Robot cognitive system and method based on physical dynamic-semantic causal collaborative field model

By constructing a robot cognitive system based on a physical dynamic-semantic causal collaborative field model, a highly efficient unity of perception, cognition, and decision-making is achieved, solving the problem of large response delays in complex environments and improving the system's dynamic adaptability and learning ability.

CN122087683APending Publication Date: 2026-05-26张丽娜
View PDF 1 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
张丽娜
Filing Date
2025-12-24
Publication Date
2026-05-26

AI Technical Summary

Technical Problem

Existing robot autonomous operation systems in complex dynamic environments suffer from large response delays and lack a unified perception-cognition-decision closed-loop framework, resulting in poor adaptability to sudden dynamic changes.

Method used

We construct a robot cognitive system based on a physical dynamic-semantic causal cooperative field model. Through the cooperative field computing unit, we realize the essential fusion of multimodal perception and physical-inspired deep reasoning. Combined with a dual-channel memory system, we support the compressed storage and associated retrieval of scene experience.

Benefits of technology

It achieves a highly efficient unified response of perception, cognition, and decision-making, shortens the end-to-end response delay of sudden dynamic obstacles, enhances the system's generalization and continuous learning capabilities, and possesses excellent dynamic adaptability and interpretable causal discovery capabilities.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure FT_1
    Figure FT_1
  • Figure FT_2
    Figure FT_2
  • Figure FT_3
    Figure FT_3
Patent Text Reader

Abstract

The invention discloses a robot cognitive system and method based on a physical dynamic-semantic causal collaborative field model, and belongs to the field of intelligent and robot autonomous decision making. The system comprises a multi-modal sensing unit, a collaborative field calculation unit, a cognitive decision module, a behavior execution unit and a dual-channel memory system. The core lies in a unified microgradable model which is constructed by the collaborative field calculation unit and is formed by coupling a physical dynamic field (psi p) and a semantic causal field (psi s) through a bidirectional collaborative protocol. Psi p is an implicit neural field embedded with physical equation constraints, and environment dynamic prediction is realized; psi is a dynamic attribute graph and represents the semantic and causal relationship. The two fields are interlocked and evolved through the protocols of physically-driven semantics and semantically-guided physics. According to the method, attention control based on field gradient, causal reasoning of physical inspiration and simulation verification planning in the field are realized. According to the method, the defects of slow response and poor adaptability of the traditional modular architecture are overcome, the end-to-end response delay in the test is less than or equal to 150ms (the actually measured mean value is 126ms), and the model migration training efficiency is improved to 4.2 times of that of a baseline model (that is, the training time is reduced by 76%).
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Cognitive regulation and training system based on multi-modal fusion

    CN114557677A