Adaptive Asynchronous Control Using Meta-Learned Neural Ordinary Differential Equations

Model-based reinforcement learning and control have demonstrated great potential in various sequential decision making problem domains, including in robotics settings. However, real-world robotics systems often present challenges that limit the applicability of those methods. In particular, we note...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on robotics Vol. 40; pp. 403 - 420
Main Authors: Salehi, Achkan, Ruhl, Steffen, Doncieux, Stephane
Format: Journal Article
Language:English
Published: New York IEEE 01.01.2024
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Subjects:
Adaptive control
Adaptive systems
Computer Science
Differential equations
Industrial robots
learning and adaptive systems
Machine Learning
Metalearning
model learning for control
Reinforcement learning
Robot control
Robot learning
Robotics
robust/adaptive control of robotic systems
Service robots
Industrial robots
Model Learning for Control
learning and adaptive systems
Robust/Adaptive Control of Robotic Systems
ISSN:1552-3098, 1941-0468
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Model-based reinforcement learning and control have demonstrated great potential in various sequential decision making problem domains, including in robotics settings. However, real-world robotics systems often present challenges that limit the applicability of those methods. In particular, we note two problems that jointly happen in many industrial systems: first, irregular/asynchronous observations and actions and, second, dramatic changes in environment dynamics from an episode to another (e.g .<inline-formula><tex-math notation="LaTeX">,</tex-math></inline-formula> varying payload inertial properties). We propose a general framework that overcomes those difficulties by meta-learning adaptive dynamics models for continuous-time prediction and control. The proposed approach is task-agnostic and can be adapted to new tasks in a straight-forward manner. We present evaluations in two different robot simulations and on a real industrial robot.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1552-3098
1941-0468
DOI:10.1109/TRO.2023.3326350