A cognitive system for human-machine interaction or inter-robot communication based on four-dimensional discrete space-time system

Abstract

This invention discloses a cognitive architecture for human-computer interaction and robot-to-robot communication based on a four-dimensional discrete spatiotemporal cognitive system, belonging to the field of artificial intelligence human-computer interaction and robot-to-robot communication technology. The four-dimensional discrete spatiotemporal cognitive system is embedded in a large language model and a robot control system, including a time coordinate axis, a context coordinate axis, a logical coordinate axis, and a device coordinate axis. The method includes: the cognitive system automatically generating four-dimensional discrete coordinates for each token, where the time coordinate is based on real-time time, the context coordinate is based on the dialogue sequence number, the logical coordinate is based on the token's sequential position in the interaction information, and the device coordinate is based on the device's globally unique identifier; in the attention mechanism of the large language model and the robot control system, the attention weight is adjusted according to the coordinate differences between tokens. This invention uses the four-dimensional discrete spatiotemporal cognitive system as a fixed underlying architecture for human-computer interaction and robot-to-robot communication, effectively solving problems such as context confusion, timing control, and multi-device collaboration in multi-turn dialogues during human-computer interaction and robot-to-robot communication.

Description

Technical Field

[0001] The present invention aims to provide a cognitive system for human-computer interaction and robot-to-robot communication, in order to solve problems such as inconsistent interactive data formats and lack of explicit cognition of key information in the prior art, and to systematically improve the cognitive ability of machines and large models. Background Technology

[0002] Existing large language models and robot control systems, when processing natural language commands or multimodal interactions in human-computer interaction and robot communication, typically treat input data as a one-dimensional sequence. Due to a lack of explicit, structured understanding of time, speaking order, logical steps, and devices, problems such as contextual confusion, command misinterpretation, and multi-turn tracking failures frequently arise. Therefore, with the development of embodied intelligence, there is an urgent need for a low-level cognitive architecture embedded in large language models and robot control systems. Summary of the Invention

[0003] The technical problem that the invention aims to solve This invention aims to provide a four-dimensional discrete spatiotemporal cognitive architecture embedded in large language model systems and robot control systems, in order to solve the problem of the lack of structured cognitive ability in human-computer interaction and robot-to-robot communication in the prior art.

[0004] Technical solution A four-dimensional discrete-time spatiotemporal cognitive system for large language model systems and robot control systems, characterized in that the cognitive system is embedded in the large language model system and the robot control system, comprising: Coordinate system construction module: Used to construct four-dimensional discrete-time coordinate systems in large language modeling systems and robot control systems. Time coordinate (T): The value is the real-time absolute timestamp when the Token was generated, based on a unified reference time source (such as UTC). Context coordinates (C): The value is the message sequence number. All tokens within the same message share the same C value. C changes monotonically with each new message. Logical coordinate (L): The value is the sequential position of the token within its respective speech, and it changes incrementally starting from a preset starting value; Device Coordinates (D): The value is a globally unique identifier of the physical entity (including robots, sensors, peripherals, etc.) associated with the Token.

[0005] Coordinate mapping module: used to automatically map each token received by the large language model and the robot to the corresponding coordinate point in the four-dimensional discrete spatiotemporal coordinate system, and obtain the four-dimensional discrete spatiotemporal coordinates of the token; Cognitive processing module: Used in the processing unit of large language model systems and robot control systems to adjust attention weights or feature extraction paths based on the coordinate differences between tokens;

[0006] Furthermore, the arrangement order, name changes, and coordinate variation rules of the coordinate axes do not constitute a limitation on the present invention.

[0007] Furthermore, the name change of the four-dimensional discrete spatiotemporal cognitive system does not constitute a limitation of the present invention.

[0008] Beneficial effects Unified Format: The cognitive architecture of this invention is built into the large language model system and robot control system, and does not depend on the format of the input data. All interaction tokens carry four-dimensional coordinates, forming a unified data format that allows interoperability between different systems.

[0009] Scalable: In addition to four-dimensional coordinates, the dialog box dimension and other coordinate dimensions required by algorithms can be added as needed to adapt to future needs.

[0010] Decoupling from algorithms: This format only defines data specifications and does not limit any parsing or usage methods. Each system can implement it freely.

[0011] This architecture is applicable to all types of large language model systems and various robot systems such as industrial robots, service robots, embodied intelligent robots, and autonomous driving robots.

[0012] Enhancing large language models and robot environmental perception and task execution capabilities: Through explicit spatiotemporal coordinates, robots can accurately understand the time constraints, step sequence, and multi-device collaborative relationships of tasks.

Claims

1. Claim 1 (Independent claim, framework) A four-dimensional discrete spatiotemporal cognitive architecture for human-computer interaction and robot-to-robot communication, characterized in that, The cognitive architecture is embedded in the large language model and robot control system, including: The coordinate system construction module is used to construct a four-dimensional discrete spatiotemporal coordinate system in large language models and robot control systems. The four-dimensional discrete spatiotemporal coordinate system includes a time coordinate axis, a context coordinate axis, a logical coordinate axis, and a device coordinate axis. The coordinate mapping module is used to automatically map each token received by the large language model and the robot to the corresponding coordinate point in the four-dimensional discrete spatiotemporal coordinate system to obtain the four-dimensional discrete spatiotemporal coordinates of the token. The mapping does not depend on the explicit labels in the input data. The cognitive processing module is used in the processing units of large language models and robot control systems to adjust attention weights or feature extraction paths based on the coordinate differences between tokens.

2. Claim 2 The cognitive architecture according to claim 1 is characterized in that, The time axis value is the real-time absolute timestamp when the token is generated or received, based on a unified reference time source.

3. Claim 3 The cognitive architecture according to claim 1 is characterized in that, The context coordinate axis is set to either the speech sequence number or the task sequence number, and it increases monotonically with each new input.

4. Claim 4 The cognitive architecture according to claim 1 is characterized in that, The logical coordinate axis is defined by the sequential position of the Token within its input sequence.

5. Claim 5 The cognitive architecture according to claim 1 is characterized in that, The values ​​of the device coordinate axes are globally unique identifiers of the physical entities (including robots, sensors, peripherals, etc.) associated with the Token.

6. Claim 6 The cognitive architecture according to claim 1 is characterized in that, The coordinate mapping module does not rely on explicit special markers in the input data, but rather the coordinates are automatically calculated internally by the robot control system.

7. Claim 7 The cognitive architecture according to claim 1 is characterized in that, The cognitive processing module adjusts attention weights in ways including, but not limited to, enhancing attention between tokens that have the same context coordinates or similar logical coordinates.

8. Claim 8 The cognitive architecture according to claim 1 is characterized in that, The arrangement order, name changes, and coordinate variation rules of the coordinate axes do not constitute a limitation on the present invention, and any arrangement order, name changes, and coordinate variation rules of the coordinate axes are considered equivalent.

9. Claim 9 The cognitive architecture according to claim 1 is characterized in that, The name of the four-dimensional discrete spatiotemporal cognitive system does not constitute a limitation on the present invention, and any change of name shall be regarded as equivalent.

10. Claim 10 A machine language model and robot control system comprising the cognitive architecture described in any one of claims 1 to 9, characterized in that, The system includes at least one processor and memory, the processor and memory having computer programs implementing the cognitive architecture.