52 results about "Simulated reality" patented technology

The invention relates to a method for obtaining a hint injection attack test case for a large language model, which combines a conditional variation auto-encoder cVAE and a Markov chain, and obtains a large language model test case through data generation and context modeling, gradual exposure of malicious instructions and simulation of multiple rounds of dialogue attacks in reality. Realizing multiple rounds of dialogue attacks on the large language model, and challenging the defense capability of the large language model; the design scheme introduces a concealment technology, role play attack, state transition and other technologies, improves the complexity and concealment of attacks, has the core advantages of automation, higher concealment, wide coverage, batch testing and the like, can evaluate the security defense capability of the large language model more truly and comprehensively, and improves the security defense capability of the large language model. The defects of an existing defense mechanism are found, and research on multi-round prompt injection attacks and improvement of a security defense mechanism are promoted.

In various examples, a language model may be trained and used as part of an interface for a simulation system. For instance, user inputs may be applied to the language model and the language model may be trained to generate code, make API calls, or perform any other operations to interact with and / or control various aspects of the simulation. In some examples, the language model may generate code for, among other things, creating and / or customizing a virtual environment associated with the simulation. For instance, the generated code may include, but is not limited to, code for rendering the virtual environment, code for rendering and simulating behaviors of virtual agents (e.g., pedestrians, vehicles, animals, etc.) and / or any other objects (e.g., road signs, buildings, trees, etc.) within the virtual environment, code for recreating and simulating real-world events from recorded sensor data, etc.

The invention provides a modular and multifunctional karst collapse physical simulation experiment platform. Comprising a rigid frame, a bearing device, a moving device, a loading device, a lifting device, a pile foundation construction dynamic load simulation device, a ground traveling dynamic load simulation device, a rainfall simulation device, a computer vision monitoring device, an underground water pumping and draining simulation device and an underground space excavation disturbance simulation device. The moving device is arranged on the bearing device, the lifting device, the loading device and the pile foundation construction dynamic load simulation device are arranged on the moving device, the travelling crane dynamic load simulation device is connected with a traction device of the moving trolley, and the camera shooting monitoring device and the rainfall simulation device are arranged on the rigid frame through a support. According to the method, key disaster-causing conditions in the urban environment are fully considered, the working condition scene coverage is wide, and the applicability is high; modularized design is adopted as a whole, all the modules can be flexibly loaded according to requirements, the real environment can be simulated more truly, and the catastrophe evolution process of surface collapse can be reproduced.

The invention provides a porcelain insulator quality evaluation system based on multi-physics field coupling analysis, and relates to the technical field of rod-column porcelain insulator data processing.The porcelain insulator quality evaluation system adopts an electricity-heat-force multi-physics field joint simulation model to comprehensively reflect the heat accumulation and stress deformation effect caused by electric field concentration; an electric field concentration coefficient, a temperature rise trend coefficient and a stress gradient coefficient at a sharp point can be accurately captured, and degradation path prediction in a real operation environment is formed; according to the technology, the thread coupling abnormal index F alpha i of the ith rod-type column porcelain insulator is constructed, whether the thread structure has the risk of partial discharge, flashover or breakdown or not can be dynamically evaluated, and by introducing equivalent dielectric layer simulation of various environmental media, the simulation of the influence of various pollutants, attachments and climates on electric field concentration in reality is promoted; quantitative evaluation of the corona discharge risk of the sharp point is realized, and the reliability of the structural design of the insulator is promoted and improved.

The invention provides an autonomous parking method and system based on end-to-end reinforcement learning, a medium and equipment. The method comprises the following steps: acquiring an image of a real scene, laser point cloud information and inertial measurement unit data; constructing a scene model according to the real scene image, the laser point cloud information and the inertial measurement unit data; under the scene model, inputting the rendered look-around image into a preset end-to-end reinforcement learning network, setting a preset auxiliary task, guiding the preset end-to-end reinforcement learning network to train by adopting a progressive guide reduction strategy and a soft prediction collision penalty mechanism, and determining a trained end-to-end reinforcement learning network model; and inputting the real look-around image information of the target vehicle into the trained end-to-end reinforcement learning network model, and determining a parking prediction trajectory of the target vehicle. According to the method and the device, an end-to-end reinforcement learning mode is adopted, error accumulation is reduced, the parking precision and the parking capability upper limit are improved, and reality-simulation-reality scene seamless migration is realized.

The invention discloses a chat system with functions of user psychological exclusive virtual interaction and emotional value providing, and belongs to the field of virtual interaction. Comprising a registration and login module, a psychological analysis module, a personality generation module, an interaction matching module, an interaction cultivation module, an exclusive management module, a stage management module, a relation authentication module, an account logout module and an offline interaction module. According to the method, the interaction between the user and the most matched AI robot can be ensured, the real and stable social experience can be provided, the chapter interaction feeling is reduced, the user belonging feeling and the interaction viscidity are enhanced, and separation and new connection in real social contact can be simulated more truly.

The invention relates to the technical field of data expansion, in particular to a malicious prompt data set expansion method based on Mongolian. According to the method, high-frequency roots and affixes in a Mongolian basic malicious prompt corpus and a universal corpus are extracted, morphological analysis is combined, targeted malicious prompt samples can be constructed, the diversity and authenticity of a Mongolian safety evaluation data set are enhanced, candidate malicious prompt samples are optimized by adopting a genetic algorithm, and the safety evaluation accuracy of the Mongolian safety evaluation data set is improved. According to the method, the expansion sample is enabled to better conform to syntax and semantic rules of the Mongolian, effectiveness of the attack sample in the Mongolian scene is ensured, the expansion data set is enhanced by using the antagonism generation strategy, robustness and antagonism of the data set are improved, attack behaviors possibly occurring in a real scene can be simulated, and the robustness and the antagonism of the data set are improved. Therefore, an effective sample generation tool is provided for safety evaluation in a Mongolian scene, and the anti-attack capability of the Mongolian model is improved.

The invention relates to the technical field of automatic driving, and discloses a mixed reality automatic driving algorithm simulation method for pedestrian safety, and the method comprises the steps: S10, analyzing a road data file based on a simulation engine to obtain road data, and constructing a simulation scene through the road data; s20, placing the automatic driving vehicle model and a virtual substitute in the simulation scene, and obtaining dynamic behavior simulation data based on the simulation engine; s30, embedding a to-be-tested automatic driving algorithm into the simulation engine, and simulating a traffic flow in a real road condition scene; s40, the behavior information of a real pedestrian is mapped to the virtual substitute through a sensor, and a simulation data set generated by the interaction behavior of the virtual substitute and the traffic flow is acquired; and S50, storing the simulation data set through a database, analyzing and processing the simulation data set by adopting a data analysis algorithm, and generating a simulation test result based on the analyzed and processed simulation data set.

An innovative augmented reality / virtual reality (AR / VR) crossbow system that includes a bow frame assembly, a stock, a sight assembly, a cocking assembly, a resistance assembly and an electronic device assembly with a mobile device. These components are configured to structurally interact to provide the user with an experience of loading, aiming and firing a virtual crossbow arrow that closely simulates real-life archery and crossbow scenarios without firing an actual crossbow arrow. Due to its unique components and their layout, the AR / VR crossbow system is not configured to fire real arrows. An interactive crossbow application is loaded on the mobile device and simulate real life archery scenarios such as hunting scenarios or target competition scenarios. The mobile device also includes a gyroscope to analyze the orientation of the crossbow frame assembly and a microphone to determine when a simulated arrow is fired within the crossbow application.

In one implementation, a non-transitory computer-readable storage medium stores program instructions computer-executable on one or more processors of an electronic device to perform operations. The operations include presenting, on a display of the electronic device, a simulated reality (SR) environment at a first immersion level, where the first immersion level is associated with a first location of a reality boundary. Using a rotatable input device of the electronic device, input is received representing a request to change the first immersion level to a second immersion level. In accordance with receiving the input, the SR environment is presented at the second immersion level, where the second immersion level is associated with a second location of the reality boundary.

The present application relates to the technical field of graphic simulation system, specifically to a hydraulic fracturing process simulation and optimization system based on digital twin technology; containing real-time input, probabilistic simulation, adaptive assimilation and risk analysis and control unit; the system generates prior fracture probability cloud by combining agent model with real-time data; the core is to generate posterior fracture probability cloud by assimilating monitoring data, and to calculate simulation-realistic credibility index and probabilistic risk boundary accordingly; the system generates adaptive control signals according to the index and feeds back to the assimilation unit; the present application overcomes the false accuracy of traditional deterministic simulation, and realizes the quantification and intuitive display of fracture uncertainty.

The utility model discloses a limited space operation and rescue drilling platform, which belongs to the technical field of drilling platforms, and comprises a platform main body and a simulation pipeline assembly arranged at the bottom of the platform main body, the top of the platform main body is provided with a main well mouth and side well mouths positioned on two sides of the main well mouth; the simulation pipeline assembly comprises a main well pipe, a side well pipe, a transverse well pipe and a maintenance bin, the main well pipe, the side well pipe and the transverse well pipe are the same in structure and each comprise two same semicircular pipes, and the semicircular pipe located on the front face is a transparent pipe. Through the arrangement of the platform main body and the simulation pipeline assembly, real ground and underground pipelines are simulated, multi-person drilling operation can be carried out, and the use is flexible; through the arrangement of the main well mouth, the main well pipe and the maintenance bin, the situation of entering the maintenance bin for drilling can be simulated, and through the arrangement of the side well pipe, when drilling rescue work is simulated, ventilation from the side well pipe and rescue from the main well pipe can be simulated.

The utility model relates to the technical field of VR (virtual reality) equipment, and discloses a bracelet interaction device for dodging ball games, which comprises a VR head-mounted display and a bracelet, a central control unit is arranged in the VR head-mounted display, four infrared sensors which are distributed in a rectangular shape are fixedly mounted on the front side of the VR head-mounted display, and the central control unit is connected with the VR head-mounted display. The four infrared sensors are all connected with the central control unit, a first Bluetooth connection unit is arranged in the VR head-mounted display, and the first Bluetooth connection unit is connected with the central control unit. According to the utility model, the positioning device is worn on the wrist through the hand strap, the purpose of hitting a virtual ball in the eyes of a user on the VR head-mounted display to petals by simulating up-down hand swinging in reality and through the hand swinging process is achieved, ball swinging or ball throwing correspondingly achieves fitness, and the purpose of exercise can be achieved without going out; an infrared emission unit in the positioning device emits a signal, a corresponding infrared sensor receives the signal, and a dodging ball is hit in real time, so that the positioning and tracking effects are achieved.

The invention belongs to the field of federated learning and image classification, and particularly relates to an image classification algorithm based on combination of self-adaptive comparative learning and federated learning. According to the algorithm, local data of participants are distributed through a data division strategy, and non-uniformity of distributed data in a real scene is simulated; then, the heterogeneity of the data is controlled through Dirichlet distribution, and the flexibility of data division is further enhanced; each participant constructs a local convolutional neural network according to the distributed data, a dynamic temperature adjustment mechanism is introduced in the training process, temperature parameters are dynamically adjusted according to gradient variance and personalized features of clients, sample similarity calculation in the learning process is optimized and compared, and the adaptability of the model to different client data is improved. In addition, efficient and accurate image classification is realized by comparing a learning loss function (such as InfoNCE loss) with knowledge distillation of a historical model pool.

An application programming interface, API, and its implementing functions provide software code that can be incorporated into a simulated reality application and that determine a total calibration gain that is to be applied during run time of the application, to the audio signal of a particular digital audio asset that has been selected by the author of the simulated reality application. The total calibration gain includes at least a loudness correction gain that is determined based on a user-specified sound level parameter indicated through the API by the author. Other aspects are also described and claimed.

Computer-readable medium containing instructions that configure a computer to: present an application programming interface (API), wherein the API includes a user-specified sound level parameter; receive, via the API, one or more values ​​for the user-specified sound level parameter, which are assigned to a digital audio asset;and incorporating code into a simulated reality application that is currently being assembled, wherein the code determines a loudness correction gain for the digital audio asset by determining a difference between i) the one or more values ​​for the user-specified sound level parameter and ii) a parameter of the playback hardware device that represents a known or predefined sound level that can be produced by a playback hardware device on which the simulated reality application is running, wherein the loudness correction gain compensates for the difference, and the loudness correction gain is then to be applied to the digital audio asset during the runtime of the simulated reality application.

The present invention provides a modular, multifunctional karst collapse physical simulation experimental platform, including a rigid frame, a bearing device, a mobile device, a loading device, a lifting device, a pile foundation construction dynamic load simulation device, a ground vehicle dynamic load simulation device, a rainfall simulation device, a computer vision monitoring device, a groundwater pumping simulation device, and an underground space excavation disturbance simulation device. The mobile device is arranged on the bearing device, the lifting device, the loading device, and the pile foundation construction dynamic load simulation device are arranged on the mobile device, the vehicle dynamic load simulation device is connected to the traction device of the mobile vehicle, and the camera monitoring device and the rainfall simulation device are arranged on the rigid frame through a bracket. The present invention fully considers the key disaster-causing conditions in the urban environment, covers a wide range of working scenarios, and has strong applicability; the overall modular design is adopted, and each module can be flexibly loaded according to demand, which can more realistically simulate the real environment and reproduce the catastrophic evolution process of ground collapse.

The present invention provides a multi-parameter testing system for a water pump assembly, relating to the technical field of water pump testing. This multi-parameter testing system for a water pump assembly includes a flow detection module, a suction lift detection module, a head detection module, a torque detection module, an electrical parameter detection module, and a master control module. The flow detection module includes an inlet flowmeter and an outlet flowmeter, each of which has an inlet flange and an outlet flange at its water inlet and outlet. The suction lift detection module and the head detection module of the present invention have a simple structure and low installation cost, and can simulate real-world pipeline resistance, making the data closer to reality and improving detection accuracy.

The invention relates to the technical field of graphic simulation systems, in particular to a hydraulic fracturing process simulation and optimization system based on a digital twin technology. The system comprises a real-time input unit, a probabilistic simulation unit, a self-adaptive assimilation unit and a risk analysis and control unit. The system generates a prior crack probability cloud by acquiring real-time data and combining with the proxy model; the core of the method is that monitoring data assimilation is utilized to generate a posterior crack probability cloud, and a simulation-reality credibility index and a probabilistic risk boundary are solved according to the posterior crack probability cloud; the system generates an adaptive control signal according to the index and feeds back the adaptive control signal to the assimilation unit; according to the method, false accuracy of traditional deterministic simulation is overcome, and quantification and visual display of crack uncertainty are realized.

An application programming interface, API, and its implementing functions provide software code that can be incorporated into a simulated reality application and that determine a total calibration gain that is to be applied during run time of the application, to the audio signal of a particular digital audio asset that has been selected by the author of the simulated reality application. The total calibration gain includes at least a loudness correction gain that is determined based on a user-specified sound level parameter indicated through the API by the author. Other aspects are also described and claimed.

The invention relates to a social network subject simulation system and method based on a large language model, and belongs to the technical field of cognitive social simulation. In order to fill up evolvable psychological and action mechanisms existing in the cognitive social simulation field, the social network subject simulation system and method are provided, virtual subjects with dynamic memory, reflection and social behavior capabilities are constructed, and interaction and information spreading of multiple subjects in a virtual social environment are realized. Therefore, the public opinion forming and spreading process in a real social network is simulated.

The application provides a defect detection and positioning method based on regional anomaly generation and multi-level reverse distillation, and relates to the technical field of industrial defect detection. The new synthetic anomaly method provided by the application simulates the randomness of the abnormal area in the real scene by generating pseudo-anomaly images of different sizes and shapes, provides rich training data for the model, proposes a single-scale repair strategy including convolutional transformer processing and C-CBAM attention mechanism, enables the student network to efficiently capture global and local features, and realizes basic feature repair, further proposes a multi-scale repair strategy, removes redundant features by using a similarity redundancy removal module to enhance cross-layer feature integration, retains key information, and removes redundant features, simultaneously uses a multi-scale patch block fusion module to strengthen local fine-grained anomaly feature reconstruction, and finally realizes image anomaly detection and positioning through loss calculation and anomaly scoring. The method has good defect positioning effect and can meet the needs of industrial production.

In one implementation, a non-transitory computer-readable storage medium stores program instructions computer-executable on one or more processors of an electronic device to perform operations. The operations include presenting, on a display of the electronic device, a simulated reality (SR) environment at a first immersion level, where the first immersion level is associated with a first location of a reality boundary. Using a rotatable input device of the electronic device, input is received representing a request to change the first immersion level to a second immersion level. In accordance with receiving the input, the SR environment is presented at the second immersion level, where the second immersion level is associated with a second location of the reality boundary.

A method is provided for evaluation of data, where it was surprisingly found out that raw data can be simulated using hardware devices. The method actively operates and control hardware components to establish a network that provides hardware components that simulate the relationships between real world concepts. The method is able to automatically perform technical processes such that firstly no human ado is required and secondly the resulting data is not prone to errors. The method suggests iterations on evolving data sets and hence a bias is excluded or at least minimized in each iteration. Also provided is a respectively arranged system along with a computer program product and a computer-readable medium.

The invention belongs to the field of intelligent unmanned cluster control, and particularly relates to a bipartite cooperative control method of an intelligent unmanned cluster system, which is used for cooperative control in an unmanned aerial vehicle cluster system with two state updating modes. The communication weight is a positive value when the intra-group unmanned aerial vehicles communicate, the communication weight is a negative value when the inter-group unmanned aerial vehicles communicate, and an unmanned aerial vehicle adjacency matrix is established according to the communication weight; one group of unmanned aerial vehicles updates the state based on behavior dynamics, one group of unmanned aerial vehicles updates the state based on a protocol, and optimal state updating is obtained by solving Nash equilibrium according to an unmanned aerial vehicle state updating strategy and an adjacent matrix; and judging whether the intelligent unmanned cluster system converges or not and determining a steady-state position, and performing unmanned aerial vehicle control according to the obtained steady-state position. According to the method, the dependence of a traditional method on a system dynamics model is overcome, and real reality scenes with dynamics mixing performance such as unmanned driving and human driving in real life are simulated.

Methods and systems are provided for digitally replicating a real-world infrastructure to simulate the behavior of a real-world object, the methods and systems supporting the ability to extend a wireless network by providing services implemented in the network, and supporting optimization of network usage and digital representation accuracy. In one embodiment, the method includes instantiating one or more digital representation (D-Rep) instances corresponding to a digital replica of the real world object included in the real world infrastructure to form a digital infrastructure (D-Inf), and generating one or more digital representation (D-Rep) instances corresponding to a digital replica of the real world object included in the real world infrastructure. The method further includes providing a communication plane to provide a service to the D-Inf. The method further includes connecting to the real world object using the communication plane to collect real world data defining a state of the real world object, and associating the real world data with the D-Rep instance; the behavior of the real-world object is simulated by the D-Rep instance using a simulator that can be accessed by the D-Inf instance using the communication plane.

A first world is one of a real world and a virtual world simulating the real world, and a second world is another of them. A first image is captured by a first camera in the first world, and a second image is captured by a second camera in the second world. A visual positioning system executes common processing that generates a scene graph representing a positional relationship between objects included in the image and extracts a feature amount of the scene graph. The visual positioning system performs matching between a first feature amount extracted by the common processing on the first image and a second feature amount extracted by the common processing on the second image, and then associates the first camera position in the first world and the second camera position in the second world with each other based on a result of the matching.

The invention provides a defect detection and positioning method based on regional anomaly generation and multi-level reverse distillation, and relates to the technical field of industrial defect detection. According to the new anomaly synthesis method provided by the invention, the randomness of an abnormal region in a real scene is simulated by generating pseudo-anomaly images with different sizes and shapes, and rich training data is provided for a model; a single-scale repair strategy is proposed, the single-scale repair strategy comprises convolution transformer processing and a C-CBAM attention mechanism, so that the student network efficiently captures global and local features, and basic feature repair is realized; a multi-scale repair strategy is also provided, cross-layer feature integration is enhanced by means of a similarity redundancy removal module, key information is retained, redundant features are removed, and local fine-grained abnormal feature reconstruction is enhanced by means of a multi-scale patch block fusion module; and finally, through loss calculation and anomaly scoring, realizing anomaly detection and positioning of the image. The method has a good defect positioning effect and can meet the requirements of industrial production.

In some embodiments, an advanced packaging design and cost estimation system is provided. It may include a user interface incorporating a 3D canvas area where users can drag and drop packaging objects to model real world chiplets and packaging architecture scenarios and receive feedback on the predicted costs.

The embodiment of the invention discloses an action simulation method and a related device, on one hand, a target preparation action can be executed before the target action is executed, so that the authenticity of a game is improved by simulating the preparation action required for executing the target action in a real scene; the situation that the action authenticity is reduced due to direct execution of the action indicated by the instruction is avoided; and on the other hand, in the process of executing the target preparation action, the lens position can change along with the change of the posture of the game object, so that the visual change caused by the posture change is simulated through the change of the game scene area displayed by the game interface, the sense of reality of the target preparation action is further improved from the visual dimension, and the user experience is improved. And more real game experience is brought to players.