Chat robot interaction method and system based on LLM model

By utilizing non-acoustic sensor data and large language models, generating natural language descriptions and controlling virtual character performance, the problem of insufficient utilization of the environment and device status of chat robots is solved, and the interaction effect and task execution efficiency are improved.

CN119091872APending Publication Date: 2024-12-06SHANDONG INSPUR SCI RES INST CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411026765.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-30
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

Existing interactive chatbots have limitations in responding and providing content, and lack full utilization of the environment and device status, resulting in their responses not meeting user needs.

Method used

By obtaining non-acoustic sensor data of client devices and their environment, classifying and generating natural language descriptions, combining pre-trained large language model (LLM) output, controlling the facial expressions, sounds and gestures of virtual characters, enhancing the interactive effect of chat robots .

Benefits of technology

It improves the interactive resonance between chat robots and users, enhances the autonomy and efficiency of the robots in performing tasks, and provides a more emotional interactive experience.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119091872A_ABST
    Figure CN119091872A_ABST
Patent Text Reader

Abstract

The invention discloses a chat robot interaction method and system based on an LLM model, and relates to the technical field of robot control, and the method comprises the steps: obtaining non-acoustic sensor data of a client device and an environment where the client device is located; classifying the acquired data into equipment state data and environment state data of the client equipment, and then generating natural language description of an equipment state and an environment state according to a preset condition; inputting the acquired data and / or the generated description into a pre-trained LLM model to generate corresponding LLM output; the chat robot generates a natural language declaration responsive to the non-acoustic sensor data based on the LLM output, controls facial expressions, sounds, and gestures of the virtual character, and presents to the user through a voice interface or graphical interface of the client device. According to the invention, the performance of the interactive chat robot is improved by using the sensor data and the LLM model, so that the dialogue of the interactive chat robot can resonate with the user, and the interaction effect is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of robot control technology, and in particular to a chat robot interaction method and system based on an LLM model. Background Art

[0002] A robot is a machine designed to perform a task and can come in a variety of form factors, including humanoid robots. These robots can be operated via teleoperation systems, where they simulate human behavior and / or are designed as dedicated robots for specific tasks or general-purpose robots for multiple tasks. Humans perform a wide range of tasks in their personal and professional lives, ranging from mundane chores to complex analytical tasks. There is a strong desire to automate these tasks to allow humans to focus on other activities. Large Language Models (LLMs) are a type of artificial intelligence trained on large amounts of text data that generates human-like text responses to natural language inputs. These models are particularly useful in providing text-based user interfaces that facilitate chat-like interactions between users and computers.

[0003] Currently, some interactive chatbots have limitations in responding and providing content. For example, their responses are often based on preconfigured sentences or only on the user's natural language input, and lack full utilization of the environment and device status. Summary of the invention

[0004] In view of the needs and shortcomings of current technological development, the present invention provides a chat robot interaction method and system based on the LLM model, which improves the performance of the interactive chat robot by utilizing sensor data and the LLM model, so that its dialogue can better resonate with users and improve the interaction effect, thereby enhancing the robot's autonomy and efficiency in performing various tasks.

[0005] In the first aspect, the present invention provides a chat robot interaction method based on the LLM model. The technical solution adopted to solve the above technical problems is as follows:

[0006] A chat robot interaction method based on the LLM model comprises the following steps:

[0007] Acquire non-acoustic sensor data about the client device and its environment;

[0008] classifying the acquired non-acoustic sensor data into device state data and environment state data of the client device, and then generating a natural language description of the device state and the environment state according to preset conditions;

[0009] Input the acquired non-acoustic sensor data and / or the generated natural language description into the pre-trained LLM model to generate corresponding LLM output;

[0010] Based on the LLM output, the chatbot generates natural language statements in response to non-acoustic sensor data, controls the facial expressions, sounds, and gestures of the virtual character, and presents them to the user through the voice interface or graphical interface of the client device.

[0011] Optionally, the preset conditions are based on: a specified type and content of the non-acoustic sensor data, a category or service of the chatbot on the client device, selection of a virtual character, acoustic input from the user, non-acoustic input from the user, a state of the client device or environment, and a value of the non-acoustic sensor data satisfying a threshold.

[0012] Further optionally, non-acoustic sensor data of the client device and its environment are obtained, and at the same time, user acoustic input from the acoustic sensor and / or user non-acoustic input are obtained;

[0013] At least one of the acquired non-acoustic sensor data, the generated natural language description, the user acoustic input from the acoustic sensor, and the user's non-acoustic input is input into the pre-trained LLM model to generate a corresponding LLM output.

[0014] Further optionally, the user interacts with the local interactive chatbot via a client device, and all processing and interaction are completed on the client device;

[0015] The local interactive chatbot consists of the following components:

[0016] A sensor data processing engine, responsible for converting non-acoustic sensor data into natural language text describing device status, detecting and parsing user acoustic input, and selecting at least one of the non-acoustic sensor data, natural language text describing device status, and user acoustic input to be processed according to preset conditions;

[0017] Access the pre-trained LLM engine, responsible for sending the selected processing data to the pre-trained LLM model and generating LLM output;

[0018] Content rendering engine, responsible for controlling the facial expressions, sounds, and gestures of the avatar based on LLM output;

[0019] Automatic speech recognition engine ASR, which is responsible for converting the user's acoustic input into text;

[0020] Natural language understanding engine NLU, responsible for understanding the text entered by the user and extracting intent and parameters;

[0021] The text-to-speech engine TTS is responsible for converting the LLM output into speech and outputting it through a local interactive chatbot.

[0022] Further optionally, the user interacts with the cloud-based interactive chatbot via a client device, during which the client device acts as an interface to send the user's acoustic input to the cloud and receive a response from the cloud-based interactive chatbot from the cloud;

[0023] The cloud-based interactive chatbot includes the following components:

[0024] The cloud-based engine ASR is responsible for recognizing user acoustic input in the cloud and converting it into text;

[0025] The cloud-based engine NLU is responsible for understanding the text input by users in the cloud and extracting intent and parameters;

[0026] Cloud-based content rendering engine, responsible for generating and processing content presentation in the cloud;

[0027] The cloud-based LLM engine is responsible for processing data using pre-trained LLM in the cloud and generating LLM output;

[0028] The cloud-based engine TTS is responsible for converting LLM output into speech in the cloud and outputting it through a cloud-based interactive chatbot.

[0029] In the second aspect, the present invention provides a chat robot interaction system based on the LLM model, and the technical solution adopted to solve the above technical problems is as follows:

[0030] A chat robot interaction system based on an LLM model, comprising:

[0031] A data acquisition module for acquiring non-acoustic sensor data of the client device and its environment;

[0032] a data classification module, for classifying the acquired non-acoustic sensor data into device state data and environment state data of the client device;

[0033] A condition preset module, used to preset the conditions for generating natural language descriptions from non-acoustic sensor data;

[0034] A description generation module is used to generate a natural language description of the device status and the environment status according to preset conditions;

[0035] A pre-trained LLM model for taking the acquired non-acoustic sensor data and / or the generated natural language description as input and generating corresponding LLM output;

[0036] A chatbot is used to generate natural language statements in response to non-acoustic sensor data based on the LLM output, control the facial expressions, sounds and gestures of the virtual character, and present them to the user through a voice interface or a graphical interface of a client device.

[0037] Optionally, conditions for generating natural language descriptions from non-acoustic sensor data are preset through a condition preset module, and the preset conditions are based on: the specified type and content of the non-acoustic sensor data, the category or service of the chatbot on the client device, the selection of the virtual character, the user's acoustic input, the user's non-acoustic input, the state of the client device or environment, and the numerical value of the non-acoustic sensor data satisfying a threshold.

[0038] Further optionally, the data acquisition module involved acquires non-acoustic sensor data of the client device and its environment, and at the same time, acquires user acoustic input from the acoustic sensor and / or user non-acoustic input;

[0039] The pre-trained LLM model takes at least one of the acquired non-acoustic sensor data, the generated natural language description, the user acoustic input from the acoustic sensor, and the user's non-acoustic input as input, and generates a corresponding LLM output.

[0040] Further optionally, the user interacts with the local interactive chatbot via a client device, and all processing and interaction are completed on the client device;

[0041] The local interactive chatbot consists of the following components:

[0042] A sensor data processing engine, responsible for converting non-acoustic sensor data into natural language text describing device status, detecting and parsing user acoustic input, and selecting at least one of the non-acoustic sensor data, natural language text describing device status, and user acoustic input to be processed according to preset conditions;

[0043] Access the pre-trained LLM engine, responsible for sending the selected processing data to the pre-trained LLM model and generating LLM output;

[0044] Content rendering engine, responsible for controlling the facial expressions, sounds, and gestures of the avatar based on LLM output;

[0045] Automatic speech recognition engine ASR, which is responsible for converting the user's acoustic input into text;

[0046] Natural language understanding engine NLU, responsible for understanding the text entered by the user and extracting intent and parameters;

[0047] The text-to-speech engine TTS is responsible for converting the LLM output into speech and outputting it through a local interactive chatbot.

[0048] Further optionally, the user interacts with the cloud-based interactive chatbot via a client device, during which the client device acts as an interface to send the user's acoustic input to the cloud and receive a response from the cloud-based interactive chatbot from the cloud;

[0049] The cloud-based interactive chatbot includes the following components:

[0050] The cloud-based engine ASR is responsible for recognizing user acoustic input in the cloud and converting it into text;

[0051] The cloud-based engine NLU is responsible for understanding the text input by users in the cloud and extracting intent and parameters;

[0052] Cloud-based content rendering engine, responsible for generating and processing content presentation in the cloud;

[0053] The cloud-based LLM engine is responsible for processing data using pre-trained LLM in the cloud and generating LLM output;

[0054] The cloud-based engine TTS is responsible for converting LLM output into speech in the cloud and outputting it through a cloud-based interactive chatbot.

[0055] Compared with the prior art, the chat robot interaction method and system based on the LLM model of the present invention have the following beneficial effects:

[0056] 1. The present invention improves the performance of interactive chatbots by using sensor data and LLM models, making their dialogues more resonant with users and improving the interaction effect, thereby enhancing the autonomy and efficiency of the robots in performing various tasks;

[0057] 2. The chat robot of the present invention can control the facial expressions, voices, gestures, etc. of the virtual character according to the output of the LLM model, so that it can enhance the interactive experience through emotional performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Attached Figure 1 is a flow chart of a method according to Embodiment 1 of the present invention;

[0059] Attached Figure 2 This is a module connection block diagram of the second embodiment of the present invention. DETAILED DESCRIPTION

[0060] In order to make the technical solution, the technical problem solved and the technical effect of the present invention more clearly understood, the technical solution of the present invention is clearly and completely described below in conjunction with specific embodiments.

[0061] Embodiment 1:

[0062] Combined with Figure 1 This embodiment proposes a chat robot interaction method based on the LLM model, which includes the following steps:

[0063] S1. Acquire non-acoustic sensor data of a client device and its environment, and / or acquire user acoustic input from an acoustic sensor and / or user non-acoustic input.

[0064] S2, classifying the acquired non-acoustic sensor data into device state data and environment state data of the client device, and then generating a natural language description of the device state and the environment state according to preset conditions;

[0065] The preset conditions are based on: the specified type and content of the non-acoustic sensor data, the category or service of the chatbot on the client device, the selection of the virtual character, the user's acoustic input, the user's non-acoustic input, the state of the client device or environment, and the value of the non-acoustic sensor data meeting the threshold.

[0066] S3. Input at least one of the acquired non-acoustic sensor data, the generated natural language description, the user acoustic input from the acoustic sensor, and the user's non-acoustic input into the pre-trained LLM model to generate a corresponding LLM output.

[0067] S4. Based on the LLM output, the chatbot generates natural language statements in response to non-acoustic sensor data, controls the facial expressions, sounds, and gestures of the virtual character, and presents them to the user through the voice interface or graphical interface of the client device.

[0068] In the process of executing the above steps S1-S4, ① if the ambient light data detected by the ambient light sensor is "30%", and the set ambient light data threshold is "35%", then the corresponding natural language description generated is "it's getting dark" or "the current ambient light is very dim", and it is input into the pre-trained LLM model, and the generated LLM output may be "how about turning on some lights"; ② if the device status data detected by the acceleration sensor is "acceleration: x=0.01g, y=-9.0g, z=0.04g", then the corresponding natural language description generated is "the device is falling freely", and it is input into the pre-trained LLM model, and the generated LLM output may be "Oh, so exciting! Where will we fall?" ③ If the user's acoustic input or non-acoustic input is "Is there any way to make it brighter here", it is input into the pre-trained LLM model, and the generated LLM output may be "how about turning on some lights" or "how about putting some flowers or colorful pillows".

[0069] It is necessary to add that:

[0070] (A) Users interact with the local interactive chatbot through client devices, and all processing and interaction are completed on the client devices;

[0071] The local interactive chatbot consists of the following components:

[0072] A sensor data processing engine, responsible for converting non-acoustic sensor data into natural language text describing device status, detecting and parsing user acoustic input, and selecting at least one of the non-acoustic sensor data, natural language text describing device status, and user acoustic input to be processed according to preset conditions;

[0073] Access the pre-trained LLM engine, responsible for sending the selected processing data to the pre-trained LLM model and generating LLM output;

[0074] Content rendering engine, responsible for controlling the facial expressions, sounds, and gestures of the avatar based on LLM output;

[0075] Automatic speech recognition engine ASR, responsible for converting the user's acoustic input into text;

[0076] The natural language understanding engine NLU is responsible for understanding the text entered by the user and extracting the intent and parameters;

[0077] The text-to-speech engine TTS is responsible for converting the LLM output into speech and outputting it through a local interactive chatbot.

[0078] (B) A user interacts with the cloud-based interactive chatbot via a client device, and during the interaction, the client device acts as an interface to send the user's acoustic input to the cloud and receive a response from the cloud-based interactive chatbot from the cloud;

[0079] The cloud-based interactive chatbot includes the following components:

[0080] The cloud-based engine ASR is responsible for recognizing user acoustic input in the cloud and converting it into text;

[0081] The cloud-based engine NLU is responsible for understanding the text input by users in the cloud and extracting intent and parameters;

[0082] Cloud-based content rendering engine, responsible for generating and processing content presentation in the cloud;

[0083] The cloud-based LLM engine is responsible for processing data using pre-trained LLM in the cloud and generating LLM output;

[0084] The cloud-based engine TTS is responsible for converting LLM output into speech in the cloud and outputting it through a cloud-based interactive chatbot.

[0085] Embodiment 2:

[0086] Combined with Figure 2 This embodiment proposes a chat robot interaction system based on the LLM model, which includes:

[0087] A data acquisition module for acquiring non-acoustic sensor data of the client device and its environment;

[0088] a data classification module, for classifying the acquired non-acoustic sensor data into device state data and environment state data of the client device;

[0089] A condition preset module, used to preset the conditions for generating natural language descriptions from non-acoustic sensor data;

[0090] A description generation module is used to generate a natural language description of the device status and the environment status according to preset conditions;

[0091] A pre-trained LLM model for taking the acquired non-acoustic sensor data and / or the generated natural language description as input and generating corresponding LLM output;

[0092] A chatbot is used to generate natural language statements in response to non-acoustic sensor data based on the LLM output, control the facial expressions, sounds and gestures of the virtual character, and present them to the user through a voice interface or a graphical interface of a client device.

[0093] In this embodiment, the conditions for generating natural language descriptions from non-acoustic sensor data are preset through a condition preset module, and the preset conditions are based on: the specified type and content of the non-acoustic sensor data, the category or service of the chatbot on the client device, the selection of the virtual character, the user's acoustic input, the user's non-acoustic input, the state of the client device or environment, and the value of the non-acoustic sensor data meets the threshold.

[0094] In this embodiment, the data acquisition module can not only acquire the non-acoustic sensor data of the client device and its environment, but also acquire the user acoustic input from the acoustic sensor and / or the user's non-acoustic input. The pre-trained LLM model takes at least one of the acquired non-acoustic sensor data, the generated natural language description, the user acoustic input from the acoustic sensor, and the user's non-acoustic input as input to generate corresponding LLM output.

[0095] Based on the chat robot interaction system of the present embodiment, ① if the ambient light data detected by the ambient light sensor is "30%", and the set ambient light data threshold is "35%", then the corresponding generated natural language description is "it's getting dark" or "the current ambient light is very dim", and it is input into the pre-trained LLM model, and the generated LLM output may be "how about turning on some lights"; ② if the device status data detected by the acceleration sensor is "acceleration: x=0.01g, y=-9.0g, z=0.04g", then the corresponding generated natural language description is "the device is falling freely", and it is input into the pre-trained LLM model, and the generated LLM output may be "Oh, so exciting! Where will we fall?" ③ If the user's acoustic input or non-acoustic input is "Is there any way to make it brighter here", it is input into the pre-trained LLM model, and the generated LLM output may be "how about turning on some lights" or "how about putting some flowers or colorful pillows".

[0096] It is necessary to add that:

[0097] (A) Users interact with the local interactive chatbot through client devices, and all processing and interaction are completed on the client devices;

[0098] The local interactive chatbot consists of the following components:

[0099] A sensor data processing engine, responsible for converting non-acoustic sensor data into natural language text describing device status, detecting and parsing user acoustic input, and selecting at least one of the non-acoustic sensor data, natural language text describing device status, and user acoustic input to be processed according to preset conditions;

[0100] Access the pre-trained LLM engine, responsible for sending the selected processing data to the pre-trained LLM model and generating LLM output;

[0101] Content rendering engine, responsible for controlling the facial expressions, sounds, and gestures of the avatar based on LLM output;

[0102] Automatic speech recognition engine ASR, which is responsible for converting the user's acoustic input into text;

[0103] Natural language understanding engine NLU, responsible for understanding the text entered by the user and extracting intent and parameters;

[0104] The text-to-speech engine TTS is responsible for converting the LLM output into speech and outputting it through a local interactive chatbot.

[0105] (B) A user interacts with the cloud-based interactive chatbot via a client device, and during the interaction, the client device acts as an interface to send the user's acoustic input to the cloud and receive a response from the cloud-based interactive chatbot from the cloud;

[0106] The cloud-based interactive chatbot includes the following components:

[0107] The cloud-based engine ASR is responsible for recognizing user acoustic input in the cloud and converting it into text;

[0108] The cloud-based engine NLU is responsible for understanding the text input by users in the cloud and extracting intent and parameters;

[0109] Cloud-based content rendering engine, responsible for generating and processing content presentation in the cloud;

[0110] The cloud-based LLM engine is responsible for processing data using pre-trained LLM in the cloud and generating LLM output;

[0111] The cloud-based engine TTS is responsible for converting LLM output into speech in the cloud and outputting it through a cloud-based interactive chatbot.

[0112] In summary, the chat robot interaction method and system based on the LLM model of the present invention improves the performance of the interactive chat robot by utilizing sensor data and the LLM model, so that its dialogue can better resonate with the user and improve the interaction effect, thereby enhancing the autonomy and efficiency of the robot in performing various tasks;

[0113] The above specific examples are used to explain the principles and implementation methods of the present invention in detail. These examples are only used to help understand the core technical content of the present invention. Based on the above specific embodiments of the present invention, any improvements and modifications made to the present invention by technicians in this technical field without departing from the principles of the present invention should fall within the scope of patent protection of the present invention.

Claims

1. A chat robot interaction method based on the LLM model, characterized in that: It includes the following steps: Acquire non-acoustic sensor data about the client device and its environment; classifying the acquired non-acoustic sensor data into device state data and environment state data of the client device, and then generating a natural language description of the device state and the environment state according to preset conditions; Input the acquired non-acoustic sensor data and / or the generated natural language description into the pre-trained LLM model to generate corresponding LLM output; Based on the LLM output, the chatbot generates natural language statements in response to non-acoustic sensor data, controls the facial expressions, sounds, and gestures of the virtual character, and presents them to the user through the voice interface or graphical interface of the client device.

2. A chat robot interaction method based on the LLM model according to claim 1, characterized in that: The preset conditions are based on: the specified type and content of the non-acoustic sensor data, the category or service of the chatbot on the client device, the selection of the virtual character, the user's acoustic input, the user's non-acoustic input, the state of the client device or environment, and the value of the non-acoustic sensor data meeting the threshold.

3. A chat robot interaction method based on the LLM model according to claim 2, characterized in that: Acquire non-acoustic sensor data of the client device and its environment, and acquire user acoustic input from the acoustic sensor and / or user non-acoustic input; At least one of the acquired non-acoustic sensor data, the generated natural language description, the user acoustic input from the acoustic sensor, and the user's non-acoustic input is input into the pre-trained LLM model to generate a corresponding LLM output.

4. A chat robot interaction method based on the LLM model according to claim 3, characterized in that: Users interact with local interactive chatbots through client devices, and all processing and interaction are completed on the client devices; The local interactive chat robot includes the following components: A sensor data processing engine, responsible for converting non-acoustic sensor data into natural language text describing device status, detecting and parsing user acoustic input, and selecting at least one of the non-acoustic sensor data, natural language text describing device status, and user acoustic input to be processed according to preset conditions; Access the pre-trained LLM engine, responsible for sending the selected processing data to the pre-trained LLM model and generating LLM output; Content rendering engine, responsible for controlling the facial expressions, sounds, and gestures of the avatar based on LLM output; Automatic speech recognition engine ASR, which is responsible for converting the user's acoustic input into text; Natural language understanding engine NLU, responsible for understanding the text entered by the user and extracting intent and parameters; The text-to-speech engine TTS is responsible for converting the LLM output into speech and outputting it through a local interactive chatbot.

5. The chat robot interaction method based on the LLM model according to claim 3 is characterized in that: The user interacts with the cloud-based interactive chatbot through a client device. During the interaction, the client device acts as an interface to send the user's acoustic input to the cloud and receive the cloud-based interactive chatbot's response from the cloud. The cloud-based interactive chatbot includes the following components: The cloud-based engine ASR is responsible for recognizing user acoustic input in the cloud and converting it into text; The cloud-based engine NLU is responsible for understanding the text input by users in the cloud and extracting intent and parameters; Cloud-based content rendering engine, responsible for generating and processing content presentation in the cloud; The cloud-based LLM engine is responsible for processing data using pre-trained LLM in the cloud and generating LLM output; The cloud-based engine TTS is responsible for converting LLM output into speech in the cloud and outputting it through a cloud-based interactive chatbot.

6. A chat robot interaction system based on the LLM model, characterized in that: It includes: A data acquisition module for acquiring non-acoustic sensor data of the client device and its environment; a data classification module, for classifying the acquired non-acoustic sensor data into device state data and environment state data of the client device; A condition preset module, used to preset the conditions for generating natural language descriptions from non-acoustic sensor data; A description generation module is used to generate a natural language description of the device status and the environment status according to preset conditions; A pre-trained LLM model for taking the acquired non-acoustic sensor data and / or the generated natural language description as input and generating corresponding LLM output; A chatbot is used to generate natural language statements in response to non-acoustic sensor data based on the LLM output, control the facial expressions, sounds and gestures of the virtual character, and present them to the user through a voice interface or a graphical interface of a client device.

7. A chat robot interaction system based on the LLM model according to claim 6, characterized in that: The conditions for generating natural language descriptions from non-acoustic sensor data are preset through a condition preset module, and the preset conditions are based on: the specified type and content of the non-acoustic sensor data, the category or service of the chatbot on the client device, the selection of the virtual character, the user's acoustic input, the user's non-acoustic input, the state of the client device or environment, and the numerical value of the non-acoustic sensor data meets the threshold.

8. A chat robot interaction system based on the LLM model according to claim 7, characterized in that: The data acquisition module acquires non-acoustic sensor data of the client device and its environment, and at the same time, acquires user acoustic input from the acoustic sensor and / or user non-acoustic input; The pre-trained LLM model takes at least one of the acquired non-acoustic sensor data, the generated natural language description, the user acoustic input from the acoustic sensor, and the user's non-acoustic input as input, and generates a corresponding LLM output.

9. A chat robot interaction system based on the LLM model according to claim 8, characterized in that: Users interact with local interactive chatbots through client devices, and all processing and interaction are completed on the client devices; The local interactive chat robot includes the following components: A sensor data processing engine, responsible for converting non-acoustic sensor data into natural language text describing device status, detecting and parsing user acoustic input, and selecting at least one of the non-acoustic sensor data, natural language text describing device status, and user acoustic input to be processed according to preset conditions; Access the pre-trained LLM engine, responsible for sending the selected processing data to the pre-trained LLM model and generating LLM output; Content rendering engine, responsible for controlling the facial expressions, sounds, and gestures of the avatar based on LLM output; Automatic speech recognition engine ASR, which is responsible for converting the user's acoustic input into text; Natural language understanding engine NLU, responsible for understanding the text entered by the user and extracting intent and parameters; The text-to-speech engine TTS is responsible for converting the LLM output into speech and outputting it through a local interactive chatbot.

10. A chat robot interaction system based on the LLM model according to claim 8, characterized in that: The user interacts with the cloud-based interactive chatbot through a client device. During the interaction, the client device acts as an interface to send the user's acoustic input to the cloud and receive the cloud-based interactive chatbot's response from the cloud. The cloud-based interactive chatbot includes the following components: The cloud-based engine ASR is responsible for recognizing user acoustic input in the cloud and converting it into text; The cloud-based engine NLU is responsible for understanding the text input by users in the cloud and extracting intent and parameters; Cloud-based content rendering engine, responsible for generating and processing content presentation in the cloud; The cloud-based LLM engine is responsible for processing data using pre-trained LLM in the cloud and generating LLM output; The cloud-based engine TTS is responsible for converting LLM output into speech in the cloud and outputting it through a cloud-based interactive chatbot.