Robot autonomous tool construction method, system and related equipment based on graph neural network

Abstract

The invention discloses a graph neural network-based robot autonomous tool construction method, system and related equipment. The method generates training data in a simulation environment, takes the depth map containing the target tool and the depth map containing multiple candidate parts as input, and first uses a deep network to select candidate parts, connect points, and build pose regression. The graph neural network is used to connect each part pair, and finally the optimal candidate part is selected comprehensively for building the target tool. The data generation method in the simulation environment used in this method effectively reduces the labor load of data collection and expands the volume of the data set. The invention can help the robot to effectively complete the automatic tool construction task in the scene of pure visual input, and increase the consideration of the robot to the integrity of the scene, which is of great significance to the development of the fusion robot project.

Description

technical field [0001] The invention belongs to the field of computer vision and fusion robots, and in particular relates to a method, system and related equipment for constructing robot autonomous tools based on a graph neural network. Background technique [0002] Compared with the original tool construction method, the robot autonomous tool construction method based on graph neural network greatly improves the flexibility, accuracy and adaptability of tool construction. Robot autonomous tool construction can be interpreted as given a reference tool, the robot automatically selects the most suitable part from some candidate parts to build the tool. Most of the existing tool construction methods first divide the reference tool into a functional part and a grasping part, then match the candidate part with the segmented part, select the most similar part, and finally splice the part. This method makes the splicing of tools only in the mode of functional part and grasping par...

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to overcome the disadvantages of low flexibility, poor adaptability, and lack of macro understanding of the scene in the original technology, and provide a robot autonomous tool construction method, system and related equipment based on graph neural network, which can better understand Self-contained tools build scenarios to complete the task more flexibly, comprehensively, efficiently and accurately

Images