11 results about "Web navigation" patented technology

Aspects of the subject disclosure may include, for example, a device, including: a processing system including a processor; and a memory that stores executable instructions that, when executed by the processing system, facilitate performance of operations of identifying a public land mobile network (PLMN) identifier associated with a communications network; determining a tracking area code (TAC) range of a plurality of cell sites in a radio access network (RAN) associated with the PLMN identifier; identifying transport network circuits of a transport network connecting the plurality of cell sites with a core network of the communications network; and displaying a user interface comprising a graphical representation of the core network, the transport network and the RAN. Other embodiments are disclosed.

Machine learning models for autonomously navigating the web and / or apps are provided that receive (i) visual input that includes one or more images representing the current, and optionally one or more prior, visual states of the website or app being navigated, and (ii) textual input that represents a user request to be implemented by the model, the current HTML or other code defining the website or app, and optionally additional text like prior model outputs that represent prior steps in the navigation of the website or app by the model. The visual input and textual input are transformed into respective sets of visual and textual tokens, and this multi-modal token set is applied to an encoder. The output of the encoder is applied to a decoder to generate textual output that represents an action to take, relative to the code defining the website or app, to pursue the user request.

A method includes obtaining an indication of a user input and determining a targeted element class associated with an application having a plurality of application elements based on the indication of the user input. The method also includes identifying a subset of the plurality of application elements based on the targeted element class and synthesizing a plurality of speech segments respectively associated with the subset of the plurality of application elements.

According to the mobile robot navigation method based on deep reinforcement learning provided by the invention, guidance is realized in the early stage of deep reinforcement learning strategy training to accelerate convergence, so that the mobile robot can complete safe and efficient autonomous navigation only by depending on an airborne sensor in a complex dynamic environment. The method comprises the following steps: S1, collecting data online: S11, training a navigation model in a simulation environment; defining a state space of the mobile robot in the simulation environment, and forming a state vector; s12, sending the state vector into a graph attention network to obtain a state representation; s13, the strategy network takes the state representation as an input and outputs a control action; reference actions of the line velocity and the angular velocity are calculated through an artificial potential field method; s2, in the training stage of the navigation model, training a commentator network, a strategy network and a graph attention network; and S3, after the navigation model is trained, deploying the navigation model to the mobile robot.

The present disclosure is generally directed to methods, apparatus, and computer-readable media (transitory and non-transitory) for learning to automatically navigate interactive web documents and / or websites. More particularly, various approaches are presented for training various deep Q network (DQN) agents to perform various tasks associated with reinforcement learning, including hierarchical reinforcement learning, in challenging web navigation environments with sparse rewards and large state and action spaces. These agents include a web navigation agent that can use learned value function(s) to automatically navigate through interactive web documents, as well as a training agent, referred to herein as a “meta-trainer,” that can be trained to generate synthetic training examples. Some approaches described herein may be implemented when expert demonstrations are available. Other approaches described herein may be implemented when expert demonstrations are not available. In either case, dense, potential-based rewards may be used to augment the training.

The invention discloses a voxel and sparse convolution-based local scene attention network navigation method and system for an indoor service robot, and the method comprises the following steps: S1, collecting original point cloud data of an indoor environment in real time through a three-dimensional laser radar carried by the robot, obtaining the information of an object in the environment, and forming an original point cloud data set; s2, dividing a three-dimensional space corresponding to the original point cloud data collected in the step S1 into regular 3D voxel grids, performing feature aggregation on all the point cloud data in each voxel grid, and calculating a voxel feature mean value VFM and a voxel feature covariance VFC; s3, constructing a sparse convolution U-Net network, and performing multi-scale feature extraction on the feature data after voxelization in the step S2; s4, dynamically generating a local region of interest based on the features extracted in the step S3 in combination with the real-time motion state of the robot; s5, performing attention weight calculation and feature weighting on the local region of interest generated in the step S4; and S6, generating a control instruction according to the features obtained in the step S5.

The invention discloses a network navigation optimal state estimation method and system, and relates to the technical field of network navigation. The method is provided for overcoming the defects of existing research in the aspects of complex dynamic system modeling, multi-source information fusion processing and the like. The technical key points are as follows: network navigation is an important theoretical framework for researching a multi-agent system collaborative navigation positioning problem, system dynamics and measurement behaviors are represented mainly by means of a network graph theory, and the optimal state estimation of the system is obtained by combining a network topology structure and fusing processing measurement information. According to the method, a super network model is provided to describe a coupling relationship among multiple states of an intelligent agent, system dynamics constraint and measurement information are converted into a structured Fischer information matrix and an information vector, and neat corresponding relationships exist between the structured Fischer information matrix and the information vector and a network topology; secondly, four optimal state estimation algorithms are deduced based on a Cramer-Rao lower bound inequality, the four optimal state estimation algorithms are a centralized algorithm, a decentralized algorithm and a recursive and hybrid algorithm with a distributed structure, and the algorithm processes can be visualized through a fusion rule of network information. Finally, numerical simulation analysis verifies the feasibility of the model and the algorithm provided by the invention.

Aspects of the subject disclosure may include, for example, a device, including: a processing system including a processor; and a memory that stores executable instructions that, when executed by the processing system, facilitate performance of operations of identifying a public land mobile network (PLMN) identifier associated with a communications network; determining a tracking area code (TAC) range of a plurality of cell sites in a radio access network (RAN) associated with the PLMN identifier; identifying transport network circuits of a transport network connecting the plurality of cell sites with a core network of the communications network; and displaying a user interface comprising a graphical representation of the core network, the transport network and the RAN. Other embodiments are disclosed.

The present invention relates to a technology for identifying a plurality of interactive elements on a web page when a user sets a goal, automatically generating an execution plan including at least one action to achieve the goal by identifying attributes of the identified interactive elements, and performing web navigation accordingly.

The present invention relates to a method and system for providing a web navigation service on the basis of an action agent , in which an AI model is trained with demonstration data in which actual web page manipulation is recorded, an action plan to be automatically performed by the trained AI is generated, and a web page is actually manipulated according to the generated plan to complete a requested task.

The present disclosure is generally directed to methods, apparatus, and computer-readable media (transitory and non-transitory) for leaming to automatically navigate interactive web documents and / or websites. More particularly, various approaches are presented for training various deep Q network (DQN) agents to perform various tasks associated with reinforcement learning, including hierarchical reinforcement learning, in challenging web navigation environments with sparse rewards and large state and action spaces. These agents include a web navigation agent that can use learned value function(s) to automatically navigate through interactive web documents, as well as a training agent, referred to herein as a "meta-trainer," that can be trained to generate synthetic training examples. Some approaches described herein may be implemented when expert demonstrations are available. Other approaches described herein may be implemented when expert demonstrations are not available. In either case, dense, potential-based rewards may be used to augment the training.