256 results about "Target–action" patented technology

The invention relates to the technical field of artificial intelligence, can be applied to business scenes such as personal intelligence, financial science and technology and medical health, and discloses a mechanical arm task planning and control method, device, equipment and medium. Encoding visual information and task description language information to obtain visual features and language features, generating a target action sequence by using a language model, generating a basic path through dynamic planning under an environment model, adjusting the basic path according to environment change in combination with reinforcement learning, generating a final path, and decoding the target action sequence into a joint control instruction. And generating a motion control signal based on the final path, and optimizing a joint control instruction in combination with environment feedback. According to the method, through multi-modal information fusion and mixed path planning, the control instruction is dynamically optimized in combination with environment feedback, and the operation efficiency and task generalization ability of the mechanical arm in a complex environment are improved.

The invention provides a training method and device for a robot control model, a control method and device and electronic equipment, and relates to the technical field of equipment control, and the method comprises the steps: obtaining a preset motion data set, and employing a preset large language model in the robot control model to obtain a preset motion data set; generating an action sequence text and a prediction target direction corresponding to the sample natural language command based on the sample natural language command; and encoding the action sequence text by adopting an initial text encoder in the robot control model, determining a target action code corresponding to the action code from a preset action codebook, and performing action planning according to the target action code and a predicted target direction by adopting an initial action planning model in the robot control model. Generating action strategy information corresponding to the sample natural language command; and parameter adjustment training is carried out on the initial motion planning model in the robot control model. The generalization ability of the robot control model is improved, and control over the robot is more accurate.

The embodiment of the invention provides an action sequence execution method, electronic equipment, a storage medium and a program product. The method comprises the steps that a global action sequence is generated based on semantic features representing user instructions, and the global action sequence comprises a plurality of node actions with an execution sequence; after executing each node action, performing anomaly identification on each node action to obtain an identification result corresponding to each node action; generating a target action sequence taking the target node action as a starting point under the condition that the identification result indicates that the target node action is abnormal; the target action sequence is used for updating the global action sequence; according to the updated global action sequence, interface elements in the operation interface are processed until the execution target of the user instruction is completed. According to the technical scheme of the embodiment of the invention, the technical problem that the success rate of the automatic execution process is not ideal in the prior art is solved.

The invention provides a model training method, a silicon wafer processing method, a silicon wafer processing device, a storage medium and equipment. The model training method comprises the following steps: acquiring initial state information and target state information of a silicon wafer to be processed; through an action decision model, according to the initial state information and the target state information, machining action information for the machining machine table is generated; determining an award parameter corresponding to the processing action information, wherein the award parameter represents the processing quality and efficiency when the processing machine processes the to-be-processed silicon wafer based on the processing action information; if the reward parameter is not more than the preset condition, updating the initial action decision model according to the reward parameter to obtain an updated action decision model; and based on the updated action decision model, returning to execute the step of generating the processing action information through the action decision model according to the initial state information and the target state information until the reward parameter is greater than the preset condition, and obtaining a target action decision model. According to the invention, the processing machine table can be accurately adjusted.

The invention provides a robot control method, a robot and electronic equipment, and the method comprises the steps: planning an initial action sequence of the robot according to the environment information of the robot, correcting a part of action sequences in the initial action sequence according to a target action state of the robot at a current moment, and obtaining a target action sequence; and finally, the robot is controlled according to the target action sequence, the collection moment of the environment information is before the current moment, the initial action sequence comprises initial action states of the robot at multiple moments after the collection moment, and the partial action sequence is the action sequence located after the current moment in the initial action sequence. According to the method, the two steps of planning the action sequence and executing the action sequence can be executed at the same time, and the control efficiency is ensured.

The invention discloses a heterogeneous agent action strategy game method and a heterogeneous agent action strategy game system, which are used for solving the technical problem of poor adaptability of a heterogeneous agent system in a complex environment caused by an existing heterogeneous multi-agent-oriented action strategy game method. The method comprises the steps that when the number of heterogeneous agents meets static confrontation environment conditions, modeling is conducted on the multiple heterogeneous agents, a game problem is generated, and a synthetic unit intelligent game confrontation behavior model is constructed according to the game problem and preset dynamic change factors; solving the synthetic unit intelligent game confrontation behavior model according to the state information and the environment information corresponding to each heterogeneous agent, and generating a plurality of heterogeneous agent target actions; when the number of the heterogeneous agents meets the dynamic confrontation environment condition, a target reinforcement learning algorithm network and a hierarchical convolutional neural network are adopted to generate heterogeneous agent target actions corresponding to the target state information according to the target state information and the environment information of the target game confrontation environment.

Aspects of the present disclosure provide techniques for machine learning based action prediction. Embodiments include providing, as inputs to a recurrent neural network (RNN), an ordered sequence of strings representing actions performed by a user within a software application, the RNN having been trained through a supervised learning process to generate embeddings of the ordered sequence of strings and generate a numerical score relating to a target action based on the embeddings and an order of the ordered sequence of strings. Embodiments include providing, as respective inputs to a tree-based classification machine learning model, the numerical score and an additional feature relating to the user and receiving, as a respective output from the tree-based classification machine learning model in response to the respective inputs, a propensity score indicating a likelihood of the user to perform the target action.

The invention relates to the technical field of intelligent control, and discloses a dexterous hand motion planning and control method based on a visual language motion model, and the method comprises the steps: S1, collecting data, and carrying out the time sequence synchronization and consistency verification of multi-source data; s2, performing dynamic adaptive exponential moving average filtering and calculation processing on the joint sequence, and reconstructing a processing result into adaptive structured features; s3, fusing joint states, inputting the fused joint states into an action head network, splicing online and offline samples into a training batch according to a preset proportion, and determining results to jointly optimize model parameters; s4, performing model reasoning to obtain target action information, performing processing in sequence to generate a smooth low-jitter control sequence, and sending the smooth low-jitter control sequence to the dexterous hand for execution; and S5, the deployment thread issues a control instruction by aligning the annular buffer and the timestamp, triggers a rollback track and allows manual intervention when the control instruction exceeds a threshold value, and writes manual intervention data into an offline buffer. According to the method, efficient, stable and smooth motion control of the dexterous hand can be realized under the condition of extremely few teaching data sets.

The invention relates to the technical field of deep reinforcement learning, and discloses a deep reinforcement learning decision optimization method and device, computer equipment and a medium, and the method comprises the steps: detecting whether a candidate network cluster is in a strategy evaluation state or not; if in the strategy evaluation state, obtaining an evaluation result corresponding to each training node output by the strategy evaluation node; training nodes in the candidate network cluster are optimized according to the evaluation result until the candidate network cluster meets a preset termination condition, and a target action decision network is obtained; and determining a target action node in the target action decision network, and taking an action strategy corresponding to the target action node as a target action strategy. According to the method, the problems that a deep reinforcement learning decision optimization scheme is low in training efficiency, local optimum and experience forgetting are likely to occur, the intelligent agent correlation utilization of a multi-node scheme is insufficient, and a single-node scheme is not suitable for multi-node training are solved.

The invention provides a task processing method, device and equipment based on a virtual auxiliary screen, and the method comprises the steps: displaying a user interaction interface through a main screen, and collecting multimode data from the user interaction interface; determining a user intention based on the multimode data, and generating a task execution sequence based on the user intention; generating a target action sequence based on the task execution sequence; obtaining a target virtual secondary screen corresponding to the target action sequence, sending the target action sequence to the target virtual secondary screen, and performing task processing based on the target action sequence through the target virtual secondary screen; and obtaining a target execution result of the target virtual auxiliary screen for the target action sequence, and displaying the target execution result on a user interaction interface of the main screen. Through the technical scheme of the invention, rapid AI response can be realized, and smooth conversation between the user and the AI agent can be ensured even under the condition that the network is unstable.

The invention discloses a processing method and device, and relates to the technical field of artificial intelligence and equipment application, and the disclosed processing device comprises an action processing module which is used for obtaining function units of at least one application program on electronic equipment to form an action set, and the action set represents actions which can be executed by the electronic equipment; the intelligent processing module is used for responding to input data and determining a target action matched with the intention represented by the input data from the action set through a target model; the intention is an intention related to an action.

The embodiment of the invention discloses an action processing method, device, system and equipment, a storage medium and a computer program product. The method comprises the following steps: acquiring point cloud data corresponding to each target object in a target space; recognizing the action of the target object based on the point cloud data to obtain the action corresponding to the target object; if the action is the target action, indicating the movable equipment to move to an area corresponding to the target object, and obtaining feedback information obtained by interaction between the movable equipment and the target object; and processing an event corresponding to the target action based on the target action and the feedback information. According to the method, action recognition is carried out by acquiring the point cloud data of the target object, a user does not need to wear any equipment, user privacy leakage is avoided while the action recognition accuracy is improved, meanwhile, the mobile equipment is linked to interact with the user, the event is processed according to the feedback information obtained through interaction, and the user experience is improved. And the action processing efficiency and the event processing capability are improved.

The application discloses a blockchain task allocation method and device based on DQN, equipment and medium, relates to the field of task allocation, comprising: determining the state vector of the DQN model, selecting the target action according to the state vector, and determining the predicted execution time and the actual execution time of the target action; the predicted execution time and the actual execution time are used as the input of the preset time length reward formula to obtain the time length reward; a plurality of experiences are randomly extracted, the target Q value of the DQN model is determined based on the plurality of experiences, and the DQN model is updated through the observation Q value and the target Q value of the DQN model; multiple rounds of updating are performed until the DQN model meets the preset requirements to obtain the target DQN model; the target blockchain task allocation model is constructed through the target DQN model, and after receiving the blockchain task allocation request, the target blockchain task allocation model is used for intelligent task allocation. Therefore, the task allocation efficiency can be improved, the dynamic environment change can be responded to, and the manual intervention can be reduced.

The invention provides an interaction method, an intelligent agent training method and device and electronic equipment, and the method comprises the steps: receiving a user instruction; generating a first target action associated with the first user interface according to the first page description information of the first user interface and the user instruction; and executing the first target action on the first user interface to realize the interactive response to the user instruction, thereby realizing the intellectualization and automation of user interface interaction, remarkably improving the speed and accuracy of the response to the user instruction, and improving the user experience. Meanwhile, user interfaces (including complex multi-element interfaces or simple single-function interfaces) with different structures and designs can be flexibly coped with, the consistency and satisfaction degree of user experience are enhanced, and operation under various application scenes is smoother and more natural.

The invention discloses a multi-step task reasoning and planning method and electronic equipment, and belongs to the technical field of robot control. The method comprises the steps of obtaining target task input, and determining a first action plan based on the target task input; candidate actions are obtained, a composite scoring function is generated, the candidate actions are associated with the first action plan, and the composite scoring function is used for scoring the candidate actions; decomposing the first action plan into multi-level tasks by using a multi-level planning algorithm, calculating the total cost, and selecting an execution path with the minimum total cost as a target action plan; and calculating a state difference value, and determining whether to trigger re-planning or not based on the state difference value. According to the invention, the robot control efficiency and accuracy are improved, and the robustness is enhanced.

The invention relates to the technical field of data set augmentation, and particularly discloses a data augmentation method for a training data set of an action recognition model. According to the method, the target action instance can be generated according to actual needs by utilizing the action generation large model, the initial data set is obtained according to the target action instance, and the initial data set is generalized by virtue of the visual large model to generate the augmented data set as the target training data, so that the use value of the data in the A I field is improved; the defects that traditional simulation data is poor in generalization, small in scale and high in generation cost are overcome; and the generation method does not depend on manual data labeling, so that the problems of wrong labeling, inconsistent labeling and the like caused by manual experience errors are avoided, the defects that manual data labeling is difficult to scale, the precision is poor, the cost is high and the efficiency is low are overcome, and the data scale, the accuracy, the data generalization and the construction efficiency of the data set are improved.

The invention discloses an available time pushing method and device and a storage medium, and the method comprises the steps: firstly receiving request information used for requesting to push target available total time, and then obtaining the predicted available time of a target object for carrying out a target action at each candidate place in each unit time period according to the request information, obtaining reachable information between every two candidate places; according to all the predicted available time and all the reachable information, the total available time of the target is obtained through calculation, dynamic time planning is carried out by comprehensively considering the space-time relation between time and places, and the sum of the available time of the target object for carrying out target actions in multiple unit time periods can be calculated more accurately and comprehensively. According to the method, the processing efficiency of time planning and the reasonability of time planning are effectively improved, so that a more accurate time planning scheme with a better optimization effect can be pushed to the target object, and the satisfaction degree of the target object for time planning is improved.

The application provides a virtual object control method, a virtual object control device, a computer readable storage medium and an electronic device, and relates to the technical field of virtual interaction. In the method, a user can trigger a sliding operation from a first control to a second control to control a virtual object to perform a target action related to the first control and the second control. Compared with the related art, the user does not need to press the first control and the second control respectively to trigger the target action. The user only needs to slide from the first control to the second control to achieve "one-key" triggering of the target action (such as a continuous attack action). This can improve the casting efficiency of the target action and reduce the cost of the user learning to cast the target action.

The embodiment of the invention discloses a task execution method and device, electronic equipment, a computer storage readable medium and a program product. The method comprises the following steps: acquiring a plurality of to-be-executed actions in a current interaction state; executing multiple rounds of action screening operation; wherein each action screening operation is used for screening candidate actions meeting the cost requirement in the current round from the action list determined in the last round, and determining the action list of the current round based on the candidate actions meeting the cost requirement in the current round; in the first round, candidate actions meeting the cost requirement in the round are obtained from a plurality of actions to be executed; taking a father node action corresponding to the candidate action screened out by the last round of action screening operation in the plurality of actions to be executed as a target action to be executed in the current interaction state; and in the current interaction state, executing the target task in response to the target action.

The invention relates to a task processing method and system for a body-equipped robot based on context self-adaption. The method comprises the following steps: receiving user instructions, and classifying the user instructions to obtain instruction types; under the condition that the instruction type is the fuzzy instruction type, outputting an intention guiding suggestion for the user instruction; the intention guidance suggestion is generated based on instruction context information corresponding to the user instruction, and the intention guidance suggestion is used for guiding the user to define a task intention; based on the intention guiding suggestion, reconstructing the instruction context information to obtain reconstructed context information; receiving a feedback instruction of the user for the intention guidance suggestion, and performing semantic analysis on the feedback instruction based on the reconstructed context information to obtain a feedback task intention; and under the condition that the feedback task intention indicates execution of the target task, a target action sequence is generated based on the target task, and the robot with the body is controlled to execute the target action sequence. By adopting the method, the task processing efficiency can be improved.

The invention discloses a generation method, model and device and electronic equipment. The method comprises the steps that action description information input by a user is obtained; determining text features based on the action description information; processing the text features, and gradually obtaining an action feature set with multiple time sequence lengths according to a pre-trained codebook; the action feature set of the time sequence length comprises a plurality of action features with different time sequences; the codebook is used for storing a mapping relation between a plurality of discrete code values and a plurality of action characteristics; and processing the action feature set with the multiple time sequence lengths to obtain a target action sequence.

The invention provides an integration test method and device, equipment, a medium and a product, and relates to the technical field of automatic testing, and the specific technical scheme is as follows: obtaining an action library corresponding to a test object of an integration test service, and screening a target action from test actions in the interface test dimension, the file test dimension, the database test dimension and the script test dimension of the action library, and generating a test case based on the target action. And performing collection processing according to the test case and a preset case to obtain a test task, and running the integrated test task based on the execution dependency relationship to obtain an integrated test result of the test object. According to the method, on one hand, the action library comprises the test actions under multiple test dimensions, and the test applicability is wider; and on the other hand, based on the execution dependency relationship, the integration test task is automatically operated, the integration test result of the test object is obtained, manual intervention on triggering and sequence control between cases is not needed, automatic test is achieved, and the test efficiency is improved.

The present application discloses an action generation method and related devices, an electronic device, and a storage medium. Among them, the action generation method includes: obtaining first feature representations respectively characterizing a plurality of individuals in a plurality of action frames, and obtaining second feature representations respectively characterizing the plurality of individuals with respect to a target action category; performing relationship modeling based on the first feature representations and the second feature representations to obtain fused feature representations of each individual in each action frame; wherein the type of relationship modeling is related to the first total number of the plurality of individuals; performing action mapping based on the fused feature representations to obtain an action sequence of the plurality of individuals with respect to the target action category; wherein the action sequence includes a plurality of action frames, and each action frame includes action representations of each individual. The above solution can be compatible with two application scenarios of a single individual and multiple individuals on the premise of improving the action generation efficiency.

The embodiments of the present disclosure disclose a method and apparatus for training a model. A specific implementation of the method includes: obtaining a sample state transition trajectory data set; performing the following training steps: selecting a target state value from the sample state transition trajectory data, and using the target state value, performing the following processing steps: inputting the target state value into the initial action selection model to obtain a target action value; inputting the target state value and the target action value into the environment model to obtain a simulation subsequent state value and a simulation reward value; using the simulation subsequent state value as the target state value, repeatedly performing the processing steps until the preset processing conditions are met, and obtaining simulation state transition trajectory data; forming a training sample set from the sample single-step state transition data and the simulation single-step state transition data; and using the training sample set to train the initial action selection model to obtain a trained action selection model. This implementation can train an action selection model with stronger generalization capabilities.

This disclosure provides a method and an apparatus for programming. The method includes: identifying a first target profile from a set of profiles according to information related to a target user. The set of profiles is obtained according to information related to users, and each profile in the set of profiles corresponds to a category of the information related to the users. The method further includes determining one or more candidate models from a model database according to the first target profile. The models in the model database are trained models obtained through subsets of a training dataset. The models are used for programming tasks. The subsets are obtained by dividing the training dataset based on the set of profiles. The method further includes obtaining a target action from one or more candidate actions generated by applying the one or more candidate models to context.

The invention provides a model training method and device, a virtual object control method and device, equipment and a storage medium, and relates to the technical field of data processing. The method comprises the steps of obtaining a target instruction sent by a first intelligent virtual object in a first camp; determining a target action of a second intelligent virtual object in the first camp by adopting a preset decision model according to the target instruction; controlling the second intelligent virtual object to execute the target action to obtain an action execution result of the second intelligent virtual object; and training the preset decision model according to the action execution result to obtain a target decision model. Compared with the prior art, the problem that man-machine cooperation is not good enough is avoided.

A game action generation method and apparatus, a storage medium, and an electronic device. The method comprises: displaying an action editing interface in a graphical user interface provided when a game program is run (S210); on the basis of an editing operation in the action editing interface, determining from among a plurality of preset actions provided by the game program at least two actions to be spliced (S220); acquiring playback parameters set for the actions to be spliced (S230); and generating target action data on the basis of the actions to be spliced and the playback parameters corresponding to the actions to be spliced (S240). A user can conveniently edit a custom action, and the labor and time costs of game action making are reduced.

The present invention discloses a method and device for positioning and interacting UI elements based on keyword driving, comprising: obtaining a written test case; using a keyword syntax matching algorithm to obtain the parent element where the target element is located, the child element where the target element is located, and the characteristic value of the target element in the test case, and obtaining the characteristic value of the target action in the test case; using a positioning algorithm to obtain the parent element XPATH that matches the characteristic value of the parent element, and to obtain the child element XPATH that matches the characteristic value of the child element and the characteristic value of the target element; splicing the child element XPATH to the right side of the parent element XPATH to obtain the target element XPATH; based on the characteristic value of the target element XPATH and the target action, executing the corresponding interactive action. The present invention can not only reduce a large amount of manpower input, but also solve the redundant manual testing workload, and greatly reduce the maintenance cost.

The disclosure provides a large model-based data generation method and device, equipment and a storage medium, relating to the technical field of computers, especially the technical fields of large models, data processing, artificial intelligence and the like. The specific implementation scheme is: inputting target prompt information to a generation large model, so that the generation large model generates first question information based on expert knowledge and a target action sequence in the target prompt information; the target action sequence is used to constrain the execution path of the generation large model; the first question information includes text data, and the text data is a clue required to answer the first question information; the first question information is verified based on an evaluation large model to obtain a difficulty level; the difficulty level is used to represent the difficulty of the target large model in answering the first question information; and in the case that the difficulty level is a target level, a training sample of the target large model is constructed based on the first question information.

The present application discloses a data search method, apparatus, electronic device and storage medium, relating to the technical field of data processing. The method includes: when receiving a data query request, obtaining the data corresponding to the target module based on the trigger point and the received three-layer model data; obtaining an array of target changed attribute names based on the data corresponding to the target module and the enumeration index; obtaining the target action data based on the data corresponding to the target module and the enumeration index. Therefore, the present application only needs to search for keywords once to directly locate all information, having the advantages of saving development labor costs and improving the development experience of developers.