Markerless Visual Robot Programming by Demonstration

TL;DR

This paper introduces a method for teaching robots to imitate human actions in a kitchen using markerless visual observation and scene understanding, enabling robots to learn tasks by demonstration without markers.

Contribution

The approach combines convolutional pose estimation, scene ontology, and spatial constraints to translate human demonstrations into executable robot commands.

Findings

Successfully learned a kitchen task through demonstration

Achieved markerless human pose and object detection in real-time

Enabled robot to imitate human behavior based on visual cues

Abstract

In this paper we present an approach for learning to imitate human behavior on a semantic level by markerless visual observation. We analyze a set of spatial constraints on human pose data extracted using convolutional pose machines and object informations extracted from 2D image sequences. A scene analysis, based on an ontology of objects and affordances, is combined with continuous human pose estimation and spatial object relations. Using a set of constraints we associate the observed human actions with a set of executable robot commands. We demonstrate our approach in a kitchen task, where the robot learns to prepare a meal.