An AI native intelligent interaction terminal with active sensing and emergency response capability

Abstract

The application discloses an AI native intelligent interaction terminal with active sensing and emergency response capability, and belongs to the technical field of artificial intelligence terminals and man-machine interaction. The terminal integrates a multi-modal interaction module, a local AI inference engine capable of offline voice recognition, negotiation strategy generation and multi-language real-time translation, and an end-cloud collaborative scheduling unit. The terminal adopts an independent transfer box form with integrated charging function or an integrated terminal form with integrated charging bin, and can provide charging and power supply functions for matched earphones, glasses and other peripherals. The core feature of the terminal is the built-in motion state sensor and environmental sound recognition model, which can actively sense user falls, impacts or dangerous acoustic events, and automatically trigger an emergency response program. The terminal also supports a covert real-time word prompting function based on voiceprint recognition and cognitive state detection, provides long-term personalized services through a localized user cognitive model, and supports seamless task relay with a local AI host. The application deeply integrates intelligent interaction, active safety guardianship and privacy protection, and can be widely applied to various independent AI interaction scenes.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence terminals and human-computer interaction technology, specifically to an AI-native intelligent interaction terminal that integrates multimodal interaction, local intelligent reasoning, and proactive security protection. Background Technology

[0002] With the increasing popularity of AI applications such as real-time translation, negotiation assistance, and emotional interaction, users are increasingly demanding a portable smart interactive terminal that can be used independently of a smartphone. Existing portable devices mainly fall into two categories: one is Bluetooth accessories that rely on smartphones and lack independent calling and intelligent processing capabilities; the other is single-function translators or voice recorders that lack deep semantic understanding and strategy generation capabilities. In terms of security, existing portable terminals generally lack reliable hardware-level security mechanisms. Their data storage and core computing reside within a general-purpose operating system environment, leaving sensitive conversations and user voiceprints vulnerable to theft by malicious applications. Regarding user experience, existing terminals are passively responsive tools, unable to proactively perceive the user's environment, physical condition, or sudden dangerous situations. Furthermore, existing devices lack long-term cognitive learning capabilities tailored to users and the ability to seamlessly relay tasks with a home AI host. Existing emergency response plans lack clear data retention rules after triggering, leaving users unsure whether on-site data will be automatically destroyed after transmission. In offline states, the emergency communication protection mechanisms of existing devices are also inadequate. Summary of the Invention

[0003] The purpose of this invention is to provide an AI-native intelligent interactive terminal with proactive sensing and emergency response capabilities: The core improvement of this invention lies in the following: First, it integrates local AI reasoning capabilities, enabling it to independently complete all core tasks, including speech recognition, semantic understanding, intent analysis, negotiation strategy generation, and real-time multilingual translation, even in a completely offline state. The negotiation strategy generation function is based on a lightweight strategy model trained using a pre-built game theory rule base and historical user negotiation data. Strategy types include concession pace suggestions, key point reiteration prompts, and risk warnings. The model parameters are adapted to the terminal's local computing power, supporting stable operation even in offline environments. Secondly, it possesses proactive safety protection and emergency response capabilities. Built-in motion sensors and an ambient sound recognition model enable the terminal to monitor the user's posture in real time, autonomously determining whether the user has experienced at least one of the following: a fall, impact, or dangerous acoustic event, and automatically triggering the emergency response procedure. After the emergency response is triggered, the system can determine the user's safe response through preset voice commands or specific gestures. In offline mode, the terminal sends a distress signal via Bluetooth directly to a preset device. The on-site audio is automatically destroyed after sending the distress signal, retaining only metadata such as timestamps and location information for subsequent model optimization, complying with data privacy protection regulations. Third, regarding personalized experiences, the terminal continuously learns users' behavioral habits and language styles locally through long-term user cognitive model units. Fourth, the terminal also supports seamless task handover with the local AI host, handling complex tasks at home and automatically switching to the personal terminal when the user is away. When the user approaches the in-vehicle system or smart home devices, the terminal automatically detects the proximity signal and proactively pushes the current context information, achieving seamless cross-device collaboration. Attached Figure Description

[0004] Figure 1 This is a schematic diagram of the terminal system architecture of the present invention; Figure 2 This is a schematic diagram of the active sensing and emergency response process of the present invention; Figure 3 This is a schematic diagram of the cognitive lag detection and hidden prompting process of the present invention; Figure 4 This is a schematic diagram illustrating the seamless task relay process between the present invention and the local AI host. Detailed Implementation

[0005] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and embodiments: Example 1: Core Interaction and Active Protection (Integrated Terminal Form with Charging Case) Users wear this terminal (integrated with a charging case) and the accompanying Bluetooth earphones. During business meetings, the multimodal interaction module captures voice information via a microphone array and visual information via external smart glasses. A local AI inference engine performs real-time translation and negotiation strategy generation within the terminal, outputting the results through the multimodal interaction module. When the user places the earphones back into the charging case, the terminal automatically charges them. After the meeting, the user walked home. The terminal's built-in accelerometer and gyroscope continuously monitored the user's movement. When the user slipped and fell, the motion sensor detected a peak instantaneous acceleration exceeding the normal range and the body posture changed from upright to horizontal; the microphone simultaneously captured the impact sound. The local AI inference engine determined that the user had fallen and immediately initiated the emergency response procedure. The system played a low-volume questioning voice through the earpiece, requiring the user to respond safely using a preset voice command or specific gesture. If the user did not respond safely within a preset time, the system automatically sent a distress message containing the current location to a preset emergency contact. The audio was automatically destroyed after sending the distress message, retaining only metadata such as the timestamp and location information for subsequent model optimization. Upon triggering an emergency response, the system automatically retrieves the user's medical data, such as blood type and allergy history (via a standardized secure base interface), packages it, and sends it to the emergency response platform, significantly shortening emergency response time. When a user enters a sensitive location such as a conference room, a privacy sandbox mode can be activated with a single click: the system disables all sensors, retaining only basic call functions, and hardware-level indicator lights display the privacy status, meeting data security regulatory requirements. Example 2: Business Negotiation Assistance in a Completely Offline Environment (Standalone Transfer Box Form) A businessperson boarded the plane carrying this terminal (a standalone transit box with integrated charging). With no internet connection available on the plane, the terminal's local AI inference engine independently completed all tasks offline. During cross-border business negotiations, the terminal used a large local model for real-time multilingual translation. Combining a pre-built game theory rule base and a lightweight strategy model trained on the user's historical negotiation data, it analyzed the other party's tone of voice, emotional state, and negotiation intentions in real time. It automatically generated negotiation strategy suggestions, including concession pace recommendations, key point reiteration prompts, and risk warnings, which were then pushed through the accompanying bone conduction headphones. The parameter count of the strategy model was adapted to the terminal's local computing power, ensuring stable and efficient operation even offline. Throughout the negotiation process, all voice, text, and strategy data are processed locally on the terminal without being uploaded to any cloud. The terminal also provides charging functionality for the accompanying bone conduction headphones, ensuring sufficient power for extended negotiations. After the negotiation concludes, the system automatically generates negotiation minutes, noting key decision points, changes in the other party's emotions, and records of adjustments to one's own strategy. Example 3: Cognitive Assistance in Product Introduction A salesperson was introducing a new product to a customer at a trade show. When discussing certain parameters, the salesperson paused. The local AI inference engine used voiceprint recognition technology to separate the salesperson's speech from the noisy environment, detecting that the pause duration and speech rate changes exceeded normal ranges. The system then automatically matched subsequent content from a pre-imported product knowledge base based on the salesperson's last complete statement, generating a connecting script, which was then pushed to the salesperson via a multimodal interaction module. When a customer paused in front of the booth to think, the system determined that the customer was not a target user, and their silence did not trigger any false alarms. As an example configuration, the cognitive stuttering threshold can be set when the pause duration exceeds a preset duration and the speech rate decreases by more than a preset percentage. The system can dynamically adjust this threshold based on historical interaction data. When multiple cognitive stuttering feature judgments conflict, a sudden drop in the frequency of micro-expression changes is given the highest priority to ensure the accuracy of prompting. After a user has used the terminal for a long time, the system analyzes the changing trends of the user's language fluency and reaction time, and generates a cognitive decline risk report locally. All raw data is not uploaded to the cloud, and only the user can view the report. Example 4: Seamless Task Handover with Local AI Host Users perform complex market analysis tasks at home using a local AI host. When a user needs to leave the house, the terminal automatically detects that the connection with the local host is about to be lost. The task relay module synchronizes the context data of the currently ongoing market analysis task to the terminal via a local encrypted data channel. As an example configuration, the encryption of the synchronized data can use national cryptographic algorithms, and the key is generated through near-field communication. After the user leaves the house, the task continues to be processed on the terminal, with the context fully synchronized and the user experiencing no sense of transition. When the user returns home, the terminal automatically detects that the local host is available again, and the ongoing task automatically switches back to local host processing. The local host has stronger computing power and can complete complex inference faster. The entire process requires no manual operation from the user. When a user approaches an in-vehicle system or smart home device, the terminal automatically detects the proximity signal of the device and actively pushes the current context information, without the need for manual wake-up, thus achieving seamless cross-device collaboration. The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An AI-native intelligent interactive terminal with proactive sensing and emergency response capabilities, characterized in that, It includes the terminal body, multimodal interaction module, local AI inference engine and edge-cloud collaborative scheduling unit; The terminal body is used to accommodate and carry the internal functional modules and provide power supply for the connected peripherals; The multimodal interaction module is electrically connected to the local AI inference engine and is used to collect the target user's voice and visual data, as well as output translation results, script suggestions, and interactive feedback content. The multimodal interaction module also integrates a motion state sensor to monitor the target user's motion state and body posture in real time. The local AI inference engine is used to complete at least one inference task among speech recognition, semantic understanding, intent analysis and strategy generation based on the collected multimodal data in an offline state. The strategy generation includes generating negotiation strategies in real time in a completely offline environment. The local AI inference engine is also used to determine whether the target user has fallen, collided or experienced a dangerous acoustic event based on the data collected by the motion state sensor and combined with the environmental sound recognition model, and automatically trigger an emergency response procedure after the determination occurs. The edge-cloud collaborative scheduling unit is connected to the local AI inference engine and is used to offload non-sensitive and high-computing-power-requirement tasks to the cloud for computation based on preset data sensitivity level rules and the current network status, while restricting sensitive and private data to be processed locally on the terminal throughout the process.

2. The terminal according to claim 1, characterized in that, The terminal body can be any one of the following forms: an independent transfer box with integrated charging function, an integrated terminal with integrated charging compartment, or a wearable form; when it adopts the form of an independent transfer box with integrated charging function or an integrated terminal with integrated charging compartment, the terminal body is also used to provide charging and power supply functions for matching peripherals such as headphones or glasses.

3. The terminal according to claim 1, characterized in that, It also includes a standardized security base interface, integrated inside the terminal body, used to establish a secure data channel with external security management devices or built-in hardware root of trust modules, providing a protected execution environment that cannot be read externally for the core operation of the local AI inference engine; the standardized security base interface automatically identifies the type of security environment it is connected to and selects the corresponding security protocol for communication.

4. The terminal according to claim 1, characterized in that, The multimodal interaction module collects voice data that filters out non-target speaker speech through voice separation, voiceprint locking, and beamforming technologies; the collected visual data includes at least one of facial expressions, lip reading, gestures, and scene status information.

5. The terminal according to claim 1, characterized in that, The local AI inference engine includes a user cognitive model unit, which is used to continuously learn the target user's behavioral habits, language style and decision-making preferences in long-term interaction, and to build and update the target user's exclusive cognitive baseline model; all learning data used to build the cognitive baseline model is stored and processed locally on the terminal.

6. The terminal according to claim 1, characterized in that, The data sensitivity level rules of the edge-cloud collaborative scheduling unit are preset with fixed boundaries. Identity data, biometric data, and personal private behavior data are classified as local exclusive privacy data, which are processed only on the terminal. General model calculations and public data query content are classified as data that can be processed in the cloud.

7. The terminal according to claim 1, characterized in that, When the network is disconnected, the terminal automatically switches to a fully local offline operation mode, and all functions are independently completed by the local AI inference engine, including multilingual real-time translation and negotiation strategy generation in offline mode; after the network is restored, incremental synchronization of non-sensitive data is silently completed.

8. The terminal according to claim 1, characterized in that, The terminal supports multi-device secure collaboration, and can achieve secure synchronization and collaborative processing of user cognitive models, scene data and unfinished tasks with other terminals of the same model through encrypted data channels. During the synchronization process, all data does not pass through the cloud.

9. The terminal according to claim 1, characterized in that, The terminal is used in at least one of the following scenarios: business negotiation, cross-language communication, meeting assistance, public speaking, and product introduction.

10. The terminal according to claim 1, characterized in that, The local AI inference engine is also used to identify the target user through biometric recognition technology and extract the cognitive state features of the target user; when the cognitive lag feature of the target user is detected to exceed a preset threshold, the engine locates the subsequent content in the preset scene-related knowledge base and generates a prompt message based on the complete sentence last output by the target user. When multiple cognitive lag features conflict, the sudden drop in the frequency of micro-expression changes is given the highest priority. The terminal does not trigger a prompt when it detects a pause by a non-target user.

11. The terminal according to claim 1, characterized in that, The terminal also includes a task relay module, which is used to automatically detect the connection status with the local AI host; when the user leaves home, complex tasks being performed on the local AI host are automatically switched to the terminal for continued processing, and the context is fully synchronized during the switching process; when the user returns home, the task is automatically switched back to the local AI host for processing.

Images