System for identifying and relating elements of scene

Abstract

Disclosed is a system for retrieving semantic data regarding a space, contextualising an environment and making inferences in real time, using artificial intelligence, and which, by means of one or more sensors, artificial intelligence hardware units and a cloud viewing and management platform, allows semantic data to be gathered and the relational dynamics of the space to be understood, for the meaning of a context.

Description

[0001] Title: Method and System for understanding context and obtaining real-time inferences about future events using Artificial Intelligence order

[0002] Method and System for understanding context and obtaining inferences in real time about future events using edge Artificial Intelligence

[0003] DESCRIPTIVE MEMORY

[0004] This invention patent application is directed to a system and method that provides a technique for the recovery of semantic data of spaces, contextualization of environments, and development of inferences (future states) in real time, through the use of Edge Artificial Intelligence, which through data from one or more video, audio, point cloud, temperature, or other type of sensors, Edge Artificial Intelligence (Edge AI) hardware units and a Platform for visualization and management in the Cloud (Cloud), allow the increase of semantic data and understanding of the relational dynamics for the significance of a context, a representation of the temporal state space enriched with complex information, using fewer computing and energy resources, for low-latency decision-making locally, at the place where the data originates. PREVIOUS ART

[0005] Edge AI, or Edge Artificial Intelligence, refers to the implementation of artificial intelligence (AI) algorithms and models directly on edge devices, such as smartphones, IoT (Internet of Things) devices, cameras, and other embedded systems, rather than relying on a centralized cloud server for predictive model execution or processing. This brings processing power and intelligence closer to the data source, reducing latency, improving privacy, and enabling real-time decision-making.

[0006] Key features and considerations of Edge AI include:

[0007] Low latency: By processing data locally on the edge device, Edge AI reduces the time it takes for information to travel back and forth to a centralized server. This is crucial for applications that require real-time responses, such as autonomous vehicles, surveillance, automation, and medical solutions.

[0008] Privacy and Surveillance: Edge AI can improve privacy by keeping sensitive data on the edge device, minimizing the need to transmit data to the cloud. This is particularly important for applications such as video surveillance, security, and healthcare, where data privacy is a major concern.

[0009] Bandwidth efficiency: Edge data processing reduces the need to send large volumes of raw data to the cloud, leading to more efficient use of bandwidth. This is especially beneficial in scenarios with limited network connectivity or high data transfer costs.

[0010] Offline Functionality: Edge AI enables devices to perform AI tasks even when they are not connected to the internet. This is useful in situations where a reliable internet connection may not be available, such as for autonomous vehicle navigation.

[0011] Energy efficiency: Local processing can reduce the need for constant communication with the cloud, saving power and increasing the battery life of edge devices.

[0012] Scalability: Edge AI systems can be designed to scale horizontally by distributing intelligence across a network of edge devices. This makes it easier to handle increasing workloads without relying solely on a centralized infrastructure.

[0013] Edge AI applications span diverse industries , including mining, manufacturing, retail, smart cities , construction , and more . Examples include real-time speech and image recognition on smartphones and edge devices , predictive maintenance in industrial equipment , and monitoring and analyzing IoT sensor data .

[0014] Edge AI is a rapidly evolving field, with continuous advancements in hardware design, machine learning algorithms, and deployment strategies. As the technology continues to advance, we can expect further integration of AI capabilities into edge devices for a wide range of applications.

[0015] In summary, the state of the art of edge artificial intelligence technology is rapidly evolving, offering innovative and efficient solutions for a wide range of applications across various industries. Patent publication US20220019889 describes an Edge AI method and device. The device includes one or more first neural network layers and one or more second neural network layers, the device is configured to perform operation based on first weight information of the one or more first neural network layers received from and trained off the device, and the device is configured to perform operation based on second weight information of the one or more second neural network layers, based on training the second weight information by the device.

[0016] For its part, patent publication IN202241056847 describes an intelligent accident detection system based on artificial intelligence and computer vision to prevent accidents through video surveillance using machine learning and deep learning algorithms. Real-world traffic surveillance information must be reviewed periodically, which requires regular human supervision that is prone to errors and is time-consuming. As a result, the deep learning-based system makes it possible to quickly and accurately determine the exact location of car accidents. The system makes use of a sequence-to-sequence long short-term memory autoencoder and a spatio-temporal autoencoder, respectively, to model the spatial and temporal representations of a video. The method employs a one-dimensional method to categorize elements.The approach produces qualitatively and statistically intriguing results when applied to real-world video traffic surveillance data sets.

[0017] Patent publication WO / 2022 / 005646 describes a machine perception computing system utilizing video / image sensors in a service / edge computing system architecture, which is deployed at a physical location and receives an input from one or more image / video sensing mechanisms. The edge computing system executes artificial intelligence image / video processing modules on the received image / video streams and generates metrics by performing spatial analysis on the image / video streams. The metrics are provided to a multi-tenant service computing system where additional artificial intelligence (AI) modules are executed on the metrics to execute perceptual analysis. Client applications may then run on the output of the AI ​​modules in the multi-tenant service computing system.Patent publication US20220086052 relates to an intelligent management method and system based on edge computing and belongs to the field of edge computing technology. By adopting the method, the time required for data analysis and processing can be shortened, and thus the delay is smaller. In addition, in the event of network interruption and cloud downtime, it can ensure that edge computing can not only run local computing software offline in an internal network to implement local autonomy, but also can implement, through the local edge computing cluster running offline, the interconnection and intercommunication between machine devices to construct a complete intelligent network and management system.Meanwhile, it can save bandwidth resources for data transmission when networking and effectively alleviate the problems of data calculation and storage pressure, high traffic, difficulty in application development, and insufficient intelligence of an underlying machine device when using cloud computing.

[0018] Patent publication WO / 2017 / 007626 discloses an object detection method comprising identifying a context label from an entire image. The method also includes selecting a set of likely regions for detecting objects of interest in the image based on an identified context label. Another aspect of the patent discloses an apparatus for object detection having a memory and at least one processor coupled to the memory. The processor is configured to identify a context label from an entire image. The processor is also configured to select a set of likely regions for detecting objects of interest in the image based on an identified context label.

[0019] Another aspect describes a non-transitory computer-readable medium for object detection. The non-transitory computer-readable medium has non-transitory program code recorded thereon that, when executed by the processor(s), causes the processor(s) to perform operations of identifying a context label from an entire image. The program code also causes the processor to select a set of likely regions for detecting objects of interest in the image based on an identified context label. Another aspect describes an apparatus for object detection and includes means for identifying a context label from an entire image. The method also includes means for selecting a set of likely regions for detecting objects of interest in the image based on an identified context label.

[0020] Patent publication US20190318170 describes a workplace safety method for an industrial processing facility includes an image processing and artificial intelligence based workforce safety system that first receives image data from cameras observing work zones, including a first camera showing an individual in a first work zone. From the image data, the current location of the individual is determined. A current minimum requirement for Personal Protective Equipment (PPE) is determined based on the current location by referencing a database having work environments with hazardous conditions, including for the current location,the determination of a current work environment that has a current hazardous condition and the PPE required for the current hazardous condition . Image data is analyzed to identify the PPE currently being worn by the individual . When it is determined that the individual does not currently meet the PPE requirement by comparing the PPE they are currently wearing to the minimum PPE requirement , an alert is generated in response to the unsafe condition . The differences observed between the present invention and prior art documents would be given because the solution being searched for, Semantic data recovery of spaces, contextualization of environments and development of inferences (future states) , aims to obtain a spatiotemporal representation enriched with complex information, by acquiring video, audio, temperature datapoint cloud or other data from sensors feeding into edge Artificial Intelligence (Edge AI) hardware units, which by running predictive Machine Learning models develop inferences to obtain information about future spaces and contexts, and all managed from a Cloud Platform or from a local machine, where contextualized spaces, enriched scenarios, elements, objects, regions, and behaviors are recorded with relational dynamics, providing a more complex representation of the state space.

[0021] The data on the Platform coming from the EDGE AI or Artificial Intelligence Edge hardware layer, are intended to be analyzed, processed and presented to the user in graphical form with information on the main metrics involved in determining the future context, where you can view the amplified context with more complex information, interpretations, inferences, warnings and alerts that can be consulted in the form of indicators, control panels or dashboards and develop management on the information.

[0022] From the data received at the edge, the use of Predictive Models based on neural networks and calculations for the complex recognition of patterns in images, video, point cloud, sound and other types of data corresponding to the temporal space, which is directly related to the result of inferences to obtain an enriched representation of the state space, although publication WO / 2017 / 007626 mentions a method to identify a context label, it is not seen that it is associated with the generation of an enriched state space with additional complex information and that they have been executed in any way to obtain semantic data and understand relational dynamics in real time.

[0023] Another difference is that the system of the invention incorporates algorithms for the construction and dynamic management of regions with elements and relational dynamics between objects, which facilitates the implementation of hierarchical validations between elements of a region or between regions, which as a result allows the identification of a context and / or activity and the enrichment of the scene in real time with information on future stages of the environment.

[0024] Another difference is that the system of the invention incorporates a mechanism for the execution of predictive models and advanced mathematical calculations in specialized hardware units within the Edge AI unit, such as CPU, GPU, NPU and DSP, which allows obtaining inferences at very high speed and latency close to zero with low energy consumption.

[0025] Another difference is that the system of the invention incorporates the understanding of the scene or context through predictive models of edge Artificial Intelligence developing high-speed inferences, using a minimum number of frames, providing an enriched spatiotemporal representation for an unlimited number of spaces, regions and objects, which provide information on future states of the environment or space, making it easier for organizations, companies and end users to benefit from decisions with augmented information for the implementation of real-time actions on the current scenario or context, although publication US20190318170 mentions a workplace safety method for an industrial processing facility based on image processing and artificial intelligence for the detection of personal protective elements (PPE), it is not appreciated that it incorporates Edge AI features,Element identification using markers, dynamic generation of regions, validation of hierarchies (relationships between elements), probability calculation and platform for visualizing results and alerts.

[0026] BRIEF DESCRIPTION OF THE FIGURES

[0027] FIGURE 1: Corresponds to an illustrative diagram of the method and system that provides a technique for the recovery of semantic data of spaces, contextualization of environments, and development of inferences (future states) in real time through the use of Edge Artificial Intelligence (Edge AI). FIGURE 2: Corresponds to an illustrative diagram of the EDGE AI system and its modules, which allow the recovery of semantic data of spaces, contextualization of environments, and development of inferences (future states).

[0028] FIG.2

[0029] FIGURE 3: Corresponds to a diagram of the system software module called Model and Inference where a previously trained and quantized model outside the Edge AI unit is parameterized for the development of inferences and obtaining elements.

[0030]

[0031] FIGURE 4: Corresponds to a diagram of the system software module called Identification, responsible for identifying binary markers and selecting the best candidate for identifying an element.

[0032] FIG. 4

[0033] FIGURE 5: Corresponds to a diagram of the system software module called Positioning and Regions, which is responsible for managing regions and elements based on a routine to be developed for each data entry or each frame in the case of a video data source.

[0034] FIG. 5

[0035] FIGURE 6: Corresponds to a diagram of software modules of the system called Hierarchy and Context, and Logic and Probability, where the first is responsible for recognizing the hierarchy based on a match with the previously defined ones, validating dependent and independent elements and their relationships defined in the hierarchy, and the second is responsible for checking the probability of occurrence of the hierarchy event.

[0036] Fig. 6

[0037] FIGURE 7: Corresponds to a diagram of the complete operating framework of the method and system, from the acquisition of data from the source (500) to the deployment of results on the Cloud platform (300).

[0038] FIGURE 8: Corresponds to a diagram of the method with several explanatory figures of the different modes of use within the framework of the use of the method and system in the context of mining and industry, also describing the preferred mode 610.

[0039] FIGURE 9: A diagram depicting industrial workers on scaffolding, highlighting the modern industrial environment and the complexity of their elevated tasks. The drawing depicts what is displayed on the Edge AI unit's video output interface using the method and system. The video is obtained for calibration and adjustment of the method and system.

[0040] 710

[0041] FIGURE 10: Corresponds to a diagram of one of the views of the user interface, in which a list of alerts is displayed from inferences and probability calculations developed by the method and system.

[0042]

[0043] DESCRIPTION OF THE INVENTION

[0044] An object of the present invention is to provide a system and method (500) for the recovery of semantic data of spaces, contextualization of environments and development of inferences using sensor data (400) and Artificial Intelligence tools, to obtain the future state of an environment, based on the increase of semantic and relational data, as shown in Figure 1.

[0045] Another object of the present invention is to provide a system that, through predictive models of Machine Learning (104), algorithms for the dynamic construction of regions (108), and validation of hierarchies (110), allows organizations, corporations or companies to identify future events with the objective of making correct decisions in real time, reducing costs, increasing visibility, increasing security and productivity.

[0046] The system comprises an EDGE AI unit (100) for deploying and executing one or more Artificial Intelligence models (104), sensors for collecting audio, video, point cloud, temperature, or other data (400), and a cloud platform (300) for managing the enriched information and data generated by the system. The method of the invention considers the implementation of an Edge Artificial Intelligence or Edge AI device (100) for developing inferences about future states of a scenario, provided with interfaces for consuming video, audio, temperature, point cloud, or other data, where the activities of the context or environment, elements, objects, people, or any other element of the temporal space are recorded second by second.

[0047] The Edge AI device provides inferences, probability of occurrence of future events and enriched context recording and feeds this data to a management platform (300), from where the user views the enriched context for different spaces and specifically the indicators of future events.

[0048] Additionally, and as can be seen in Figure 7, the inferences and enriched context information are processed in real time by the Edge AI unit (100), which simultaneously, based on the records and results obtained, sends data corresponding to the enriched context to the local management platform (120), and audio or visual alerts in real time (122) to the scenario or site through alert means (124), in order to warn of taking actions in real time to produce changes in the current state, mitigating undesired future states.The data collected by the Cloud platform (300) are intended to be analyzed, processed and presented to the user in a metric graphic form (dashboards) where they can view enriched context indicators in dashboards, such as the one shown in Figures 9 and 10, from where the user can easily and user-friendly consult the conditions of the space, elements, regions, enriched context and inferences and probabilities about future states.

[0049] Access to the cloud platform (300) is provided to the organization for the user to track operational conditions, warnings, alerts, events and / or behaviors in general.

[0050] The inference and obtaining of enriched context is developed in the Edge AI unit (100) recursively at a regular time interval in milliseconds, each iteration is composed of executions of the predictive model(s), execution of calculation routines, data structuring and triggering of alerts, all executed on processors of the CPU (802), DSP (804), GPU (806) and NPU (808) type. Throughout the process, the Edge AI unit (100) is responsible for executing the processes automatically and reestablishing operation in case of failures. The enriched context environment (ECE) as shown in Figure 9, is presented to the user through a screen connected to the video output interface of the edge Artificial Intelligence unit, or through the platform, and allows the deployment of the enriched context for testing and adjustment purposes.The rich context environment includes a digital equivalent of the scene (point cloud) or a video of the scene, on which information (Metadata) corresponding to the elements, objects, regions, hierarchies and inferences is indexed.

[0051] DESCRIPTION OF THE PREFERRED MODALITY

[0052] The invention consists of a system composed of sensors

[0053] (400) for the acquisition of data from the scene or physical space, Edge AI (100) edge Artificial Intelligence devices, and Deep Learning predictive models (104) that are correlated with an inference method (500) for the prediction of future states, which allow the recovery of semantic data of spaces, contextualization of environments and development of inferences to predict future states, allowing the increase of environmental data and understanding of the relational dynamics of its elements, significance of the content, a representation of the state space enriched with complex information and real-time inferences about future events.

[0054] The system comprises an Edge Artificial Intelligence or Edge AI unit (100) that is placed at the site of use, space or physical location, for the acquisition of data from sensors (400) installed at said location, so that, as a result, the Edge AI device consumes said data and feeds the AI ​​predictive model(s) with them to obtain highly reliable inferences and context enrichment that is then presented in a visualization environment or Cloud platform (300), from where the user can follow up and obtain recommendations.

[0055] As can be seen in Figure 8, where different usage modes are described, from the installation of the Edge AI device on site, data is collected from sensors, and one or more predictive models (104) allow inferences to be made locally from the data, detecting elements or objects in the scene, while routines are executed for hierarchy recognition and context identification (110) and determination of the probability of occurrence of an event (112), and a location calculation engine based on binary markers (106) allows the location of objects, elements or people to be estimated (108).Simultaneously, an identifier (binary marker) can be positioned on the element, person or object of the scene to record its space-time position in the scenario or physical environment and correlate this data with the inferences made by artificial intelligence models, as shown in Figure 9 where workers wear helmets with a binary marker (710) for identification.

[0056] The capture of scenario data using sensors (400) collects information on video, audio, image, temperature, point cloud, among others. The Edge AI device (100) consumes the data and delivers it to the corresponding predictive model (104), which is responsible for developing the inferences.

[0057] The system also includes at least one portable device, such as a bracelet, card, or watch, worn by the person or item moving through the environment or surroundings, allowing them to receive real-time alerts about future situations or deviations.

[0058] As can be seen in Figure 7, one or more predictive models of future states with advanced calculation components are preloaded in the model and inference module (104), which in an embedded manner executes them at high speed in the most appropriate available processing unit based on the expected latency and resource consumption (802-808) and allows the detection of future events related to the detected context elements.

[0059] The system also incorporates a cloud management platform (300) that centrally controls the Edge AI devices (100), the predictive models (104), manages the licenses for using the system and method of the invention, records user information, stores and processes information on the inferences made, facilitating the improvement of the predictive models and their real-time adjustment as the system operates. More particularly, the management platform manages the predictive models that are loaded into the Edge AI units, base hierarchies, base contexts, warning and alert configuration, and, on the other hand, concentrates the inference data generated by the Edge AI units that are arranged in the space or environment, making it easier for Artificial Intelligence algorithms to develop recommendations that facilitate an improvement in safety, productivity, and operational efficiency in real time.

[0060] This generates high-speed data consumption for real-time processing and inferences based on edge artificial intelligence (Edge AI), calculation execution, probability estimation, and alert activation, for recognition adjustments, training, and recalibration of predictive models by specialists.

[0061] As can be seen in Figure 3, the system comprises an online data load and predictive models of Artificial Intelligence from the Local Management Environment (200) to the Edge AI unit, which when executed and consuming data from different sensors and sources, allows obtaining inferences that translate into enriched information about the environment, to make a recommendation for improvements in real time, which can determine the success of an operation and in this way constitute the entire method for the recovery of semantic data of spaces, contextualization of environments, and development of inferences (future states). With the above, a deployment of the inference and feedback mechanism is carried out in real time for low latency decision making in the environment or space.

[0062] The connection between the local Management Environment and the Edge AI Unit is done through an integration API.

[0063] In accordance with one aspect of the present invention, the system of the invention comprises integrations with analog and digital media for collecting scene data before and during the inference process.

[0064] The integration feature is provided through the integration manager of the Local Management Environment (200) with other platforms that provide information about the environment and elements that participate in it, providing additional information to the Edge AI device for complete decision making and generation of alerts based on knowledge of multiple systems.

[0065] The digital data capture feature is provided by the system devices in charge of acquiring data from the environment (400), which read space-time information, where the data generated by sensors feeds the Edge AI unit (100) where data is collected from different sources, making it possible to develop inferences about the future state of the scene or environment.

[0066] A separate video display (LCD Screen) (700) connected to the Edge AI device or unit is provided, so that the system administrator can monitor the development of inferences, probabilities, regions and obtaining future states and developing the necessary adjustment.

[0067] Regarding the method of the invention, during the inference process the data of the scene or context are subjected to predictive models ( 104 ) where inferences and probability calculations are obtained ( 112 ) in order to obtain the context enriched with future information, which allows for data-rich decision making. Alternatively to the predictive model, the invention will incorporate advanced probability calculations ( 112 ) to increase the enrichment of the context, while the results of the inferences and possible future states will be provided to the management platform for the development of improvements.

[0068] In accordance with a related aspect of the present invention, the predictive models are managed centrally by the platform (300) or from the Local Management Environment (200).

[0069] The Edge AI system (100) can operate independently of the cloud and internet management platform, facilitating the obtaining of inferences, warnings and alerts in real time without the need for connection to the internet network and without the need for a monitoring platform.

Claims

Title: Method and System for understanding context and obtaining real-time inferences about future events using edge Artificial Intelligence RE-UNIONIZATIONS 1. System that provides a technique for the recovery of semantic data of physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, CHARACTERIZED because the data generated by sensors feeds the Edge AI unit where data is collected from different sources and the development of inferences about the future state of the scene or physical space is possible.

2. System that provides a technique for the recovery of semantic data of physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, CHARACTERIZED because the capture of characteristics or features from the data allows the increase of semantic data and understanding of the relational dynamics for the meaning of the content, a representation of the state space enriched with complex information and real-time inferences.

3. System that provides a technique for the recovery of semantic data from physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, CHARACTERIZED because the inference and obtaining of enriched context is developed in the Edge AI unit recursively at a regular interval of multiple times per second, where each iteration is composed of executions of the predictive model(s), execution of calculation routines, data structuring and triggering of alerts.

4. System that provides a technique for the recovery of semantic data of physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, CHARACTERIZED because it has predictive models of Deep Learning that correlate with a method for the prediction of future states and the improvement of operational and security indicators of an organization or corporation.

5. System that provides a technique for the recovery of semantic data from physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, CHARACTERIZED because it has an EDGE AI hardware unit for the execution of predictive models at high speed, being able to execute them on CPU, DSP, GPU and NPU or in a combination of the above.

6. System that provides a technique for the recovery of semantic data of physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, CHARACTERIZED because it executes artificial intelligence algorithms directly in the Edge AI units, where operational data is collected, instead of sending all that data to a centralized location for processing in the cloud, allowing latency-free processing, immediate decision-making, and reducing dependence on constant network connectivity.

7. System that provides a technique for the recovery of semantic data of physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, CHARACTERIZED because it allows the detection of elements or objects in scene generation automatic region recognition, hierarchy recognition and context identification, which allow a determination of the probability of occurrence of a future event.

8. System that provides a technique for the recovery of semantic data of physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, CHARACTERIZED because it has modules of Hierarchization and Context and Logic and Probability, where the first is in charge of recognizing the hierarchy based on a match with the previously defined ones, validating dependent and independent elements and their relationships defined in the hierarchy and the second is in charge of checking the probability of occurrence of the event of the hierarchy.

9. System that provides a technique for the recovery of semantic data of physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, CHARACTERIZED because it consumes video, audio, point cloud, temperature data, among others, and feeds this data to the AI ​​predictive model(s) to obtain inferences. highly reliable for context enrichment.

10. System that provides a technique for the recovery of semantic data of physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, CHARACTERIZED because it incorporates algorithms for the construction and dynamic management of regions with elements and relational dynamics between objects, which allows the implementation of hierarchical validations between elements of a region or between regions, which as a result allows the identification of a context and / or activity.

11. System that provides a technique for the recovery of semantic data of physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, CHARACTERIZED because for video data sources it implements the visualization of the video of the scene in real time, inferences, elements, regions, contexts and probability of occurrence of events, and when registering alerts it records the video of the alert with the metadata.

12. System that provides a technique for the recovery of semantic data of physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, CHARACTERIZED because it has a location calculation engine based on binary markers that allows estimating the location of objects, elements or people.

13. System that provides a technique for the recovery of semantic data from physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, according to claim 1, CHARACTERIZED in that said Artificial Intelligence unit is deployed at the edge or site of interest and has the resistance and certifications to operate in industrial environments.

14. System that provides a technique for the recovery of semantic data of physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, according to claim 1 CHARACTERIZED in that said edge unit has hardware and software developed to execute Artificial Intelligence models with a minimum energy consumption which translates into a lower environmental impact.

15. System that provides a technique for the recovery of semantic data of physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, according to claim 1 CHARACTERIZED in that the Edge AI unit has physical interfaces of the analog and digital type for the activation of notices and alerts to the physical environment in real time.

16. System that provides a technique for the recovery of semantic data from physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, CHARACTERIZED because it operates based on a module for local management of the EDGE AI unit, predictive models and parameterization from a local management environment machine or alternatively delegate this control to the cloud.

17. System that provides a technique for the recovery of semantic data from physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, CHARACTERIZED because the data already processed and collected by the Cloud platform are presented to the user in a metric graphical form (dashboards) where they can view enriched context indicators, warnings and alerts.

18. System that provides a technique for the recovery of semantic data of physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, CHARACTERIZED because it comprises at least one portable device such as a bracelet, card or watch that is carried by the element or person who moves in the environment or surroundings and that allows them to receive real-time alerts about situations.

19. System that provides a technique for the recovery of semantic data of physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, according to claim 1 CHARACTERIZED in that said Edge AI device has a video output interface (HDMI) for the connection of a video monitor that a system administrator or supervisor can use to view the metadata and augmented context.

20. System that provides a technique for the recovery of semantic data from physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, in accordance with claim 1 CHARACTERIZED because it considers a preloading of parameters of hierarchies and contexts which can be carried out from a local server or from the cloud.

21. System that provides a technique for the recovery of semantic data of physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, according to claim 5 CHARACTERIZED said models in their output deliver inference results (predictions), which are deployed in a management platform in the cloud (Cloud), from where the augmented context is recorded, and availability of control panels with information on deviations and specifically indicators of warnings and alerts.

22. System that provides a technique for the recovery of semantic data of physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, according to claim 16 CHARACTERIZED because said platform controls Edge AI devices, manages predictive models, administers licenses for using the system and method of the invention, records prediction information, stores and facilitates visualization for rapid decision-making.

23. System that provides a technique for the recovery of semantic data of physical spaces, contextualization of environments, and development of inferences (future states) through the use of Artificial Intelligence, according to claim 16 CHARACTERIZED because said local management environment allows integration with other platforms through an API that provides information about the environment and elements that participate in it, facilitating complete decision-making and generation of alerts based on information from multiple systems.

Images