Robot collision detection optimization method and apparatus, electronic device, and storage medium

By employing two collision detection methods and self-learning model optimization, the problem of robot detection models being unable to adapt to complex environments was solved, thereby improving the safety and robustness of robot operation.

WO2025001931A1PCT designated stage expired Publication Date: 2025-01-02SHANGHAI JIEKA ROBOT TECH CO LTD
View PDF 6 Cites 2 Cited by

Patent Information

Application Number
PCT/CN2024/099946
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-06-28
Filing Date
2024-06-18
Publication Date
2025-01-02

AI Technical Summary

Technical Problem

Existing robot collision detection methods suffer from frequent false collision reports due to the complexity and variability of on-site conditions and the inability of factory-preset detection models to accurately adapt to application scenarios, affecting the safety and robustness of robot operation.

Method used

A two-stage collision detection approach is adopted. After initial detection using a pre-configured collision detection model, the collision detection model is optimized by combining the robot's historical operation data. Reinforcement learning is performed using a pre-trained self-learning model to adjust the model parameters to adapt to specific application scenarios.

Benefits of technology

It reduces the probability of false collisions during robot detection, improves the safety and robustness of robot operation, and adapts to the needs of different application scenarios.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2024099946_02012025_PF_FP_ABST
    Figure CN2024099946_02012025_PF_FP_ABST
Patent Text Reader

Abstract

A robot collision detection optimization method and apparatus, an electronic device, and a computer-readable storage medium. The collision detection optimization method comprises the following steps: by means of a pre-configured collision detection model, performing collision detection on a robot to obtain a first collision detection result; when the first collision detection result indicates that it is detected that the robot has come into collision, determining whether the robot has come into collision indeed to obtain a determination result; and it is determined on the basis of the determination result that the robot has come into collision indeed, optimizing the collision detection model on the basis of the determination result and acquired historical operation data of the robot.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Remote automatic damage assessment and collision detection method and system for vehicle

    CN105976449A

  • Robot collision detection method and device, detection equipment and readable storage medium

    CN114474076A

  • Collision detection method, device and equipment

    CN114565107A

  • Cantilever crane anti-collision detection method, electronic equipment and multi-cantilever crane equipment

    CN115932886A

  • Collision detection optimization method and device of robot, electronic equipment and storage medium

    CN116690649A

Cited By

  • Abnormal collision detection and control method of robot and related device

    CN122087666A

  • Abnormal collision detection and control method of robot and related device

    CN122087666B