Interactive intelligent assistant robot, interaction system based on AI large model and field interaction method

Abstract

The invention relates to the technical field of robots and artificial intelligence, and discloses an interactive intelligent assistant robot, an interactive system based on an AI large model and a field interactive method.The robot comprises a base device, and the base device comprises a shell, a power supply unit, a sensing unit, a communication and control unit and an interactive output unit; the power supply unit is used for charging or supplying power to an external intelligent terminal and supplying power to each unit; the sensing unit is used for sensing an interaction signal; the communication and control unit is used for receiving and processing a signal of the sensing unit and carrying out interactive communication between the units in the robot and between the units and an external network; and the interactive output unit is used for outputting changed mechanical actions, light rays, audios and images. According to the intelligent assistant robot, a mode of combining hardware and software, local and cloud, on-site triggering, interactive output and remote computing control is adopted, so that the intelligent assistant robot can meet rigid requirements of users and elastic requirements of office, education, entertainment and the like at the same time.

Description

Technical Field

[0001] This invention relates to the fields of robotics and artificial intelligence, specifically to an interactive intelligent assistant robot, an interactive system based on an AI large model, and an on-site interaction method. It also relates to intelligent charging, multimodal human-computer interaction, integrated thermal design and counterweight, and edge computing and large model fusion technologies, which can be applied to various desktop / indoor scenarios such as offices, homes, education, and smart homes. Background Technology

[0002] With the rapid development of AI big data models, smart terminals, and cloud service technologies, service robots are expanding deeply from the industrial sector to the civilian sector. Desktop intelligent assistant robots have become an important sub-category suitable for close-range office, study, and home scenarios. Market demand for them is upgrading from single interaction / charging functions to multi-functional integration, intelligence, and collaboration. However, existing technologies have the following core shortcomings: 1. Defects in robot hardware design: Existing desktop charging / interactive robots separate the charging unit from the body counterweight, resulting in a high center of gravity and easy tipping; the robotic arm has limited degrees of freedom and cannot achieve multi-angle spatial positioning of the smart terminal; the shell has poor heat dissipation and has not achieved an integrated design of charging, counterweight, and heat dissipation.

[0003] 2. Limited functionality and poor collaboration: Existing robots only have basic charging or simple voice interaction functions, and cannot deeply collaborate with smart terminals such as smartphones and tablets. They also do not access the large AI model resources of cloud servers, making it difficult to meet users' diverse needs such as charging, companionship, assistance, and education.

[0004] 3. Low level of intelligence in charging strategy: Existing charging control relies solely on the hardware circuitry of the power adapter, lacking intelligent power allocation and multi-objective optimization strategies. It cannot dynamically adjust the charging scheme based on device status, user habits, and grid load, which can easily damage the battery or cause energy waste.

[0005] 4. Poor human-computer interaction experience: Existing robots have a single sensing method, only supporting single-modal interaction of voice / touch, and lack the ability to fuse multimodal perception; the interaction response is passive, unable to understand the implicit needs of users, and unable to achieve personalized interaction based on user profiles.

[0006] 5. Prominent contradictions in resource allocation: If robots rely entirely on themselves to achieve sensing, computing, and output functions, it will lead to high hardware costs and difficulty in widespread adoption; if they rely too much on external devices, it will result in limited functionality and a decline in the interactive experience.

[0007] For example, the desktop mini-robot disclosed in CN 110281245 A can only achieve basic route control and cannot collaborate with smart terminals or cloud servers; the charging system disclosed in CN 110299752 A only achieves charging control through hardware and has no intelligent strategy selection; existing desktop charging robots such as CN106210212A can only achieve simple flip charging, with low degrees of freedom of the robotic arm and no interactive functions.

[0008] In summary, existing technologies have not yet resolved the core contradictions between low resource consumption and increased functionality, and between increased functionality and reduced costs for robots. They also cannot simultaneously meet users' rigid charging needs and flexible needs such as interaction, companionship, and assistance. There is an urgent need for an integrated, intelligent, and collaborative interactive intelligent assistant robot and its supporting systems and methods. Summary of the Invention

[0009] (a) Technical issues The core technical problems to be solved by this invention mainly include: 1. How to improve functionality while reducing robot hardware costs, achieving integrated hardware for charging, weight distribution, heat dissipation, sensing, and interaction; 2. How to build a deep collaborative architecture among robots, smart terminals, and cloud servers to achieve resource sharing in sensing, computing, and output; 3. How to achieve multimodal perception fusion and user intent understanding through AI big data models, upgrading from passive response to proactive personalized intelligent interaction; 4. How to design intelligent charging strategies based on AI big data models to achieve multi-objective optimization of charging efficiency, battery life, energy consumption, and heat generation; 5. How to build a hybrid computing architecture combining local and cloud computing, balancing real-time interactive response with the intelligence of AI models; 6. How to achieve multimodal collaborative output between robots and smart terminals, enhancing the immersiveness and experience of human-computer interaction.

[0010] (II) Technical Solution To address the aforementioned technical problems, this invention provides a three-in-one technical solution integrating an interactive intelligent assistant robot, an AI-based large-scale interactive system, and a field interaction method. It adopts a design concept that combines hardware and software, local and cloud-based approaches, and field triggering and interactive output with remote computation and control, thereby achieving hardware integration, software collaboration, and a closed-loop methodology.

[0011] An interactive intelligent assistant robot includes a base device, which includes a housing, a power supply unit, a sensing unit, a communication and control unit, and an interactive output unit. The power supply unit is used to charge or supply power to the external smart terminal, and to supply power to the sensing unit, communication and control unit, and interactive output unit. It includes a power supply PCB circuit board and at least one power adapter. The power supply PCB circuit board and its onboard power adapter provide counterweight for the base device and lower the overall center of gravity of the base device. The sensing unit is used to sense interactive signals and includes an infrared sensing module, an environmental sensor module, and a voice pickup module. The communication and control unit is used to receive and process signals from the sensing unit, conduct interactive communication between various units within the robot and with external networks, and control the operation of the power supply unit and interactive output unit; the communication and control unit includes an edge computing module for performing lightweight AI inference tasks locally. The interactive output unit is used to output one or more of the following: mechanical action, light, audio, and image, under the triggering of an interactive signal or the control of a communication and control unit. It can also coordinate with an external smart terminal to output one or more of the following: mechanical action, light, audio, and image, which change in coordination with each other. The power supply unit, sensing unit, communication and control unit, and interactive output unit are electrically connected to each other.

[0012] The present invention also provides an interactive system based on an AI large model, including the above-mentioned interactive intelligent assistant robot, and also including an intelligent terminal and a cloud server. The interactive intelligent assistant robot, the intelligent terminal and the cloud server are connected and communicate through wired or wireless networks. The interactive intelligent assistant robot is used to provide physical support, charging or power supply for the intelligent terminal, establish local communication with the intelligent terminal, and then use its own sensing unit or the communication and control unit to call the sensing interface of the intelligent terminal to perceive interactive information; then send the perceived interactive information to the local communication and control unit or the AI ​​large model program of the cloud server for processing and calculation, and finally convert the returned control information into intelligent charging, power supply or interactive output information and execute it. The intelligent terminal is used to receive physical support, charging or power supply from the interactive intelligent assistant robot, and to establish communication with the interactive intelligent assistant robot or cloud server. It uses its own sensing, computing, storage and output interfaces to work in collaboration with the interactive intelligent assistant robot or cloud server to provide input data for the AI ​​large model program, or to output output information on site, including collaborative output with the interactive intelligent assistant robot. The cloud server has a built-in AI large model program, which is used to establish communication with interactive intelligent assistant robots or smart terminals, receive their input data, analyze and process it, and output intelligent charging, power supply, interactive output, and collaborative execution information. These information can be executed individually or collaboratively by the interactive intelligent assistant robot and the smart terminal to complete various interactive intelligent assistant tasks.

[0013] This invention also provides a live interaction method based on a large AI model, comprising the following steps: S1. Interactive Sensing Steps: The interactive intelligent assistant robot establishes a connection with the intelligent terminal, collects dynamic interactive sensing data of the user on site through its own sensing unit, and sends the collected interactive sensing data to the communication and control unit for preprocessing. S2, Intelligent Analysis Step: The communication and control unit sends the pre-processed interactive sensor data and information from the smart terminal to the cloud server. The AI ​​large model program analyzes the user's interactive needs, determines the optimal charging or power supply strategy, and generates charging, power supply and interactive output schemes. S3. Execution and Interaction Output Steps: The interactive intelligent assistant robot executes the aforementioned charging or power supply scheme; the interactive intelligent assistant robot and the intelligent terminal execute the aforementioned interaction output scheme separately or in concert. S4. Model Optimization Steps: The AI ​​large model program automatically memorizes and learns users' usage habits, evaluations, and feedback, and automatically upgrades and optimizes the AI ​​large model.

[0014] (III) Beneficial Effects The beneficial effects of the present invention include at least the following: 1. Reduced hardware costs and increased integration: Through the integrated counterweight and power supply architecture, and the architecture of local edge computing + cloud-based large model, the overall hardware cost of the robot body can be reduced by more than 30%; the three-section shell achieves integrated heat dissipation, light guiding and protection, and the modular design of the robotic arm and charging tray improves the device's adaptability to different scenarios.

[0015] 2. Intelligent charging, improved efficiency and safety: Intelligent power distribution and AI charging strategy optimization model significantly improve charging efficiency, extend battery life, and enable dynamic power distribution for simultaneous charging of multiple devices. The upper limit of total output power can be selected as needed.

[0016] 3. Enhanced functionality to meet diverse needs: It simultaneously enables functions such as smart charging, human body tracking, multimodal interaction, AI smart assistant, and smart home integration, catering to users' essential charging needs as well as flexible needs such as companionship, education, and office work.

[0017] 4. Enhanced interactive experience, from passive to proactive: Multimodal perception fusion and AI-based user intent understanding models enable accurate identification of users' explicit / implicit needs; personalized interaction solutions based on user profiles significantly improve user experience satisfaction.

[0018] 5. Strong collaboration and high resource utilization: The three-in-one architecture of robot-intelligent terminal-cloud server realizes the sharing of sensing, computing and output resources, and solves the contradiction between low resources and high functions of robot.

[0019] 6. High robustness and wide adaptability: The addition of an offline working mode and safety monitoring steps enables the system to continue to work normally when there is no network or equipment malfunction; the four-degree-of-freedom adjustment of the robotic arm and the modular structure design make it suitable for various scenarios such as office, education, smart home, and home entertainment.

[0020] 7. Continuous model optimization and increasing intelligence: The AI ​​big model performs incremental learning and transfer learning based on user interaction feedback, achieving self-optimization and self-upgrading, becoming smarter the more it is used.

[0021] 8. This invention adopts a design concept that combines hardware and software, local and cloud, and on-site triggering and remote computing, which solves the contradiction between low resources and increased functions, and between increased functions and reduced costs for assistant robots. It achieves reduced hardware costs and significantly improved charging efficiency, while meeting users' rigid charging needs and flexible needs such as interaction, companionship, and assistance. It can be widely used in scenarios such as office, education, smart home, and home entertainment. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the composition and connection structure of the interactive system according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the composition structure of the interactive intelligent assistant robot according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the overall three-dimensional external structure of the interactive intelligent assistant robot according to Embodiment 1 of the present invention; Figure 4 This is a schematic diagram of the three-dimensional assembly structure of the interactive intelligent assistant robot according to Embodiment 1 of the present invention; Figure 5 This is a three-dimensional external structural diagram of the base device in an embodiment of the present invention; Figure 6 This is a rear view of the base device in an embodiment of the present invention; Figure 7 This is a top view of the base device in an embodiment of the present invention; Figure 8 This is a three-dimensional assembly structure diagram of the base device in an embodiment of the present invention; Figure 9This is a top view of the internal structure of the base device in an embodiment of the present invention; Figure 10 This is a schematic diagram of the overall three-dimensional external structure of the charging tray device in an embodiment of the present invention; Figure 11 These are the front and rear views of the charging tray device in an embodiment of the present invention; Figure 12 This is a three-dimensional assembly structure diagram of the charging tray device in an embodiment of the present invention; Figure 13 This is a schematic diagram of the overall three-dimensional structure of the interactive intelligent assistant robot according to Embodiment 2 of the present invention; Figure 14 This is a schematic diagram of the three-dimensional assembly structure of the interactive intelligent assistant robot according to Embodiment 2 of the present invention; Figure 15 This is a three-dimensional external structural diagram of the robotic arm device in an embodiment of the present invention; Figure 16 This is a schematic diagram of the three-dimensional assembly structure of the robotic arm device in an embodiment of the present invention; Figure 17 This is a three-dimensional assembly structure diagram of one side component of the robotic arm device in an embodiment of the present invention; Figure 18 This is a schematic diagram of the system composition structure of a child education and companionship scenario in an embodiment of the present invention.

[0023] In the picture: 100. Interactive intelligent assistant robot; 200. Tablet computer; 300. Cloud server; 110. Base assembly; 111. Housing; 111a. Metal heat dissipation upper shell; 111b. Plastic middle frame; 111c. Metal heat dissipation lower shell; 111d. Lower joint support base; 112. Power supply unit; 112a. PCB circuit board; 112b. Power adapter; 113. Sensing unit; 113a. Infrared sensing module; 114. Communication and control unit; 114a. Communication and control circuit board; 115. Interactive output unit; 115a. Light output module; 115b. Image output module; 120. Charging tray device; 121. Charging tray; 122. Claw; 123. Joint hinge seat; 124. Wireless charging coil; 130. Robotic arm device; 131. Upper arm; 132. Forearm; 133. Lower joint; 134. Middle joint; 135. Front joint; 136. Damping pivot; 137. Damping pivot slot. Detailed Implementation

[0024] To make the technical solution of the present invention clearer and easier to implement, multiple specific embodiments are provided in conjunction with the accompanying drawings and actual application scenarios. Each embodiment covers five core scenarios: office, education, smart home, home entertainment, and mobile office. The embodiments detail the robot hardware configuration, system software configuration, workflow, and technical effects. Each embodiment is designed based on the claims of the present invention, fully covers the scope of protection, and has specific parameters that can be directly implemented.

[0025] Basic Implementation See appendix Figures 1 to 18 The interactive intelligent assistant robot provided by the present invention includes a base device, which includes a housing, a power supply unit, a sensing unit, a communication and control unit, and an interactive output unit. The power supply unit is used to charge or supply power to the external smart terminal, and to supply power to the sensing unit, communication and control unit, and interactive output unit. It includes a power supply PCB circuit board and at least one power adapter. The power supply PCB circuit board and its onboard power adapter provide counterweight for the base device and lower the overall center of gravity of the base device. The sensing unit is used to sense interactive signals and includes an infrared sensing module, an environmental sensor module, and a voice pickup module. The communication and control unit is used to receive and process signals from the sensing unit, conduct interactive communication between various units within the robot and with external networks, and control the operation of the power supply unit and interactive output unit; the communication and control unit includes an edge computing module for performing lightweight AI inference tasks locally. The interactive output unit is used to output one or more of the following: mechanical action, light, audio, and image, under the triggering of an interactive signal or the control of a communication and control unit. It can also coordinate with an external smart terminal to output one or more of the following: mechanical action, light, audio, and image, which change in coordination with each other. The power supply unit, sensing unit, communication and control unit, and interactive output unit are electrically connected to each other.

[0026] Preferably, the power supply unit includes a charging control module, multiple charging output interfaces or wireless charging coils, and multiple power adapters respectively disposed at different positions on the power supply PCB circuit board. The multiple power adapters work and dissipate heat at different positions, and use their different positions and weights to provide counterweight for the robot. At the same time, under the control of the communication and control unit, they work together to charge the external smart terminal. The charging control module supports an intelligent power distribution algorithm, which dynamically adjusts the power distribution of each output port according to the charging protocol of the access device, the battery status, and the user priority.

[0027] Preferably, the housing of the base device includes a metal heat dissipation upper shell, a plastic middle frame, and a metal heat dissipation lower shell that are nested and snapped together. The metal heat dissipation upper shell opens downward from the top, and the metal heat dissipation lower shell opens upward from the bottom, respectively nesting and snapping together with the plastic middle frame to form a three-section spliced ​​housing as a whole, forming a closed internal housing space. The power supply PCB circuit board is horizontally arranged inside the housing. The inner surfaces of the metal heat dissipation upper shell and the metal heat dissipation lower shell are provided with heat dissipation fin structures, which are thermally connected to the heat-generating elements on the power supply PCB circuit board through thermally conductive silicone pads.

[0028] Preferably, the interactive output unit includes a mechanical motion output module, a light output module, an image output module, and an audio output module; The mechanical motion output module includes a lower joint drive mechanism and a rotating shaft assembly mounted on the power supply PCB circuit board. The lower joint drive mechanism includes a motor and a reduction gear set. The rotating shaft assembly includes a lower joint damping rotating shaft, a lower damping rotating shaft holder, and a lower joint support seat. The lower damping rotating shaft holder is mounted on the lower joint damping rotating shaft, and the lower joint damping rotating shaft is mounted on the output gear of the reduction gear set. Both of them pass through the rectangular window provided in the metal heat sink upper shell and extend to the top of the metal heat sink upper shell. The lower joint support seat is mounted on the lower damping rotating shaft holder, with a gap between it and the metal heat sink upper shell, and rotates with the lower joint damping rotating shaft. The mechanical motion output module also includes a robotic arm device mounted on the lower damping shaft bracket and a charging tray device mounted at the front end of the robotic arm device, both of which rotate with the lower joint damping shaft.

[0029] Preferably, the light output module includes multiple LED beads surrounding the power supply PCB circuit board; the plastic frame is a transparent or semi-transparent material component, and has multiple transparent light guide strips, light-transmitting gaps, and light-transmitting windows below the fastening line with the upper metal heat sink and above the fastening line with the lower metal heat sink, for exporting or transmitting the light emitted by the multiple LED beads on the power supply PCB circuit board to create ambient lighting effects; the LED beads are RGB full-color LED beads, supporting 16 million colors, and can realize various lighting effects such as breathing light, flowing light, and music rhythm under the control of the communication and control unit.

[0030] Preferably, the image output module includes an LED display screen or a micro-projection module. A slope is provided on the front surface of the plastic frame, and a window is opened on the slope. The LED display screen is disposed in the window. The driving module corresponding to the LED display screen is disposed on the power supply PCB circuit board and is used to output robot operation information through the image displayed thereon, or output other changing image signals corresponding to trigger signals and control signals. The LED display screen is a touch screen that supports direct touch operation by the user.

[0031] Preferably, the audio output module includes a speaker mounted on a PCB circuit board or inside a robotic arm device, and a bass enhancement cavity mounted inside a base device, for emitting an audio signal whose output changes in accordance with the trigger signal and control signal; the audio output module supports directional sound wave output, which concentrates the sound into a specific direction through beamforming technology.

[0032] Preferably, the communication and control unit includes a communication and control circuit board, which is vertically mounted on the power supply PCB circuit board on the side close to the LED display screen. The communication and control circuit board is provided with a communication module and a computing module, which are electrically connected to each component of the power supply unit, the sensing unit and the interactive output unit, respectively, and control the operation of the power supply unit and the interactive output unit according to the sensing unit or external network signals. The communication module includes at least one of a WiFi module, a Bluetooth module, a Zigbee module, and an NFC module, and supports concurrent communication using multiple protocols. The computing module includes a local AI inference chip for performing neural network inference calculations at the edge.

[0033] Preferably, the infrared sensing module of the sensing unit includes a pyroelectric infrared sensor, which is obliquely disposed on the power supply PCB circuit board and inside the LED display screen. The upper part of its rear end face contacts the communication and control circuit board, and the lower part of its rear end face contacts the power supply PCB circuit board, for collecting infrared signals in the area above the front of the LED display screen. The sensing unit also includes an environmental sensor module, which includes a temperature sensor, a humidity sensor, an ambient light sensor, an air quality sensor, and a barometric pressure sensor, respectively mounted on a power supply PCB circuit board or a communication and control circuit board. The voice pickup module includes a microphone array that supports far-field voice recognition and sound source localization.

[0034] Preferably, the wireless charging coil of the power supply unit is disposed on the charging tray device; The charging tray device includes: a charging tray, a claw, and a joint hinge seat; the wireless charging coil is disposed inside the charging tray, near the geometric center, for wirelessly charging a smart terminal placed on the charging tray; the wireless charging coil supports the Qi standard, the PMA standard, and proprietary fast charging protocols. The articulated joint is located on the back of the charging tray, and the claws are retractably located on both sides of the charging tray for fixing smart terminals of different sizes.

[0035] Preferably, the robotic arm device includes a large arm and a forearm that are hinged together; each of the large arm and forearm includes two symmetrical U-shaped shells, which are hollow inside after being interlocked; a lower joint is provided in the lower section of the large arm shell, and the large arm is connected to the lower joint support seat and the shaft assembly of the base device through the lower damping pivot of the lower joint; the large arm and forearm are connected by a middle joint; a front joint is provided in the front section of the forearm; the charging tray device can be connected to the front joint or the lower joint respectively through the joint hinge seat on its back, and the whole moves with the front joint or the lower joint after connection.

[0036] The present invention also provides an interactive system based on an AI large model, including the above-mentioned interactive intelligent assistant robot, and also including an intelligent terminal and a cloud server. The interactive intelligent assistant robot, the intelligent terminal and the cloud server are connected and communicate through wired or wireless networks. The interactive intelligent assistant robot is used to provide physical support, charging or power supply for the intelligent terminal, establish local communication with the intelligent terminal, and then use its own sensing unit or the communication and control unit to call the sensing interface of the intelligent terminal to perceive interactive information; then send the perceived interactive information to the local communication and control unit or the AI ​​large model program of the cloud server for processing and calculation, and finally convert the returned control information into intelligent charging, power supply or interactive output information and execute it. The intelligent terminal is used to receive physical support, charging or power supply from the interactive intelligent assistant robot, and to establish communication with the interactive intelligent assistant robot or cloud server. It uses its own sensing, computing, storage and output interfaces to work in collaboration with the interactive intelligent assistant robot or cloud server to provide input data for the AI ​​large model program, or to output output information on site, including collaborative output with the interactive intelligent assistant robot. The cloud server has a built-in AI large model program, which is used to establish communication with interactive intelligent assistant robots or smart terminals, receive their input data, analyze and process it, and output intelligent charging, power supply, interactive output, and collaborative execution information. These information can be executed individually or collaboratively by the interactive intelligent assistant robot and the smart terminal to complete various interactive intelligent assistant tasks.

[0037] Preferably, the AI ​​large model program includes: A multimodal perception fusion model is used to process sensor data from interactive intelligent assistant robots and smart terminals, including visual data, audio data, environmental data, and user behavior data, to perform cross-modal feature extraction and fusion analysis. The user intent understanding model, based on the Transformer architecture, is used to identify users' natural language commands, gesture commands, and implicit needs, and generate structured user intent representations. The charging strategy optimization model is used to identify the charging and power supply needs of smart terminals and generate the optimal charging strategy based on device type, battery health status, user habits and current grid load. The interactive output decision model is used to determine what kind of interactive output to trigger based on user intent, current scene, and historical interaction records, including combinations of mechanical actions, lighting effects, audio responses, and image displays.

[0038] Preferably, the multimodal perception fusion model adopts a cross-modal encoder architecture based on Vision Transformer and BERT, mapping image features and text features to a unified semantic space; the user intent understanding model adopts a large language model based on GPT architecture, supporting context-aware dialogue understanding and multi-turn interaction management; the charging strategy optimization model adopts a deep reinforcement learning algorithm, with multiple optimization objectives of extending battery life, shortening charging time, and reducing energy consumption; and the interaction output decision model adopts a personalized recommendation algorithm based on user profiles, generating customized interactive experiences for different users.

[0039] The on-site interaction method based on an AI large model provided by this invention employs the aforementioned AI large model-based interaction system and interactive intelligent assistant robot, and includes the following steps: S1. Interactive Sensing Steps: The interactive intelligent assistant robot establishes a connection with the intelligent terminal and collects dynamic interactive sensing data of the user on site through its own sensing unit, including changes in human infrared signals, changes in environmental parameters, voice commands and gestures. The collected interactive sensing data is then sent to the communication and control unit for preprocessing. S2, Intelligent Analysis Step: The communication and control unit sends the pre-processed interactive sensing data and information from the smart terminal to the cloud server. The AI ​​large model program analyzes the user's interactive needs and determines the optimal charging or power supply strategy to achieve the optimization goals of intelligent matching of user needs, power supply unit power consumption optimization or heat generation control, and generates charging, power supply and interactive output schemes. S3. Execution and Interaction Output Steps: The interactive intelligent assistant robot executes the aforementioned charging or power supply scheme; the interactive intelligent assistant robot and the intelligent terminal execute the aforementioned interaction output scheme separately or in concert, outputting one or more of the following on-site: coordinated and changing mechanical movements, light, audio, and images; S4. Model Optimization Steps: The AI ​​large model program automatically memorizes and learns the human users, usage scenarios, smart terminals, user selection habits, user evaluations and feedback served by each interactive intelligent assistant robot, automatically upgrades and optimizes the AI ​​large model, and pre-generates more recommendation schemes to meet the diverse needs of users in work, study, companionship and entertainment.

[0040] Preferably, in the S1 interactive sensing step, the preprocessing includes: filtering and denoising the infrared sensing data, calibrating the environmental sensor data, performing endpoint detection and noise suppression on the voice data, and performing time synchronization and fusion on the multi-source sensing data.

[0041] Preferably, the S2 intelligent analysis step includes: User identification sub-step: Identify the current user's identity based on facial recognition or voiceprint recognition technology; Scene understanding sub-step: Analyze the current scene type based on the multimodal perception fusion model, including work scene, learning scene, rest scene or entertainment scene; Demand forecasting sub-step: Based on users' historical behavior data and the current scenario, predict users' possible interaction needs; Strategy generation sub-step: Based on the charging strategy optimization model and the interactive output decision model, generate multiple candidate schemes and calculate the comprehensive score of each scheme.

[0042] Preferably, in the strategy generation sub-step, the calculation of the comprehensive score considers the following factors: user preference matching degree, charging efficiency, energy consumption level, device heat generation, interaction response timeliness and user historical satisfaction feedback, and each factor is assigned a different weight coefficient according to the priority set by the user.

[0043] Preferably, in the S3 execution and interactive output step, the interactive output scheme includes: Mechanical motion output: controls the robotic arm device to perform specific posture changes or motion sequences; Lighting effect output: Controls LED beads to display specific colors, brightness changes, or dynamic lighting effects; Audio output: Play specific voice prompts, music, or sound effects; Image output: Displaying specific images, animations, or video content on an LED display screen; Collaborative output: The interactive intelligent assistant robot and the intelligent terminal execute coordinated output actions in sync to form a unified interactive experience.

[0044] Preferably, the collaborative output includes: when a user is making a video call, the interactive intelligent assistant robot automatically adjusts the posture of its robotic arm to make the camera point at the user at the best angle according to the content of the call, while adjusting the ambient lighting atmosphere according to the emotion analysis results of the other party in the call, and automatically generating a call summary at the end of the call.

[0045] Preferably, the S4 model optimization step includes: Data collection: Collect user interaction data, including sensor data, user commands, system responses, and user feedback; Feature engineering: Extracting user behavior features, scenario features, and interaction effect features from collected data; Model fine-tuning: Incremental learning or transfer learning of large AI model programs based on collected data; Personalized adaptation: Generate personalized model parameters or prompt word templates for each user; Performance evaluation: The optimization effect of the model is evaluated through A / B testing to ensure that the performance of the optimized model is improved.

[0046] Preferably, the system also includes a user confirmation step: after the S2 intelligent analysis step generates a solution, the candidate solutions are displayed to the user through a smart terminal, and the user can select or confirm them; if the smart terminal does not make a selection within a set time, the AI ​​big data model automatically selects the optimal solution with the highest score and executes it; after execution, the user is prompted to evaluate the above solution and provide feedback through an interactive intelligent assistant robot or smart terminal.

[0047] Preferably, it also includes a safety monitoring step: real-time monitoring of the working status of the interactive intelligent assistant robot, including temperature, current, voltage and mechanical component status; when an abnormality is detected, automatic safety protection measures are implemented, including reducing charging power, stopping mechanical movements, issuing an alarm to notify the user and reporting fault information to the cloud server.

[0048] Preferably, it also includes an offline working mode: when the network connection between the interactive intelligent assistant robot and the cloud server is interrupted, it automatically switches to the offline working mode, and the local edge computing module performs lightweight AI inference tasks, including basic user intent recognition, preset interactive responses and security protection functions, and synchronizes the data during the offline period to the cloud server for model updates after the network is restored.

[0049] The following describes several embodiments in more detail.

[0050] Example 1 See Figures 1 to 11 The interactive intelligent assistant robot 100 provided in this embodiment includes a base device 110 and a charging tray device 120, with the charging tray device 120 connected to the base device 110 via a joint hinge. The base device 110 includes a housing 111, a power supply unit 112, a sensing unit 113, a communication and control unit 114, and an interactive output unit 115. The housing 111 includes a nested and interlocking upper metal heat-dissipating shell 111a, a plastic middle frame 111b, and a lower metal heat-dissipating shell 111c. The upper metal heat-dissipating shell 111a opens downwards from the top, and the lower metal heat-dissipating shell 111c opens upwards from the bottom, nesting and interlocking with the plastic middle frame 111b to form a three-section spliced ​​housing assembly. The upper metal heat-dissipating shell 111a and the lower metal heat-dissipating shell 111c are made of aluminum alloy, with heat dissipation fins on their inner surfaces.

[0051] The shell adopts a three-section splicing structure. The upper metal heat dissipation shell 111a is made of 6061 aluminum alloy with a wall thickness of 2.0mm and heat dissipation fins on the inner surface. The lower metal heat dissipation shell 111c uses the same material and structure, forming a symmetrical heat dissipation structure with the upper metal heat dissipation shell 111a. The plastic middle frame 111b is made of PC+ABS transparent material with a light transmittance of ≥85%, a thickness of 1.5mm, and a fitting gap of ≤0.2mm with the metal shell.

[0052] The power supply unit 112 includes a power supply PCB circuit board 112a and two power adapters 112b (the one closer to the communication and control circuit board 114a is adapter A, and the other is adapter B). The power supply PCB circuit board 112a is horizontally arranged inside the housing 111. The two power adapters 112b are respectively located on the left and right sides of the power supply PCB circuit board 112a, using their weight to provide counterweight for the robot and lower the overall center of gravity of the base device. The power supply unit 112 also includes multiple USB charging output ports and a wireless charging coil.

[0053] The power supply unit is configured as follows: the power supply PCB circuit board 112a has a size of 100mm×80mm×1.6mm, uses FR-4 material, and has a four-layer board design.

[0054] The power adapter is configured as follows: Adapter A: 100W PD fast charging, output specifications 5V / 3A, 9V / 3A, 12V / 3A, 15V / 3A, 20V / 5A, weight 220g, dimensions 65mm×65mm×28mm; Adapter B: 65W PD fast charging, output specifications 5V / 3A, 9V / 3A, 12V / 3A, 15V / 3A, 20V / 3.25A, weight 200g, dimensions 55mm×55mm×25mm. The two adapters are symmetrically distributed along the diagonal of the PCB, with a center of gravity offset ≤3mm.

[0055] The charging output interface is configured as follows: Type-C ports × 3: Supports PD3.1 protocol, two for 100W adapters and one for 65W adapters; USB-A ports × 2: Supports QC4.0 protocol, maximum output 18W; Wireless charging coil: Supports Qi standard, maximum power 15W, charging efficiency ≥75%; The sensing unit 113 includes an infrared sensing module 113a, an environmental sensor module 113b, and a voice pickup module 113c. The infrared sensing module 113a includes a pyroelectric infrared sensor, obliquely mounted on the power supply PCB circuit board 112a. The environmental sensor module 113b includes a temperature sensor, a humidity sensor, and an ambient light sensor. The voice pickup module 113c includes a microphone array supporting far-field voice recognition.

[0056] The charging control module uses the TI BQ25713 charging management chip and supports intelligent power allocation algorithms. The power allocation strategy is as follows: Single device access: Automatically matches maximum power according to device protocol; Multiple devices can be accessed: devices are dynamically allocated based on priority, and priority rules can be customized by the user. Total power limit: When the total demand of multiple devices exceeds 165W, the power will be reduced according to the following priority: First priority: wireless charging (maintain 5W basic power supply); Second priority: high-performance devices such as laptops (maintain a minimum of 45W); Third priority: mobile phones, tablets, etc. (the remaining power will be allocated proportionally).

[0057] The heat dissipation design parameters are: The metal heat dissipation upper shell 111a and the metal heat dissipation lower shell 111c form a thermal conduction connection with the heat-generating components (power adapter, charging management chip) on the power supply PCB circuit board 112a through thermally conductive silicone pads; the overall thermal resistance is ≤8°C / W, and the surface temperature rise of the shell is ≤25°C when the full load output is 165W.

[0058] The communication and control unit 114 includes a communication and control circuit board 114a, which is vertically mounted on the power supply PCB circuit board 112a. The communication and control circuit board 114a is equipped with a WiFi module, a Bluetooth module, and a local AI inference chip.

[0059] The interactive output unit 115 includes a light output module 115a, an image output module 115b, and an audio output module 115c. The light output module 115a includes multiple RGB LEDs arranged around a power supply PCB circuit board 112a. The image output module 115b includes an LED display screen disposed on the beveled front surface of the plastic frame 111b. The audio output module 115c includes a speaker.

[0060] The charging tray device 120 includes a charging tray 121, claws 122, and a joint hinge seat 123. A wireless charging coil is disposed within the charging tray 121. The claws 122 are retractably disposed on both sides of the charging tray 121.

[0061] When the interactive intelligent assistant robot 100 in this embodiment is working, when the infrared sensor module 113a detects a human approaching, the communication and control unit 114 controls the LED beads to emit a welcoming light effect, and the speaker plays a greeting voice. When the user places their mobile phone on the charging tray 121, the power supply unit 112 automatically identifies the device type and begins charging. The user can interact with the robot via voice. The voice commands are pre-processed locally and then sent to the cloud server for AI analysis. The returned control commands drive the robot to execute the corresponding interactive output.

[0062] Example 2 See Figures 13 to 17 Based on Embodiment 1, this embodiment further provides an interactive intelligent assistant robot 100 with a robotic arm device.

[0063] The main difference between this embodiment and Embodiment 1 is the addition of a robotic arm device 130.

[0064] like Figures 15 to 17 As shown, the robotic arm device 130 includes a large arm 131 and a small arm 132 that are hinged to each other. Both the large arm 131 and the small arm 132 include two symmetrical U-shaped shells that are hollow inside when they are interlocked.

[0065] The lower section of the upper arm 131 is provided with a lower joint 133, which is connected to the base device 110 via a lower damping pivot 136. The upper arm 131 and the forearm 132 are connected by a middle joint 134. The front section of the forearm 132 is provided with a front joint 135. The charging tray device 120 can be connected to the front joint 135 or the lower joint 133 via a joint hinge seat on its back.

[0066] The lower joint 133, middle joint 134, and front joint 135 all include a damping shaft 136 and a damping shaft slot 137. Both the damping shaft 136 and the damping shaft slot 137 are insulating components. A metal shell is fitted onto the damping shaft 136, and a metal sleeve is provided inside the damping shaft slot 137. When they are engaged, they form a metal electrical connection terminal for transmitting electrical energy from the power supply unit to the robotic arm device and the charging tray device.

[0067] The U-shaped housing of the robotic arm device 130 contains multiple LED beads and speakers, which can create rich lighting and sound effects.

[0068] The robotic arm device in this embodiment has a two-section hinged robotic arm structure, consisting of a large arm and a small arm. It adopts a symmetrical U-shaped shell snap-fit ​​structure, with a hollow interior for wiring and electrical component placement. The joint configuration is as follows: lower joint (connecting the large arm to the base), middle joint (connecting the large arm to the small arm), and front joint (connecting the small arm to the charging tray), achieving four degrees of freedom of motion. The connection method is as follows: each joint is connected by a damping pivot and a damping pivot slot. The robotic arm and the base / charging tray are detachably snap-fitted together, supporting direct connection of the charging tray to the lower joint of the base (for compact scenarios). It adopts a modular design: the hinge interfaces of the large arm and small arm / damping pivot / electrical connection terminals are universal, and the two can be interchanged for assembly, enabling flexible switching of the joint layout.

[0069] In office / general scenarios, the main specifications and core dimensional parameters include: (a) Shell Dimensions Main arm (ABS+PC alloy material): Total length 150mm, U-shaped housing opening width 28mm, housing height 25mm, overall width after fastening 30mm, wall thickness 2.5mm, internal hollow cable routing cavity size 10mm×8mm; Forearm (ABS+PC alloy material): Total length 120mm, U-shaped shell opening width 25mm, shell height 22mm, overall width after fastening 27mm, wall thickness 2.5mm, internal hollow wiring cavity size 8mm×7mm; Compact size (bedside / small desktop scenario): upper arm 120mm, forearm 100mm, other shell proportions are the same as the main size.

[0070] (ii) Joint installation dimensions Lower joint: The center distance of the pivot is 20mm from the bottom of the upper arm, the pivot diameter is 8mm, and the damping groove fit depth is 15mm; Middle joint: The center of the pivot is 18mm from the top of the upper arm and 18mm from the bottom of the forearm; the pivot diameter is 6mm; and the damping groove fit depth is 12mm. Front joint: The center of the pivot is 15mm from the top of the forearm, the pivot diameter is 6mm, and the damping groove fit depth is 12mm; Joint hinge spacing: The shell gap at each joint hinge is 0.3mm, with no interference movement margin.

[0071] (iii) Adapting connection size Connection with the base: The lower joint damping shaft slot is adapted to the lower joint drive shaft slot of the base, with a snap-fit ​​tolerance of ±0.1mm; Connection with charging tray: The front joint damping pivot is adapted to the charging tray joint hinge seat slot, with a snap-fit ​​tolerance of ±0.1mm; Charging tray mounting position: The center of the front joint pivot is 40mm away from the geometric center of the charging tray, which is compatible with 5.4-7.2 inch smart terminals.

[0072] The interactive intelligent assistant robot 100 in this embodiment can achieve more flexible interactive output through the robotic arm device 130. For example, during a video call, the robotic arm can automatically adjust its angle to point the camera at the user; when playing music, the robotic arm can swing in rhythm, and the LED beads will flash synchronously.

[0073] Example 3 This embodiment, based on Embodiments 1 and 2, applies the interactive intelligent assistant robot to enterprise office and personal desktop office scenarios, focusing on multi-device intelligent charging, video conferencing assistance, and AI office assistant functions to achieve intelligent and efficient office scenarios. Its differences from Embodiments 1 and 2 are as follows: 1. Hardware Configuration Robot base assembly: Onboard 100W+65W dual power adapters, diagonal weights converge the center of gravity to the geometric center of the base; the power supply PCB has 3 Type-C ports (PD3.1 / 3.0) and 2 USB ports (QC4.1 / 4.0), supporting simultaneous charging of multiple devices such as laptops, mobile phones, and wireless headphones; the communication and control unit is equipped with an ESP32-S3 main control board + K210 edge computing module, integrating WiFi 6 and Bluetooth 5.0 modules; the sensing unit includes a PIR2020 pyroelectric infrared sensor (detection 10-40cm) and a microphone array (5-meter far-field voice); the interactive output unit includes a 1.28-inch touch LCD screen, RGB full-color LEDs, directional speakers, and a four-degree-of-freedom robotic arm (150mm upper arm and 120mm lower arm).

[0074] Charging tray device: Qi standard wireless charging coil (5W / 7.5W / 10W / 15W), retractable claws to adapt to 5.4-7.2 inch mobile phones / tablets, maximum load capacity 300g.

[0075] Smart terminals: office laptops (requires PD fast charging), smartphones, wireless earphones, and conference cameras.

[0076] 2. System Software Configuration Local edge computing: The K210 module is equipped with a lightweight office intent recognition model and a speech-to-text model to achieve real-time local voice command recognition and real-time transcription of meeting speech.

[0077] Large-scale AI model for cloud servers: User intent understanding model: A fine-tuned model for the office domain based on the GPT-3.5 architecture, supporting the understanding of office instructions such as "meeting minutes", "schedule reminders", and "document organization"; Charging strategy optimization model: Prioritizing fast charging in office scenarios while taking into account battery life, dynamically adjusting power according to the charging protocols of laptops / mobile phones; Interactive output decision model: A solution specifically designed for office scenarios, with low-brightness cool white lighting (4000K) and directional, soft audio output to avoid interference from the office environment.

[0078] Third-party service interfaces: Integrate with enterprise office automation (OA) systems, calendar systems, and cloud document systems.

[0079] 3. Work Process Device connection: The robot connects to AC220V power supply, automatically connects to the enterprise 5G WiFi, and pairs with office laptops and mobile phones via Bluetooth to complete system networking; Interactive sensing: an infrared sensor detects when a user sits down (10-40cm), automatically waking up the robot; a microphone array collects the user's voice commands; and an ambient light sensor detects the brightness of the office environment and adjusts the screen brightness accordingly. Intelligent Analysis: If the user's instruction is "Start video conference", the cloud server's AI big model recognizes the intent and generates a conference assistance plan: the charging strategy is 100W fast charging for the laptop and 15W wireless charging for the mobile phone; the interactive output is that the robotic arm adjusts the mobile phone to an angle level with the monitor (300-350mm from the table, tilt angle 75°) to serve as a secondary camera for the conference. Execution and Collaborative Output: The robot executes the charging strategy, the robotic arm adjusts to a preset angle, the directional speaker synchronizes with the audio of the mobile conference, and the LCD screen displays the conference duration and speech-to-text content in real time; the lighting remains low-brightness and cool white, without flickering interference; After the meeting: The AI ​​big data model automatically generates a meeting summary and sends it to the user's mobile phone / office OA system; the robot prompts the user to evaluate the solution, and at the same time checks the charging device status. If the laptop battery is full, it automatically switches to trickle charging to protect the battery. Model optimization: The cloud server collects data such as user reviews, meeting duration, and charging parameters to incrementally learn the office intent understanding model, thereby improving the accuracy of instruction recognition.

[0080] 4. Technical performance of the prototype in actual testing It can charge or power up 4 devices simultaneously (laptop, mobile phone, wireless headset, conference camera), with a total power of 165W, significantly improving charging efficiency; The robotic arm's four-degree-of-freedom adjustment enables optimal angle positioning of the conference camera, preventing users from looking down at their phones and reducing neck fatigue; Local speech-to-text transcription and cloud-based meeting summaries can significantly improve office efficiency; Directional audio and low-brightness lighting enable interference-free interaction in office scenarios, adapting to the needs of the office environment; With its integrated counterweight and power supply design, the robot can tilt up to 15° without the risk of tipping over, meeting the operational stability requirements of office desktops.

[0081] Example 4 This embodiment builds upon embodiments 1 to 3, further applying it to the scenario of children's educational companionship. This embodiment is suitable for family-based children's learning and education scenarios, focusing on personalized educational content recommendations, emotional support, and vision protection functions, thereby making children's learning intelligent and engaging.

[0082] like Figure 18As shown, the interactive intelligent assistant robot 100, tablet computer 200 (as an intelligent terminal), and cloud server 300 in this embodiment constitute an educational companion system.

[0083] 1. Hardware Configuration Robot base device: Based on embodiment 1, the speaker is optimized to be child-friendly, and the LED beads are equipped with an eye protection mode (color temperature 3000K, no blue light); the infrared sensor is equipped with a distance monitoring function, which triggers an alert when the child is less than 30cm away from the screen; the robotic arm is made of colored plastic shell to improve children's acceptance.

[0084] Smart devices: Children's learning tablets (with educational apps), children's smartwatches.

[0085] Other features: The charging tray is equipped with an anti-slip mat and a child safety lock to prevent children from accidentally touching it.

[0086] 2. System Software Configuration Local edge computing: The K210 module is equipped with a child behavior detection model to monitor children's sitting posture and screen viewing distance in real time, enabling local quick reminders.

[0087] Large-scale AI model for cloud servers: Multimodal perception fusion model: Integrates visual data from tablets and infrared / voice data from robots to analyze children's learning focus; User Intent Understanding Model: A child-friendly, fine-tuned model based on the GPT-4 architecture, using simple language dialogue and supporting storytelling and knowledge point explanation; Educational content recommendation model: A K-12 education domain model based on knowledge graphs, recommending learning content according to children's age, learning progress, and interests; Interactive output decision model: A solution specifically designed for children's scenarios, featuring soft, warm lighting, a robotic arm that can perform fun actions such as "clapping" and "waving," and cartoon-style audio.

[0088] Third-party service interfaces: Connect to children's education platforms, audio story platforms, and children's educational game platforms.

[0089] 3. Work Process Identity recognition: The robot confirms the child's identity through voiceprint recognition and automatically retrieves the child's learning file (age, grade, learning progress). Interactive sensing: When an infrared sensor detects that a child is sitting at the study table, the robot is automatically woken up, and the microphone array collects the child's voice commands (such as "I want to learn English"). Intelligent Analysis: The cloud server AI model, combined with the child's learning profile, generates an English learning plan; the charging strategy provides the tablet with 65W power (continuous learning without power interruption); the educational content recommended is Oxford Reading Tree leveled reading; the interactive output is a robotic arm that adjusts the tablet to an eye-protection angle (250mm from the desktop, 60° tilt), and the light switches to a warm white eye-protection mode. Execution and Collaborative Output: The robot continuously powers the tablet, adjusts the robotic arm to a preset angle, and displays a list of English learning content on the LCD screen. The robot guides children's learning through cartoon-style voice. During the learning process, the child's behavior detection model monitors in real time. If the child's posture is incorrect or too close to the screen, the LCD screen displays a reminder animation, the light flashes yellow, and the robotic arm gently swings to remind them. After the learning is completed: the robotic arm makes a "clapping" motion, the lights flash colored lights, and it plays encouraging voice messages; a learning report (learning time, concentration level, and mastery of knowledge points) is automatically generated and sent to the parent's mobile phone; Model optimization: Cloud servers collect children's learning data and parent feedback to fine-tune the educational content recommendation model and improve the accuracy of content recommendations.

[0090] 4. Technical effects of prototype testing Personalized educational content recommendations can significantly improve children's learning interest and efficiency. Real-time posture / distance monitoring effectively protects children's eyesight and reduces the risk of myopia; Through engaging and interactive activities that foster emotional connection, we can reduce children's feelings of loneliness while learning and enhance their initiative in learning. Parents can remotely view learning reports, enabling remote monitoring of their children's learning.

[0091] Example 5 This embodiment, based on embodiments 1 to 4, is applied to smart home scenarios such as living rooms and entryways, focusing on human body tracking, whole-house intelligent control, and scenario-based interactive functions to achieve deep integration between the robot and smart home devices. The specific solution is as follows: 1. Hardware Configuration Robot base device: The communication and control unit adds a Zigbee module to achieve wireless connection with smart home devices; the sensing unit adds an air quality sensor (PM2.5, formaldehyde) and a human presence sensor to achieve whole-house environmental monitoring and human tracking; the interactive output unit adds a micro projection module to project interactive information onto the wall / desktop.

[0092] Smart terminals: home smart control screens, smartphones.

[0093] Smart home devices, including smart lights, smart air conditioners, smart air purifiers, and smart curtains, all support the Zigbee protocol.

[0094] 2. System Software Configuration Local edge computing: The ESP32-S3 module is equipped with a smart home device control model, enabling real-time local control of smart lights, air conditioners and other devices without relying on cloud servers.

[0095] Large-scale AI model for cloud servers: Multimodal perception fusion model: integrates environmental / human data from robots and operational data from smart home devices to analyze the status of home scenarios; User intent understanding model: A fine-tuned smart home model based on the GPT-3.5 architecture, supporting the understanding of scene commands such as "home mode", "sleep mode" and "movie mode"; Interactive Output Decision Model: A solution specifically designed for smart home scenarios, enabling collaborative output between robots and smart home devices, such as turning off robot lights, closing smart curtains, and dimming smart lights during movie viewing mode.

[0096] Third-party service interfaces: Connect to smart home control systems, weather platforms, and security platforms.

[0097] 3. Work Process Scene trigger: The robot detects the user's return home (entrance area) through the human presence sensor and automatically triggers the home mode; Interactive sensing: Infrared sensors track user movement, environmental sensors detect indoor temperature and air quality, and microphone arrays collect user voice commands; Intelligent Analysis: The cloud server's AI big model combines environmental data and user habits to generate a "coming home" mode solution: the charging strategy is to wirelessly charge the user's mobile phone at 15W; smart home control is to turn on smart lights (warm color tone), adjust the smart air conditioner to 26℃, and activate the smart air purifier if PM2.5 levels exceed the standard; the interactive output is that a robotic arm adjusts the mobile phone to the angle of the entryway table, and the LCD screen displays indoor environmental data and weather information; Execution and Collaborative Output: The robot executes the charging strategy, controls smart home devices to start corresponding modes via the Zigbee module, and projects weather information onto the entryway wall via the micro-projection module; if the user command is "turn on movie viewing mode", the robot immediately controls the smart curtains to close, the smart lights to dim, its own lights to turn off, and the directional speakers to synchronize with the TV audio. Security monitoring: The robot monitors the operating status of smart home devices in real time. If the smart air conditioner malfunctions, it immediately sends an alarm to the user's mobile phone and displays the fault information on the LCD screen. Model optimization: The cloud server collects users' usage habits and environmental data to optimize the scene intent recognition model and improve the adaptability of scene modes.

[0098] 4. Technical effects of prototype testing It enables deep integration between robots and smart home devices, allowing for one-click triggering of scene modes and enhancing the intelligence and convenience of family life; The human body tracking function enables scene adaptation after the user moves, such as when the user moves from the entrance hall to the living room, the smart lights will automatically turn on to follow and turn on. Local edge computing enables real-time control of smart home devices with a response time of < 0.5 seconds; Whole-house environmental monitoring enables real-time display and intelligent adjustment of air quality, temperature, and humidity, enhancing the comfort of home life; Robots serve as the central control terminal for smart homes, replacing traditional smart control screens and integrating charging, interaction, and control.

[0099] Example 6 This embodiment, based on embodiments 1 to 5, is applied to scenarios such as bedside tables and bedrooms in the home. It focuses on fine-tuning of angles, low power consumption at night, and creating an immersive atmosphere, catering to users' needs for studying, watching movies, and entertainment before bed. Specifically: 1. Hardware Configuration Robot base device: The drive motor adopts a low-noise SG90 micro servo motor (operating noise < 30dB) to avoid nighttime interference; the LED beads support stepless color temperature adjustment (2700K-6500K) and fine brightness adjustment (1%-100%); the communication and control unit adds a low power consumption mode, with nighttime standby power consumption < 1W.

[0100] Robotic arm device: It adopts a compact design with a 120mm upper arm and a 100mm lower arm, and can be folded up above the base to reduce the space occupied by the bedside.

[0101] Smart terminals: mobile phones, small tablets (7.9 inches and below).

[0102] 2. System Software Configuration Local edge computing: The K210 module is equipped with a night mode recognition model, which automatically switches to a low-power night mode based on the ambient light intensity.

[0103] Large-scale AI model for cloud servers: Charging strategy optimization model: With the goal of nighttime trickle charging, the battery is protected first, and when the phone battery reaches 80%, it automatically switches to 5W trickle charging; Interactive output decision model: A bedside scene-specific solution that supports synchronization of lighting and audio / video, such as lighting changing with the main color tone of the video when watching a movie, and lighting moving in rhythm with the music when listening to music.

[0104] Third-party service interfaces: Connect to video platforms, music platforms, and audiobook platforms.

[0105] 3. Work Process Night mode triggered: When the ambient light sensor detects indoor light level < 50 lux, the robot automatically switches to night low-power mode, the LCD screen brightness is adjusted to 10%, and the speaker volume is limited to 30dB. Interactive sensing: The user gives the voice command "I want to watch a movie", the microphone array collects the command, and the infrared sensor detects the user's lying position; Intelligent Analysis: Cloud server AI large model generates viewing solutions: the charging strategy is 10W wireless charging for mobile phones; the interactive output is that the robotic arm adjusts the mobile phone to the optimal lying angle (tilt angle 45°-60°), the lighting switches to "video synchronization mode", and the directional speaker synchronizes with the mobile phone audio; Execution and Collaborative Output: When the robot executes the charging strategy, the robotic arm adjusts to a preset angle, and the lights change in real time with the main color tone of the video screen to create an immersive viewing experience; if the user is watching an audiobook, the lights switch to "breathing light mode" (alternating blue and green), and the robotic arm remains still to avoid interference; Bedtime study scenario: When the user commands "I want to study", the robot switches the light to a warm white eye-protection mode (3000K), the robotic arm adjusts the tablet to the study angle, and at the same time continuously supplies power to the tablet to avoid interruption of study; Automatic sleep mode: If the infrared sensor does not detect user operation for 5 consecutive minutes, the robot will automatically turn off the lights, fold the robotic arm to the base, and only retain trickle charging for the mobile phone, entering a deep low power mode.

[0106] 4. Prototype testing technology effectiveness Low-noise motor (<30dB) + low-power mode (standby < 1W), perfectly suited for bedside nighttime use, with no noise / light interference; The robotic arm features precise four-degree-of-freedom adjustment, achieving the optimal viewing / learning angle in a lying position, thus enhancing the user experience. The lighting is synchronized with the audio and video to create an immersive entertainment experience, replacing traditional bedside ambient lighting; The nighttime trickle charging strategy effectively protects the phone battery and extends battery life. With its compact folding design, the robot's height is less than 150mm when folded, reducing the space it occupies at the head of the bed.

[0107] The embodiments of the present invention described above are all based on the core principles of integration, intelligence, collaboration, and scenario-based application, achieving deep integration of hardware, software, and methods, and all possess the following advantages: 1. Hardware Structure: Integrated design to overcome the structural defects of traditional robots. Integrated counterweight and power supply: The power adapter is directly mounted on the power supply PCB circuit board. The weight of the adapter is used to achieve the counterweight of the base, so that the center of gravity converges to the geometric center, solving the problem of unstable center of gravity and easy tipping of traditional robots; at the same time, the two adapters are diagonally distributed to take into account heat dissipation and counterweight balance.

[0108] Integrated heat dissipation and light guiding: The three-section housing adopts a metal heat dissipation upper / lower shell + transparent plastic middle frame. The metal shell is equipped with heat dissipation fins, which form heat conduction with the heat-generating components to achieve efficient heat dissipation; the plastic middle frame is equipped with light guiding strips and light transmission windows to evenly guide the LED light, thus achieving integrated heat dissipation and light guiding.

[0109] Mechanical and electrical connections are integrated: The joints of the robotic arm adopt a damped pivot + metal shell / sleeve design, which realizes mechanical hinge and angle positioning, and forms an electrical connection channel, avoiding the exposure of external cables and realizing the integration of mechanical structure and electrical connection.

[0110] Multimodal sensing and output integration: The base device integrates multimodal sensing modules such as infrared, environment, and voice, as well as multi-dimensional output modules such as mechanical, lighting, audio, and image, to achieve hardware integration of sensing and output, thereby improving the integration level and space utilization of the device.

[0111] 2. Hardware Functionality: Multi-protocol, multi-degree-of-freedom, and multi-scenario adaptation to meet diverse needs. Multi-protocol smart charging: Supports Qi / PMA wireless charging protocol, PD3.1 / 3.0, and QC4.1 / 4.0 wired charging protocol, compatible with most smart terminals on the market, and supports simultaneous charging of multiple devices. The total output power can be selected as needed (by replacing the appropriate adapter).

[0112] The four-degree-of-freedom robotic arm enables the lower joint to rotate horizontally from 0° to 180° and pitch from 0° to 120°, the middle joint to adjust the included angle from 0° to 120°, and the front joint to rotate from 0° to 180°, allowing the charging tray to be flexibly positioned in three-dimensional space to adapt to the angle requirements of different scenarios.

[0113] Multimodal human-computer interaction: Supports multiple interaction methods such as voice, touch, gesture, and human infrared, realizing multimodal perception fusion and improving the convenience and accuracy of interaction.

[0114] Modular structure design: The robotic arm, charging tray and base are detachable and can be flexibly assembled, realizing the switching between "full-function intelligent assistant" and "compact charging assistant", adapting to various scenarios such as office, education, smart home, mobile office and so on.

[0115] 3. Software System: A hybrid architecture combining local and cloud-based systems, balancing real-time performance with intelligent operation. Edge computing + cloud-based large model: The robot is equipped with an edge computing module to achieve local lightweight AI inference (such as real-time speech recognition and device control) and ensure the real-time performance of interactive responses; the cloud server has a built-in multi-model collaborative AI large model program to achieve complex intelligent decision-making (such as intent understanding and strategy optimization) and ensure the intelligence of the system.

[0116] The multi-model collaborative AI big model program consists of four major models: multimodal perception fusion, user intent understanding, charging strategy optimization, and interactive output decision-making. Each model has its own division of labor and collaborates with each other, and data can be shared to achieve full-process intelligence from perception to decision-making.

[0117] Continuous model optimization mechanism: The cloud server performs incremental learning and transfer learning based on user interaction feedback data, realizing the self-optimization and self-upgrading of the AI ​​large model, becoming smarter with use and improving the long-term intelligence level of the system; it also supports multi-terminal collaborative communication.

[0118] The above are only some specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention; any substitutions and improvements made without departing from the concept of the present invention shall fall within the scope of protection of the present invention.

Claims

1. An interactive intelligent assistant robot, characterized in that, The device includes a base assembly, which comprises a housing, a power supply unit, a sensing unit, a communication and control unit, and an interactive output unit. The power supply unit is used to charge or supply power to the external smart terminal, and to supply power to the sensing unit, communication and control unit, and interactive output unit. It includes a power supply PCB circuit board and at least one power adapter. The power supply PCB circuit board and its onboard power adapter provide counterweight for the base device and lower the overall center of gravity of the base device. The sensing unit is used to sense interactive signals and includes at least an infrared sensing module. The communication and control unit is used to receive and process signals from the sensing unit, conduct interactive communication between various units within the robot and with external networks, and control the operation of the power supply unit and interactive output unit; the communication and control unit includes an edge computing module for performing lightweight AI inference tasks locally. The interactive output unit is used to output one or more of the following, namely, changing mechanical movements, light, audio, and images, according to the type of the interactive signal or control signal, under the triggering of the interactive signal or the control of the communication and control unit. The power supply unit, sensing unit, communication and control unit, and interactive output unit are electrically connected to each other.

2. The interactive intelligent assistant robot according to claim 1, characterized in that, The power supply unit includes a charging control module, multiple charging output interfaces or wireless charging coils, and multiple power adapters respectively set at different positions on the power supply PCB circuit board. The multiple power adapters work and dissipate heat at different positions, and use their different positions and weights to provide counterweight for the robot. At the same time, under the control of the communication and control unit, they work together to charge the external smart terminal. The charging control module supports intelligent power distribution algorithm, which dynamically adjusts the power distribution of each output port according to the charging protocol of the connected device, battery status and user priority.

3. The interactive intelligent assistant robot according to claim 1, characterized in that, The base device has a three-section splicing structure, including a metal heat dissipation upper shell, a plastic middle frame, and a metal heat dissipation lower shell that are nested and snapped together. The metal heat dissipation upper shell opens downward from the top, and the metal heat dissipation lower shell opens upward from the bottom, and is nested and snapped together with the plastic middle frame to form a three-section splicing shell as a whole, forming a closed internal housing space; the power supply PCB circuit board is horizontally arranged inside the shell.

4. The interactive intelligent assistant robot according to claim 3, characterized in that, The interactive output unit includes a mechanical motion output module, which includes a lower joint drive mechanism and a rotating shaft assembly mounted on the power supply PCB circuit board. The lower joint drive mechanism includes a motor and a reduction gear set. The rotating shaft assembly includes a lower joint damping rotating shaft, a lower damping rotating shaft holder, and a lower joint support seat. The lower damping rotating shaft holder is mounted on the lower joint damping rotating shaft, and the lower joint damping rotating shaft is mounted on the output gear of the reduction gear set. Both of them pass through a rectangular window provided on the metal heat sink upper shell and extend to the top of the metal heat sink upper shell. The lower joint support seat is mounted on the lower damping rotating shaft holder, with a gap between it and the metal heat sink upper shell, and rotates with the lower joint damping rotating shaft.

5. The interactive intelligent assistant robot according to claim 4, characterized in that, The mechanical motion output module also includes a robotic arm device mounted on the lower damping shaft holder and a charging tray device mounted at the front end of the robotic arm device, both of which rotate with the lower joint damping shaft; the robotic arm device is a multi-joint structure that can achieve four degrees of freedom adjustment: the lower joint can rotate horizontally from 0° to 180° and pitch from 0° to 120°, the middle joint can adjust the included angle from 0° to 120°, and the front joint can rotate from 0° to 180°.

6. The interactive intelligent assistant robot according to claim 4, characterized in that, The interactive output unit also includes a light output module; the light output module includes multiple LED beads surrounding the power supply PCB circuit board. The LED beads are RGB full-color LED beads, which can realize various lighting effects such as breathing light, flowing light, and music rhythm under the control of the communication and control unit. The plastic frame is a transparent or semi-transparent material component. Below the fastening line with the metal heat sink upper shell and above the fastening line with the metal heat sink lower shell, there are multiple transparent light guide strips, light transmission gaps, and light transmission windows, which are used to guide or transmit the light emitted by the multiple LED beads on the power supply PCB circuit board to create ambient lighting effects.

7. The interactive intelligent assistant robot according to claim 4, characterized in that, The interactive output unit also includes an image output module, which includes a touch-sensitive LED display / micro-projection module for displaying robot operation information and interactive images; A bevel is provided on the front end face of the plastic frame, and a window is opened on the bevel. The LED display screen is set in the window. The driving module corresponding to the LED display screen is set on the power supply PCB circuit board. It is used to output robot operation information through the image screen displayed on it, or output other image signals that change in accordance with trigger signals and control signals. The LED display screen is a touch screen that supports direct touch operation by the user.

8. The interactive intelligent assistant robot according to claim 4, characterized in that, The interactive output unit also includes an audio output module, which includes a speaker mounted on the PCB circuit board or in the robotic arm device, and a bass enhancement cavity mounted in the base device, for emitting audio signals that change output in accordance with trigger signals and control signals; the audio output module supports directional sound wave output, which uses beamforming technology to focus sound onto a specific direction.

9. The interactive intelligent assistant robot according to claim 5, characterized in that, The charging tray device includes a charging tray, a retractable claw, a joint hinge base, and a wireless charging coil. The wireless charging coil is located at the geometric center of the charging tray and supports Qi standard, PMA standard, and proprietary fast charging protocols. The claw is compatible with smart terminals of various sizes. The wireless charging coil can also be installed on a robotic arm device or a base device to achieve flexible switching of charging positions.

10. The interactive intelligent assistant robot according to claim 5, characterized in that, The robotic arm device includes a large arm and a small arm that are hinged together. Both the upper arm and the lower arm include two symmetrical U-shaped shells, which are hollow inside after being interlocked. A lower joint is provided in the lower section of the upper arm shell, and the upper arm is connected to the lower joint support seat and the shaft assembly of the base device through the lower damping shaft of the lower joint. The upper arm and forearm are connected by a middle joint; the forearm has a front joint at the front end; the charging tray device can be connected to the front joint or the lower joint respectively through the joint hinge seat on its back, and the whole moves with the front joint or the lower joint after connection.

11. The interactive intelligent assistant robot according to claim 1, characterized in that, The communication and control unit includes a communication and control circuit board, which is vertically mounted on the power supply PCB circuit board on the side close to the LED display screen. The communication and control circuit board is equipped with a communication module and a computing module, which are electrically connected to the components of the power supply unit, the sensing unit and the interactive output unit, respectively, and control the operation of the power supply unit and the interactive output unit according to the sensing unit or external network signals. The communication module includes at least one of a WiFi module, a Bluetooth module, a Zigbee module, and an NFC module, supports concurrent communication of multiple protocols, and can pair with two or more smart terminals at the same time. The computing module includes a local AI inference chip, equipped with a lightweight human detection and intent recognition model, used to perform neural network inference calculations at the edge.

12. The interactive intelligent assistant robot according to claim 11, characterized in that, The sensing unit includes an infrared sensing module, an environmental sensor module, and a voice pickup module. The infrared sensing module includes a pyroelectric infrared sensor, which is obliquely mounted on the power supply PCB circuit board and located inside the LED display screen. The upper part of its rear end face contacts the communication and control circuit board, and the lower part of its rear end face contacts the power supply PCB circuit board. It is used to collect infrared signals in the area above and in front of the LED display screen. The environmental sensor module includes a temperature sensor, a humidity sensor, an ambient light sensor, an air quality sensor, and a barometric pressure sensor, which are respectively mounted on the power supply PCB circuit board or the communication and control circuit board. The voice pickup module includes a microphone array that supports far-field voice recognition and sound source localization.

13. An interactive system based on a large AI model, characterized in that, The interactive intelligent assistant robot, as described in any one of claims 1 to 12, further includes an intelligent terminal and a cloud server, wherein the interactive intelligent assistant robot, the intelligent terminal, and the cloud server are connected and communicate via a wired or wireless network; The interactive intelligent assistant robot provides physical support, charging or power supply for the intelligent terminal. It senses interactive information through its own sensing unit or by calling the intelligent terminal's sensing interface, sends the information to its own communication and control unit or cloud server for processing, and executes control information. The intelligent terminal works in collaboration with the robot and cloud server to provide input data for the large AI model and execute collaborative outputs. The cloud server has a built-in AI large model program, which is used to analyze input data and output intelligent charging, power supply, interactive output and collaborative execution information, which are executed by the robot and the intelligent terminal alone or in collaboration.

14. The interactive system based on a large AI model according to claim 13, characterized in that, The AI ​​large-scale model program includes: A multimodal perception fusion model is used to process sensor data from interactive intelligent assistant robots and smart terminals, including visual data, audio data, environmental data, and user behavior data, to perform cross-modal feature extraction and fusion analysis. The user intent understanding model, based on the Transformer architecture, is used to identify users' natural language commands, gesture commands, and implicit needs, and generate structured user intent representations. The charging strategy optimization model is used to identify the charging and power supply needs of smart terminals. Based on device type, battery health status, user habits and current grid load, and using deep reinforcement learning algorithms, it generates the optimal charging strategy with multiple optimization objectives, such as extending battery life, shortening charging time, reducing energy consumption and controlling heat generation. The interactive output decision model is used to generate personalized mechanical action, lighting, audio, and image combinations for different users based on user intent, current scene, and historical interaction records, using a personalized recommendation algorithm based on user profiles, and preset thresholds to trigger interactive output.

15. The AI-based large-scale interactive system according to claim 13, wherein the cloud server further comprises a model training and update module, a user data management module, a device management module, and a third-party service interface module; the model training and update module performs incremental learning / transfer learning based on user interaction feedback data to achieve continuous optimization of the AI ​​large-scale model; the third-party service interface module can connect to external services such as weather queries, news information, smart home control, and online educational resources.

16. A live interaction method based on a large AI model, characterized in that, The interactive system based on an AI large model as described in any one of claims 13-15, and the interactive intelligent assistant robot as described in any one of claims 1-12, comprises the following steps: S1. Interactive sensing steps: The robot establishes a connection with the intelligent terminal and collects interactive sensing data such as human infrared, environmental parameters, voice commands, and gestures through its local sensing unit. The data is then sent to the communication and control unit for filtering, noise reduction, calibration, endpoint detection, time synchronization, and fusion preprocessing. S2, Intelligent Analysis Step: The communication and control unit sends the pre-processed data and smart terminal information to the cloud server, where the AI ​​large model program completes user identification, scene understanding, demand prediction and strategy generation, generating candidate solutions for charging or power supply and interactive output. S3. Execution and Interaction Output Steps: The robot executes the charging / power supply scheme, and the robot and the intelligent terminal execute the interaction output scheme individually or in collaboration, outputting coordinated mechanical movements, light, audio and images; S4. Model Optimization Steps: The AI ​​large model program memorizes and learns users, usage scenarios, smart terminals, user selection habits and feedback, and achieves automatic model upgrade and optimization through data collection, feature engineering, model fine-tuning, personalized adaptation and effect evaluation.

17. The on-site interaction method based on an AI large model according to claim 16, characterized in that, The S2 intelligent analysis steps include: User identification sub-step: Identify the current user's identity based on facial recognition or voiceprint recognition technology; Scene understanding sub-step: Analyze the current scene type based on the multimodal perception fusion model, including work scene, learning scene, rest scene or entertainment scene; Demand forecasting sub-step: Based on users' historical behavior data and the current scenario, predict users' possible interaction needs; Strategy generation sub-step: Based on the charging strategy optimization model and the interaction output decision model, multiple candidate schemes are generated and the comprehensive score of each scheme is calculated. When calculating the comprehensive score of the candidate schemes, the scoring factors include user preference matching degree, charging efficiency, energy consumption level, device heat generation, interaction response timeliness and user historical satisfaction feedback. Each factor is assigned a different weight coefficient according to the priority set by the user.

18. The on-site interaction method based on an AI large model according to claim 16, characterized in that, In the S3 execution and interactive output step, the interactive output scheme includes: Mechanical motion output: controls the robotic arm device to perform specific posture changes or motion sequences; Lighting effect output: Controls LED beads to display specific colors, brightness changes, or dynamic lighting effects; Audio output: Play specific voice prompts, music, or sound effects; Image output: Displaying specific images, animations, or video content on an LED display screen; Collaborative output: The interactive intelligent assistant robot and the intelligent terminal execute coordinated output actions in sync to form a unified interactive experience.

19. The on-site interaction method based on an AI large model according to claim 16, characterized in that, It also includes an offline working mode: when the network connection between the interactive intelligent assistant robot and the cloud server is interrupted, it automatically switches to the offline working mode, and the local edge computing module performs lightweight AI inference tasks, including basic user intent recognition, preset interactive responses and security protection functions, and synchronizes the data during the offline period to the cloud server for model updates after the network is restored.

Images