Voice-driven 3D automobile model recognition control system

Abstract

The invention provides a voice-driven 3D automobile model recognition control system. The voice-driven 3D automobile model recognition control system comprises a voice recognition module and a natural language processing module, the system further comprises a car model action library platform, the car model action library platform is connected with an intention recognition module and an action recognition module, and the intention recognition module is connected with the action recognition module; the manual labeling operation module is connected with the intention recognition module and the action recognition module; and the automatic simulation test module is connected to the intention recognition module. The voice-driven 3D automobile model recognition control system provided by the invention not only solves the technical difficulty of driving the 3D automobile model through voice, but also brings brand new experience to users, and creates a precedent for interaction between an AI large model and the 3D automobile model.

Description

Technical Field

[0001] The invention relates to the field of 3D automobile model control, and in particular to a voice-driven 3D automobile model recognition control system. Background Art

[0002] Speech recognition technology ASR (Automatic Speech Recognition) is a technology that converts human speech into text. Natural language processing NLP (Natural Language Processing) is a bridge between machine language and human language to achieve the purpose of human-computer communication.

[0003] At present, speech recognition technology (ASR) and natural language processing (NLP) are mature technologies widely used in commercial scenarios. The former can convert human speech into text, and the latter can recognize the intention of expression based on the text content. However, the existing technology cannot further drive the 3D car model through the intention. Summary of the invention

[0004] In order to solve the technical problem of how to further drive a 3D car model through intention, the present invention provides a voice-driven 3D car model recognition control system.

[0005] The present invention solves the above technical problems through the following technical solutions:

[0006] The present invention provides a voice-driven 3D car model recognition control system, comprising a voice recognition module and a natural language processing module; further comprising: a car model action library platform, the car model action library platform is connected to an intention recognition module and an action recognition module, and the intention recognition module is connected to the action recognition module; a manual marking operation module, the manual marking operation module is connected to the intention recognition module and the action recognition module; and an automatic simulation test module, the automatic simulation test module is connected to the intention recognition module.

[0007] In this technical solution, the intention recognition module is used to perform intention recognition, and the action recognition module performs action recognition in the car model action library platform, and then executes it to further drive the 3D car model through intention.

[0008] Preferably, the speech recognition module is connected to a microphone module, and the speech recognition module is connected to the natural language processing module; the speech recognition module converts human speech into text, and the natural language processing module recognizes the intention to be expressed based on the text content.

[0009] In this technical solution, the microphone module is used to receive human speech; the speech recognition module converts the human speech into text, and further, the natural language processing module recognizes the intention to be expressed based on the text content.

[0010] Preferably, the intention recognition module is connected to a natural language processing module, and the intention recognized by the natural language processing module is input into the intention recognition module.

[0011] In this technical solution, the intent identified by the natural language processing module enters the intent recognition module to perform intent recognition.

[0012] Preferably, the intention recognition module and the action recognition module are connected to an AI big model, the actions in the car model action library platform are associated one-to-one with the intentions recognized by the natural language processing module, and the associated information is fed to the AI ​​big model for automatic learning.

[0013] In this technical solution, the intention recognition module and the action recognition module are based on the AI ​​big model, and their capabilities are improved through the automatic learning ability of the AI ​​big model.

[0014] Preferably, the manual labeling operation module labels the associated information.

[0015] In this technical solution, the operation staff continuously performs manual labeling through the manual labeling operation module, so that the AI ​​large model has a better understanding of the user's intentions and the actions of the car model, and the feedback given is more and more accurate.

[0016] Preferably, the car model action library platform includes a standard action library and a special action library.

[0017] In this technical solution, the car model action library platform is universal for all car models, by designing a standard action library and a special action library; the actions in the standard action library support all car models; the special action library is used to adapt to the functions unique to one or some car models.

[0018] Preferably, the vehicle model action library platform also includes a maintenance operation unit for operation personnel to operate.

[0019] In this technical solution, a maintenance operation unit is used to maintain the vehicle model action library platform.

[0020] Preferably, the automatic simulation test module comprises a simulated user unit, and the simulated user unit is connected to the intention recognition module.

[0021] In this technical solution, the automatic simulation test module simulates users by simulating user units for testing and training of large AI models.

[0022] Preferably, the simulated user unit is used for simulating user asking questions.

[0023] In this technical solution, the simulated user unit is used to simulate the user to ask questions

[0024] Preferably, it also includes an action execution animation demonstration module, and the action execution animation demonstration module is connected to the action recognition module.

[0025] In this technical solution, after the intention and action are recognized, the action execution animation demonstration module is used to demonstrate and play the 3D car model action animation required by the intention for display.

[0026] On the basis of being in accordance with the common sense in the art, the above-mentioned preferred conditions can be arbitrarily combined to obtain the preferred embodiments of the present invention.

[0027] The positive and progressive effects of the present invention are:

[0028] The voice-driven 3D car model recognition control system proposed above improves the user's 3D car viewing and selection experience. It can quickly display the car in 3D according to the user's intention by recognizing the user's intention and adding an action execution animation demonstration module, thereby forming an interaction, making the user's car viewing and selection more intuitive and interactive, which greatly improves the user experience. Furthermore, it combines the characteristics of the AI ​​large model, so that it can feedback the actions of the 3D car model on the basis of feedback on the user's intention, so as to carry out display interaction. In summary, the proposed voice-driven 3D car model recognition control system not only solves the technical difficulties of driving the 3D car model by voice, but also brings a new experience to the user, creating a precedent for the interaction between the AI ​​large model and the 3D car model. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a schematic diagram of the overall principle of the present invention.

[0030] Figure 2 Schematic diagram of the architecture of the voice-driven 3D car model recognition control system of the present invention.

[0031] Figure 3 It is a schematic diagram of the architecture of the vehicle model action library platform of the present invention.

[0032] Description of Reference Numerals

[0033] 1. Speech recognition module;

[0034] 2. Natural language processing module;

[0035] 3. Car model action library platform; 301. Standard action library; 302. Special action library; 303. Maintenance operation unit;

[0036] 4. Intent recognition module;

[0037] 5. Action recognition module;

[0038] 6. Manual annotation operation module;

[0039] 7. Automatic simulation test module; 701. Simulate user unit;

[0040] 8. Microphone module;

[0041] 9. AI big model;

[0042] 10. Action execution animation demonstration module. DETAILED DESCRIPTION

[0043] The present invention is further described below by way of examples, but the present invention is not limited to the scope of the examples.

[0044] like Figure 1-3 As shown, the voice-driven 3D car model recognition control system includes a voice recognition module 1 and a natural language processing module 2; it also includes: a car model action library platform 3, the car model action library platform 3 is connected to the intention recognition module 4 and the action recognition module 5, and the intention recognition module 4 is connected to the action recognition module 5; a manual marking operation module 6, the manual marking operation module 6 is connected to the intention recognition module 4 and the action recognition module 5; an automatic simulation test module 7, the automatic simulation test module 7 is connected to the intention recognition module 4.

[0045] In the specific implementation, we will focus on the 3D car model and create action animation effects for all car models to support all car models on the market.

[0046] Among them, the action animation effects include opening and closing car doors, reclining seats, opening and closing trunks, popping out airbags, etc. Further, other functions are added to the 3D client car viewing, such as radar effect display, wind resistance effect display, AI smart navigation, etc. The above animation effects and functions are uniformly named "actions" and form a car model action library platform 3.

[0047] Among them, the car model action library platform 3 is suitable for all car models by designing a standard action library 301 and a special action library 302.

[0048] The actions in the standard action library 301 support all vehicle models, such as opening and closing doors, checking the roof, checking the steering wheel, and rotating the vehicle body; while the dedicated action library 302 is used to adapt to the functions unique to one or some vehicle models, such as checking the lidar, turning on seat ventilation, etc., or to support a function in the 3D client, such as turning on smart navigation, switching between indoor and outdoor scenes, etc.

[0049] By associating all actions and intentions in the car model action library platform 3 and further feeding them to the AI ​​big model 9 for learning, through continuous manual labeling, the AI ​​big model 9 can have a better understanding of the user's intentions and the actions of the car model, and the feedback it gives can become more and more accurate.

[0050] After the above, the whole process can be realized: user speech -> audio to text -> intention recognition -> action recognition -> action execution. For example, when the user says: "Open the main driver's door", the AI ​​will reply "The main driver's door is open" and the car model will execute the animation of the main driver's door opening.

[0051] The speech recognition module 1 is connected to a microphone module 8, and the speech recognition module 1 is connected to the natural language processing module 2; the speech recognition module 1 converts human speech into text, and the natural language processing module 2 recognizes the intention to be expressed according to the text content.

[0052] The intention recognition module 4 is connected to the natural language processing module 2 , and the intention recognized by the natural language processing module 2 is input into the intention recognition module 4 .

[0053] The intention recognition module 4 and the action recognition module 5 are connected to the AI ​​big model 9. The actions in the car model action library platform 3 are associated one by one with the intentions recognized by the natural language processing module 2, and the associated information is fed to the AI ​​big model 9 for automatic learning.

[0054] By combining the characteristics of the AI ​​big model 9, it is able to provide feedback on the user's intentions while also providing feedback on the 3D car model's actions for display interaction.

[0055] Intention and action training, the relationship between intention and action is manually labeled in the early stage, and then fed to the AI ​​model 9 for learning. Through continuous learning, testing, and labeling, the accuracy of feedback is improved.

[0056] The manual labeling operation module 6 labels the associated information.

[0057] The vehicle model action library platform 3 includes a standard action library 301 and a special action library 302 .

[0058] The vehicle model action library platform 3 also includes a maintenance operation unit 303 for operation personnel to operate.

[0059] The Car Model Action Library Platform 3 is designed to configure standard actions and special actions for all car models on the market. Figure 3 As shown, each vehicle model corresponds to multiple parts, and each part corresponds to multiple actions. The operator can add parts and actions to the platform through the maintenance operation unit 303, and can also maintain the relationship between the parts and actions of the vehicle.

[0060] The automatic simulation test module 7 comprises a simulated user unit 701, and the simulated user unit 701 is connected to the intention recognition module 4. The simulated user unit 701 is used to simulate a user asking a question.

[0061] The automated testing module is used to simulate user questions. The text of the AI ​​model 9's answer is directly transferred to the intention and training phase, and the screenshots of the model's actions are saved as documents.

[0062] It also includes an action execution animation demonstration module 10 , which is connected to the action recognition module 5 .

[0063] The voice-driven 3D car model recognition control system proposed above improves the user's 3D car viewing and selection experience. It can quickly display in 3D dimension according to the user's intention by recognizing the user's intention and adding an action execution animation demonstration module 10, forming an interaction, making the user's car viewing and selection more intuitive and highly interactive, which greatly improves the user experience.

[0064] When it comes to voice, text, intent, and action, they are not coupled to each other and are developed and maintained by professional teams in their respective fields. For example, the AI ​​team is responsible for voice and file parsing, the 3D team is responsible for model creation and animation production, the R&D team is responsible for function development, and the operations team is responsible for intent and action maintenance. This can not only fully integrate the company's resources, but also make maintenance and iterative upgrades very convenient.

[0065] The present invention is not limited to the above-mentioned embodiments. Any changes in shape or structure are within the protection scope of the present invention. The protection scope of the present invention is defined by the attached claims. Those skilled in the art can make various changes or modifications to these embodiments without departing from the principle and essence of the present invention, but these changes and modifications are within the protection scope of the present invention.

Claims

1. A voice-driven 3D car model recognition control system, comprising a voice recognition module (1) and a natural language processing module (2); characterized in that: Also includes: A vehicle model action library platform (3), wherein the vehicle model action library platform (3) is connected to an intention recognition module (4) and an action recognition module (5), and the intention recognition module (4) is connected to the action recognition module (5); A manual labeling operation module (6), wherein the manual labeling operation module (6) is connected to the intention recognition module (4) and the action recognition module (5); An automatic simulation test module (7), wherein the automatic simulation test module (7) is connected to the intention recognition module (4).

2. The voice-driven 3D car model recognition control system according to claim 1, characterized in that: The speech recognition module (1) is connected to a microphone module (8), and the speech recognition module (1) is connected to the natural language processing module (2); the speech recognition module (1) converts human speech into text, and the natural language processing module (2) recognizes the intention to be expressed based on the text content.

3. The voice-driven 3D car model recognition control system according to claim 2, characterized in that: The intention recognition module (4) is connected to the natural language processing module (2), and the intention recognized by the natural language processing module (2) is input into the intention recognition module (4).

4. The voice-driven 3D car model recognition control system according to claim 3, characterized in that: The intention recognition module (4) and the action recognition module (5) are connected to an AI large model (9), the actions in the vehicle model action library platform (3) are associated one by one with the intentions recognized by the natural language processing module (2), and the associated information is fed to the AI ​​large model (9) for automatic learning.

5. The voice-driven 3D car model recognition control system according to claim 4, characterized in that: The manual labeling operation module (6) labels the associated information.

6. The voice-driven 3D car model recognition control system according to claim 1, characterized in that: The vehicle model action library platform (3) comprises a standard action library (301) and a special action library (302).

7. The voice-driven 3D car model recognition control system according to claim 1, characterized in that: The vehicle model action library platform (3) also includes a maintenance operation unit (303) for operation personnel to operate.

8. The voice-driven 3D car model recognition control system according to claim 1, characterized in that: The automatic simulation test module (7) comprises a simulated user unit (701), and the simulated user unit (701) is connected to the intention recognition module (4).

9. The voice-driven 3D car model recognition control system according to claim 8, characterized in that: The simulated user unit (701) is used to simulate user questions.

10. The voice-driven 3D car model recognition control system according to claim 1, characterized in that: It also comprises an action execution animation demonstration module (11), wherein the action execution animation demonstration module (11) is connected to the action recognition module (5).

Images