the human-computer interaction subsystem, so that the software robot can fully understand user instructions and external environmental information, and achieve seamless connection and efficient cooperation between humans and machines; with the assistance of the human-computer collaboration subsystem, the artificial cognitive subsystem can further accurately simulate the user's cognitive behavior, and achieve understanding, analysis, and reasoning decisions on complex information, thereby completing the standardized implementation of business processes more efficiently and accurately. This application controls software robots to automatically and standardizedly implement complete business processes based on the software robot system framework, improves the standardization, intelligence, and collaboration level of software robots, realizes effective collaboration between humans and machines in the execution of business processes, effectively responds to complex and changing business challenges, and improves the automation level and execution efficiency of business processes.

[0154] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0155] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A standardized implementation method for intelligent software robots, It is characterized in that The method comprises: Based on the software robot system framework, the software robot performs human-machine interaction, human-machine collaboration and artificial cognition with users and the external environment, assists the software robot to fully understand user instructions and external environment information, and controls the software robot to automatically and standardizedly implement the complete business process; wherein the software robot system framework includes a human-machine interaction subsystem, a human-machine collaboration subsystem and an artificial cognition subsystem; The human-machine interaction subsystem is used to perform human-machine interaction through input and feedback loops between the software robot and the user and the external environment, and send the interaction data to the human-machine collaboration subsystem; The human-machine collaboration subsystem includes a perception collaboration layer, a decision collaboration layer and an execution collaboration layer, which are used to perform perception collaboration, decision collaboration and execution collaboration according to the interaction data, and assist the artificial cognition subsystem in performing human-machine collaborative cognition; The artificial cognitive subsystem includes a perception layer, a decision layer and an execution layer, which are used to convert the collected unstructured data and semi-structured data into structured data based on user instructions with the assistance of the human-machine collaborative subsystem, make decisions by analyzing and processing the structured data, and control the software robot to automatically and standardizedly implement the complete business process based on the decision results.

2. A method for implementing standardization of an intelligent software robot according to claim 1, It is characterized in that The human-computer interaction subsystem includes: a sensor module, a brain-computer interface module, a multimodal interaction module, an interface interaction module, a visual interaction module and a language interaction module.

3. A method for implementing standardization of intelligent software robots according to claim 2, It is characterized in that The sensor module is used to realize human-computer interaction between the software robot and the user and the external environment through sensors, and analyze and process environmental changes, user behavior and user identity based on data collected by the sensors; The brain-computer interface module is used to realize human-computer interaction between the software robot and the user through the brain-computer interface, and convert the information captured by the brain-computer interface into command output; The multimodal interaction module is used to realize human-computer interaction between the software robot and the user through the multimodal interaction device, and integrate the collected multimodal interaction data; The interface interaction module is used to realize human-computer interaction between the software robot and the user through direct manipulation of the interface; The visual interaction module is used to realize human-computer interaction between the software robot and the user by means of image processing; The language interaction module is used to realize human-computer interaction between the software robot and the user through natural language processing.

4. A method for implementing standardization of intelligent software robots according to claim 1, It is characterized in that The perception collaboration layer includes a supervisory control module, an information fusion module, and a remote collaboration module, which are used to assist the perception layer in the artificial cognitive subsystem in realizing human-machine collaborative perception during business execution by sharing information and collaborating with the software robot; The supervisory control module is used to mark data that is difficult for the software robot to recognize, correct the erroneous recognition results of the software robot, and select and limit the recognition range of the software robot through manual supervisory control; The information fusion module is used to describe the current environment background by means of artificial information fusion to achieve situation awareness, and to achieve feature alignment by artificially preprocessing multiple extracted key feature data; The remote collaboration module is used to achieve remote perception collaboration through virtual reality, augmented reality and mixed reality.

5. The method for implementing standardization of an intelligent software robot according to claim 1, It is characterized in that The decision-making coordination layer is used to assist the decision-making layer in the artificial cognitive subsystem in realizing human-machine collaborative decision-making during the business execution process by combining human decision-making with machine decision-making; Among them, when making decisions manually, users provide software robots with contextual supplements, experience guidance, error correction, and preference constraint information; When making machine decisions, the network data, historical interaction data, manually annotated data, and industry data collected by the software robot are used as training data sets, and the training data sets are input into a large model built through natural language processing technology for training. An explanatory text description is generated based on the trained large model, and the reasoning path and supporting basis for the machine decision are provided according to the explanatory text description.

6. A method for implementing standardization of an intelligent software robot according to claim 5, It is characterized in that After building a large model through natural language processing technology, it also includes: The large model is used to extract entities and relationships in the problem, and the extraction results are input into the knowledge graph constructed by ontology, entities, relationships and attributes for query and reasoning. The query and reasoning results based on the knowledge graph constrain the hallucination problem of the large model.

7. The method for implementing standardization of an intelligent software robot according to claim 1, It is characterized in that The execution collaboration layer includes an exception handling module, a communication and collaboration module, and an evaluation and monitoring module, which are used to assist the execution layer in the artificial cognitive subsystem to achieve human-machine collaborative execution during the business execution process by allowing users and software robots to jointly participate in the operation execution; The exception handling module is used to identify abnormal situations during the execution of software robot services and send the abnormal situations to relevant personnel via email for collaborative processing; The communication and collaboration module is used to manually handle demand changes, department collaboration, and scheduling management requirements during the execution of software robot business, and provide docking services to customers through manual customer service; The evaluation and monitoring module is used to evaluate and monitor the business execution process and results of the software robot, and optimize the execution performance of the software robot based on the evaluation and monitoring results.

8. The method for implementing standardization of an intelligent software robot according to claim 1, It is characterized in that The perception layer is used to convert the collected unstructured data and semi-structured data into structured data based on user instructions with the assistance of the perception collaboration layer, and output the structured data to the decision layer for analysis and processing; wherein the unstructured data and semi-structured data include images, voice, video, files, web pages and components; data conversion methods include optical character recognition, intelligent document processing, document object model, natural language processing, automatic speech recognition and direct component labeling.

9. A method for implementing standardization of an intelligent software robot according to claim 1, It is characterized in that The decision layer is used to obtain and analyze the structured data output by the perception layer with the assistance of the decision collaboration layer, and make decisions using three decision-making methods: rule-based decision-making, human-machine collaborative decision-making, and cognitive intelligent decision-making, and output the decision results to the execution layer for processing; The rule-based decision-making is to convert the elements on the screen into an operable interface through screen capture technology, and to perform business process operations in sequence based on the constructed flowchart, and to determine the actions to be taken under the current conditions by comparing the running status data with the pre-constructed decision table, and to process complex business logic using the rules preset by the rule engine; The human-machine collaborative decision-making is to identify the user's intention through human-machine interaction, and to collaborate with the software robot to make human-machine decisions based on the intention identification results; The cognitive intelligent decision-making is to automatically identify the specific business that needs to be executed at present through the perception of the environment, mine the business process from the interaction log, understand the various operations that need to be performed to complete the business, and schedule and divide the work according to the business characteristics, workload and system resource status, and dynamically allocate the business to the most suitable software robot.

10. A method for implementing standardization of an intelligent software robot according to claim 1, It is characterized in that The execution layer is used to control the software robot to automatically and standardizedly implement the complete business process based on the decision results output by the decision layer with the assistance of the execution collaboration layer, and store interaction logs during the implementation process, and monitor and track the status of the business process based on the interaction logs.

Images