CN-122021713-A - Task processing method and device based on robot, computer equipment and medium

CN122021713ACN 122021713 ACN122021713 ACN 122021713ACN-122021713-A

Abstract

The application belongs to the technical field of artificial intelligence, and relates to a task processing method, a device, computer equipment and a medium based on a robot, wherein the method comprises the steps of generating a digital twin data set based on multi-source data; the method comprises the steps of constructing a collaborative scene model based on a digital twin data set, constructing a virtual-real fusion data set based on real data and simulation data, constructing a data mapping relation model based on the virtual-real fusion data set, carrying out dynamic fusion interaction calculation on the virtual-real fusion data set and the collaborative scene model based on the data mapping relation model to obtain a virtual-real interaction result data set, carrying out simulation training on an algorithm to be trained based on the virtual-real interaction result data set and the virtual-real fusion data set to obtain a first algorithm, deploying a robot based on a target algorithm obtained by optimizing the first algorithm, and carrying out task processing based on the obtained target robot. The application can be applied to the robot task processing scene in the financial science and technology field and the digital medical field, and effectively improves the accuracy of task processing.

Inventors

WANG JIANZONG
SUN AOLAN

Assignees

平安科技（深圳）有限公司

Dates

Publication Date: 20260512
Application Date: 20260109

Claims (10)

1. The task processing method based on the robot is characterized by comprising the following steps of: acquiring multi-source data of a real scene related to the robot, and generating a corresponding digital twin data set based on the multi-source data; constructing a corresponding collaborative scene model based on the digital twin data set; Acquiring real data acquired based on the robot, and acquiring acquired simulation data with dimensions corresponding to the real data; constructing a corresponding virtual-real fusion data set based on the real data and the simulation data, and constructing a corresponding data mapping relation model based on the virtual-real fusion data set; based on the data mapping relation model, carrying out dynamic fusion interaction calculation processing on the virtual-real fusion data set and the collaborative scene model by using a preset fusion interaction algorithm to obtain a virtual-real interaction result data set corresponding to the robot; performing simulation training on a preset algorithm to be trained based on the virtual-real interaction result data set and the virtual-real fusion data set to obtain a corresponding first algorithm; Optimizing the first algorithm based on a preset migration adaptation module to obtain a corresponding target algorithm; And deploying the robots based on the target algorithm to obtain corresponding target robots, and processing tasks to be processed based on the target robots.
2. The robot-based task processing method of claim 1, wherein the step of constructing a corresponding collaborative scene model based on the digital twin dataset specifically comprises: constructing a basic model of the simulation scene based on the digital twin data set; Performing scene feature alignment processing on the basic model to obtain a corresponding first generation model; Performing dynamic physical engine embedding processing on the first generation model to obtain a corresponding second generation model; And taking the second generation model as the collaborative scene model.
3. The robot-based task processing method according to claim 1, wherein the step of constructing a corresponding virtual-real fusion data set based on the real data and the simulation data specifically includes: integrating the real data and the simulation data to obtain corresponding first processing data; performing time synchronization processing on the first processing data to obtain corresponding second processing data; Performing noise removal and smoothing on the second processing data to obtain corresponding third processing data; performing mapping fusion processing on the third processing data to obtain corresponding fourth processing data; Performing exception compensation processing on the fourth processed data to obtain corresponding fifth processed data; and taking the fifth processing data as the virtual-real fusion data set.
4. The robot-based task processing method according to claim 1, wherein the step of performing simulation training on a preset algorithm to be trained based on the virtual-real interaction result data set and the virtual-real fusion data set to obtain a corresponding first algorithm specifically comprises: performing scene construction processing based on the collaborative scene model to obtain a corresponding virtual environment; Deploying the algorithm to be trained into the virtual environment; integrating the virtual-real interaction result data set and the virtual-real fusion data set to obtain corresponding joint training data; Preprocessing the combined training data to obtain corresponding target training data; Training the initial algorithm deployed in the virtual environment based on the target training data to obtain a trained specified algorithm; The designated algorithm is taken as the first algorithm.
5. The robot-based task processing method according to claim 1, wherein the step of optimizing the first algorithm by the preset migration adaptation module to obtain a corresponding target algorithm specifically includes: calling a preset characteristic migration calibration algorithm and an environment self-adaption layer based on the migration adaptation module; performing weight adjustment processing on the first algorithm based on the characteristic migration calibration algorithm to obtain a corresponding second algorithm; Performing algorithm adjustment processing on the second algorithm based on the environment self-adaptive layer to obtain a corresponding third algorithm; And taking the third algorithm as the target algorithm.
6. The method for processing tasks based on robots as claimed in claim 5, wherein said step of performing weight adjustment processing on said first algorithm based on said feature migration calibration algorithm to obtain a corresponding second algorithm comprises: extracting features of the appointed layer of the first algorithm to obtain corresponding feature data; Performing differential analysis on the characteristic data based on a preset similarity algorithm to obtain corresponding differential information; Root cause analysis is carried out on the difference information, and a corresponding root cause analysis result is obtained; performing dynamic weight adjustment processing on the first algorithm based on the root cause analysis result to obtain a corresponding fourth algorithm; And taking the fourth algorithm as the second algorithm.
7. The robot-based task processing method of claim 1, wherein the step of generating a corresponding digital twin dataset based on the multi-source data, comprises: acquiring a preset data arrangement strategy; Performing arrangement treatment on the multi-source data based on the data arrangement strategy to obtain corresponding first generated data; Integrating the first generated data based on a preset target format to obtain corresponding second generated data; the second generated data is taken as the digital twin data set.
8. A robot-based task processing device, comprising: The first processing module is used for collecting multi-source data of a real scene related to the robot and generating a corresponding digital twin data set based on the multi-source data; the first construction module is used for constructing a corresponding collaborative scene model based on the digital twin data set; The acquisition module is used for acquiring real data acquired based on the robot and acquiring acquired simulation data with corresponding dimensions with the real data; the second construction module is used for constructing a corresponding virtual-real fusion data set based on the real data and the simulation data and constructing a corresponding data mapping relation model based on the virtual-real fusion data set; The calculation module is used for carrying out dynamic fusion interaction calculation processing on the virtual-real fusion data set and the collaborative scene model by using a preset fusion interaction algorithm based on the data mapping relation model to obtain a virtual-real interaction result data set corresponding to the robot; The training module is used for carrying out simulation training on a preset algorithm to be trained based on the virtual-real interaction result data set and the virtual-real fusion data set to obtain a corresponding first algorithm; the optimization module is used for carrying out optimization processing on the first algorithm based on a preset migration adaptation module to obtain a corresponding target algorithm; the second processing module is used for deploying the robots based on the target algorithm to obtain corresponding target robots and processing tasks to be processed based on the target robots.
9. A computer device comprising a memory and a processor, the memory having stored therein computer readable instructions which when executed by the processor implement the steps of the robot-based task processing method of any of claims 1 to 7.
10. A computer readable storage medium, characterized in that it has stored thereon computer readable instructions, which when executed by a processor, implement the steps of the robot-based task processing method according to any of claims 1 to 7.

Description

Task processing method and device based on robot, computer equipment and medium Technical Field The application relates to the technical field of artificial intelligence, which can be applied to the fields of financial science and technology, digital medical treatment and the like, in particular to a task processing method, a task processing device, computer equipment and a storage medium based on a robot. Background In the current technical field of the simulation of the robot with the body, the design of the existing simulation algorithm has obvious defects, so that the virtual-real fusion capability is weak. Specifically, on one hand, the existing algorithm is severely distorted in scene construction, and physical details and dynamic characteristics in a real environment cannot be fully considered. For example, when an interactive scene between a robot and an object is simulated, physical properties such as material, friction, elasticity and the like of the object and dynamic change processes (such as movement and deformation of the object) of the object are not accurately simulated, so that the simulated scene is greatly different from the real scene. On the other hand, algorithm migration efficiency is low, and when a simulation algorithm is migrated to a real environment, the performance of the robot may be suddenly reduced. The simulation environment is different from the real environment in many aspects, and the existing algorithm cannot effectively adapt to the difference, so that the accuracy of task processing of the robot after the robot integrates the algorithm is low, and the requirements of practical application cannot be met. For example, in the field of business property insurance, a patrol robot equipped with an infrared sensor needs 7×24 hours to monitor the status of factory equipment. If the existing simulation technology is adopted, the simulation scene cannot accurately simulate physical details such as heating conditions of equipment in a factory, vibration and the like when the equipment runs, and dynamic fault changes of the equipment, and when an algorithm trained based on the simulation scene is migrated to an actual inspection robot, the robot cannot accurately identify abnormal states of the equipment, and accuracy and timeliness of equipment monitoring are affected. For another example, in the medical field, a medical accompanying robot needs to accurately sense actions, forces, etc. of a patient during interaction with the patient. The existing simulation technology is difficult to accurately simulate physical characteristics, action habits and dynamic changes in the interaction process of a patient, so that after an algorithm is migrated, the problems of inaccurate actions, unsmooth interaction and the like of a robot in actual accompanying can occur, and high-quality accompanying service can not be provided for the patient. Therefore, it is desirable to provide an improved simulation technique for a robot with a body, so as to enhance the virtual-real fusion capability, improve the performance of the algorithm after the algorithm is migrated, and ensure that the robot can accurately and efficiently process tasks in practical application. Disclosure of Invention The embodiment of the application aims to provide a task processing method, a task processing device, computer equipment and a storage medium based on a robot, so as to solve the technical problem that the accuracy of task processing of the existing robot is low. In a first aspect, a task processing method based on a robot is provided, including: acquiring multi-source data of a real scene related to the robot, and generating a corresponding digital twin data set based on the multi-source data; constructing a corresponding collaborative scene model based on the digital twin data set; Acquiring real data acquired based on the robot, and acquiring acquired simulation data with dimensions corresponding to the real data; constructing a corresponding virtual-real fusion data set based on the real data and the simulation data, and constructing a corresponding data mapping relation model based on the virtual-real fusion data set; based on the data mapping relation model, carrying out dynamic fusion interaction calculation processing on the virtual-real fusion data set and the collaborative scene model by using a preset fusion interaction algorithm to obtain a virtual-real interaction result data set corresponding to the robot; performing simulation training on a preset algorithm to be trained based on the virtual-real interaction result data set and the virtual-real fusion data set to obtain a corresponding first algorithm; Optimizing the first algorithm based on a preset migration adaptation module to obtain a corresponding target algorithm; And deploying the robots based on the target algorithm to obtain corresponding target robots, and processing tasks to be processed based on the target robots. In a second aspect, a rob