Object posture judgment method based on soft grabbing lower pressure vector array analysis

By constructing pressure vector arrays and dimensionality reduction technology, combined with soft-NMS algorithm, the problem of identifying the posture and contact surface firmness of the grab object in soft clamping is solved, and high-precision object attitude determination is achieved, which is suitable for precision grasping of power systems.

CN120277485APending Publication Date: 2025-07-08NANJING YUEJING INTELLIGENT TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202510337419.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-03-21
Publication Date
2025-07-08

AI Technical Summary

Technical Problem

The existing mechanical detection methods are difficult to accurately identify the gripping posture and the firmness of the contact surface during soft clamping. Especially under complex deformation and poor lighting conditions, the macroscopic vision method is insufficient in adaptability.

Method used

By constructing a pressure vector array, conducting contact surface force vector analysis and feature modeling, combining dimensionality reduction and edge recognition methods, image-like decomposition technology is used to identify the posture of the grasped object, and border detection is optimized by using soft-NMS algorithm, and posture determination is achieved by combining rotation force and center of gravity analysis.

Benefits of technology

It realizes millimeter-level precision grasping position determination and accurate identification of two-dimensional attitude angles less than 1 degree to meet the business scenario requirements of the power system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120277485A_ABST
    Figure CN120277485A_ABST
Patent Text Reader

Abstract

The invention provides an object posture judgment method based on soft grabbing lower pressure vector array analysis, and aims at providing an object recognition and posture judgment method based on contact force feedback to solve the problems that in the grabbing process of an existing robot arm, object recognition depends on visual guidance, and vision is limited by ambient light and object shielding. The invention discloses a vector force array dimension reduction analysis method, which comprises the following steps of: disassembling a three-dimensional contact surface in a space into independent stress planes by searching boundary force characteristics of the contact surface, and reducing an angle dimension in a vector array into a two-dimensional scalar array only with the magnitude of force; by training and learning the mechanical characteristics of the contact surface, perception and posture judgment of an object in the pressing and grabbing process are realized. According to the grabbing face rapid analysis and extension posture judgment technology based on the convolutional neural network, grabbing position judgment with millimeter-level precision can be achieved, the two-dimensional posture angle of a grabbed object is smaller than 1 degree, and the requirement for grabbing a sensing service scene is met on the whole.
Need to check novelty before this filing date? Find Prior Art