Combining decision making and dynamical systems for monitoring and executing manipulation tasks

Matteo Saveriano,Justus Piater

doi:10.1007/s00502-020-00816-7

Matteo Saveriano, Justus Piater

Open Access

https://doi.org/10.1007/s00502-020-00816-7

Copy DOI

Abstract

In this paper, we propose a unified framework for online task scheduling, monitoring, and execution that integrates reconfigurable behavior trees, a decision-making framework with integrated low-level control functionalities, and reactive motion generation with stable dynamical systems. In this way, we realize a flexible and reactive system capable of coping with unexpected variations in the executive context without penalizing modularity, expressiveness, and readability of humans. The framework is evaluated in a simulated sorting task showing promising results in terms of flexibility regarding task scheduling and robustness to external disturbances.

Highlights

Robots operating in everyday environments need increased flexibility to cope with sudden and unexpected variations in their executive context
Task scheduling and execution monitoring are handled by Reconfigurable Behaviour Trees (RBTs) [4], while stable dynamical systems are exploited for online motion generation
RBTs extend the traditional behavior trees [2] with control layer features permitting the continuous monitoring of the task execution and the online replanning of the task to react to perceptual stimuli

Summary

Introduction

Robots operating in everyday environments need increased flexibility to cope with sudden and unexpected variations in their executive context. The enhanced reasoning and acting capabilities of cognitive robots have to be implemented considering facets like modularity, reusability, design effectiveness, and human-readability. RBTs extend the traditional behavior trees [2] with control layer features permitting the continuous monitoring of the task execution and the online replanning of the task to react to perceptual stimuli. This is useful to rapidly react to external perturbations in the task execution like unexpected changes in the pose of the manipulated object. The control layer of RBT is implemented using only the basic components of a traditional BT, de facto keeping the same level of expressiveness, modularity, and human-readability

Methods

Results

Conclusion