Robot small sample sorting method

The invention relates to a robot small sample sorting method. The method comprises the following steps: constructing an end-to-end robot small sample sorting network fusing recognition and grabbing; and according to a meta-learning method, training a robot small sample sorting network to enable the...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: HOU ZIYUE, BI YIFEI, LIU YI, LI WEI, GUO WEIKUN, WEI BINGSHENG, HU ZIYE
Format: Patent
Sprache:chi ; eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The invention relates to a robot small sample sorting method. The method comprises the following steps: constructing an end-to-end robot small sample sorting network fusing recognition and grabbing; and according to a meta-learning method, training a robot small sample sorting network to enable the robot to adapt to different sorting tasks. Compared with the prior art, the small-sample sorting element learning framework endows the robot with the capability of quickly learning objects which do not appear in the sorting training set, not only sorts the target objects, but also enables the robot to learn to grab the target objects according to the grabbing parts during demonstration grabbing, and meanwhile, the robot can not only grab the target objects according to the grabbing parts during demonstration grabbing when encountering new objects every time. The robot can immediately learn to grab the target part of the target object by using a proper grabbing posture only by carrying out one-time sorting demonstra