Learning Object-Action Relations from Bimanual Human Demonstration Using Graph Networks

Recognizing human actions is a vital task for a humanoid robot, especially in domains like programming by demonstration. Previous approaches on action recognition primarily focused on the overall prevalent action being executed, but we argue that bimanual human motion cannot always be described suff...

Full description

Saved in:
Bibliographic Details
Published in:IEEE robotics and automation letters Vol. 5; no. 1; pp. 187 - 194
Main Authors: Dreher, Christian R. G., Wachter, Mirko, Asfour, Tamim
Format: Journal Article
Language:English
Published: Piscataway IEEE 01.01.2020
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Subjects:
Classification
Classifiers
Domains
Feature extraction
Graphical representations
Ground truth
Human motion
Humanoid
Labels
Learning from demonstration
Machine learning
Moving object recognition
Object recognition
Programming
Robots
Segmentation
semantic scene understanding
Semantics
Task analysis
Three-dimensional displays
Video data
visual learning
ISSN:2377-3766, 2377-3766
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Recognizing human actions is a vital task for a humanoid robot, especially in domains like programming by demonstration. Previous approaches on action recognition primarily focused on the overall prevalent action being executed, but we argue that bimanual human motion cannot always be described sufficiently with a single action label. We present a system for framewise action classification and segmentation in bimanual human demonstrations. The system extracts symbolic spatial object relations from raw RGB-D video data captured from the robot's point of view in order to build graph-based scene representations. To learn object-action relations, a graph network classifier is trained using these representations together with ground truth action labels to predict the action executed by each hand. We evaluated the proposed classifier on a new RGB-D video dataset showing daily action sequences focusing on bimanual manipulation actions. It consists of 6 subjects performing 9 tasks with 10 repetitions each, which leads to 540 video recordings with 2 hours and 18 minutes total playtime and per-hand ground truth action labels for each frame. We show that the classifier is able to reliably identify (action classification macro F1-score of 0.86) the true executed action of each hand within its top 3 predictions on a frame-by-frame basis without prior temporal action segmentation.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2377-3766
2377-3766
DOI:10.1109/LRA.2019.2949221