# US 2026/0064918 A1 · Context-based safety systems for subterranean environment and methods of operating thereof > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2026/0064918 A1 - Document ID: US20260064918A1 - Status: Published application - Kind: A1 - Application number: 19/305,998 - Filed: 2025-08-21 - Published: 2026-03-05 - Earliest claimed: 2024-08-27 - Assignees: LoopX Innovation Inc. - Applicants: LoopX Innovation Inc. - Inventors: Yaodong CUI; Chao YU; Jinwei ZHANG - Web page: https://patentgrep.com/patent/US20260064918A1 ## CPC - G06F 30/27: using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) ## Abstract There is provided context-based safety systems, and methods for operating thereof, for a subterranean environment. An example context-based safety system for a subterranean environment includes: a plurality of sensors configured to collect a set of operating data; and a processor configured to: continuously receive the operating data from sensors, the set of operating data including at least one visual data; receive a user prompt from a user defining a risk assessment in respect of the operating data; apply the user prompt to a vision-language model to: identify one or more abnormal activities observed from the set of operating data, each abnormal activity being unexpected within a safety context associated with the subterranean environment and an activity type of that abnormal activity; and identify one or more risky activities from the abnormal activities for the risk assessment; and generate one or more recommendations in response to the risky activities. ## Claims ### Claim 1 A context-based safety system for a subterranean environment, the safety system comprises: a plurality of sensors configured to collect a set of operating data within the subterranean environment; and a processor in communication with the plurality of sensors and configured to: continuously receive the set of operating data from the plurality of sensors, the set of operating data comprising at least one visual data; receive a user prompt from a user defining a risk assessment in respect of the set of operating data; apply the user prompt to a vision-language model to: identify one or more abnormal activities observed from the set of operating data, each abnormal activity being unexpected within a safety context associated with the subterranean environment and an activity type of that abnormal activity; and identify one or more risky activities from the one or more abnormal activities for the risk assessment; and generate one or more recommendations in response to the one or more risky activities. ### Claim 2 (depends on 1) The context-based safety system of claim 1, wherein the processor is configured to apply the vision-language model to assign a risk level to each abnormal activity based on the safety context associated with the subterranean environment and that activity type. ### Claim 3 (depends on 1) The context-based safety system of claim 1, wherein the processor is configured to: define a set of safety contexts associated with the subterranean environment for the vision-language model, the set of safety contexts identifying one or more situations related to one or more activities requiring safety compliance within the subterranean environment. ### Claim 4 (depends on 1) The context-based safety system of claim 1, wherein the risk assessment comprises an operator performance assessment and the processor is configured to identify the one or more risky activities associated with operator performance, and generate the one or more recommendations for improving the operator performance. ### Claim 5 (depends on 1) The context-based safety system of claim 1, wherein the risk assessment comprises an environment safety assessment and the processor is configured to identify the one or more risky activities related to an unsafe environment, and generate the one or more recommendations for improving the unsafe environment. ### Claim 6 (depends on 1) The context-based safety system of claim 1, wherein the processor is further configured to: generate a safety report in compliance with regulatory requirements for summarizing at least the one or more risky activities and the one or more recommendations. ### Claim 7 (depends on 1) The context-based safety system of claim 1, wherein the plurality of sensors comprises at least one sensor coupled to a machinery operating within the subterranean environment. ### Claim 8 (depends on 1) The context-based safety system of claim 1, wherein the processor is further configured to: optimize the vision-language model with at least the set of operating data. ### Claim 9 A method for operating a context-based safety system for a subterranean environment, the method comprises: collecting, by a plurality of sensors, a set of operating data within the subterranean environment; continuously receiving the set of operating data from the plurality of sensors, the set of operating data comprising at least one visual data; receiving a user prompt from a user defining a risk assessment in respect of the set of operating data; apply the user prompt to a vision-language model to: identify one or more abnormal activities observed from the set of operating data, each abnormal activity being unexpected within a safety context associated with the subterranean environment and an activity type associated to that abnormal activity; and identify one or more risky activities from the one or more abnormal activities for the risk assessment; and generate one or more recommendations in response to the one or more risky activities. ### Claim 10 (depends on 9) The method of claim 9, wherein the processor is further operated to apply the vision-language model to assign a risk level to each abnormal activity based on the safety context associated with the subterranean environment and that activity type. ### Claim 11 (depends on 9) The method of claim 9, wherein the processor is further operated to define a set of safety contexts associated with the subterranean environment for the vision-language model, the set of safety contexts identifying one or more situations related to one or more activities requiring safety compliance within the subterranean environment. ### Claim 12 (depends on 9) The method of claim 9, wherein the risk assessment comprises an operator performance assessment and the processor is further operated to identify the one or more risky activities associated with operator performance, and generate the one or more recommendations for improving the operator performance. ### Claim 13 (depends on 9) The method of claim 9, wherein the risk assessment comprises an environment safety assessment and the processor is further operated to identify the one or more risky activities related to an unsafe environment, and generate the one or more recommendations for improving the unsafe environment. ### Claim 14 (depends on 9) The method of claim 9, wherein the processor is further operated to generate a safety report in compliance with regulatory requirements for summarizing at least the one or more risky activities and the one or more recommendations. ### Claim 15 (depends on 9) The method of claim 9, wherein the plurality of sensors comprises at least one sensor coupled to a machinery operating within the subterranean environment. ### Claim 16 (depends on 9) The method of claim 9, wherein the processor is further operated to optimize the vision-language model with at least the set of operating data.