Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Comparative analysis of motor skill acquisition in a novel bimanual task: the role of mental representation and sensorimotor feedback.

Frontiers in human neuroscience·2024

Same author

Exploring motor skill acquisition in bimanual coordination: insights from navigating a novel maze task.

Scientific reports·2024

Same author

Face Generation and Editing With StyleGAN: A Survey.

IEEE transactions on pattern analysis and machine intelligence·2024

Same author

Perception and simulation during concept learning.

Psychological review·2023

Same author

Decentralized control and local information for robust and adaptive decentralized Deep Reinforcement Learning.

Neural networks : the official journal of the International Neural Network Society·2021

Same author

Toward More Robust Hand Gesture Recognition on EIT Data.

Frontiers in neurorobotics·2021

Same journal

Editorial: Synergizing large language models and computational intelligence for advanced robotic systems.

Frontiers in robotics and AI·2026

Same journal

Editorial: Innovations in industry 4.0: advancing mobility and manipulation in robotics.

Frontiers in robotics and AI·2026

Same journal

MPM-based simulation and bounded-error compression of material points for magnetic tactile sensors.

Frontiers in robotics and AI·2026

Same journal

Torque-sensorless control of a high-ratio, backdrivable Wolfrom-gearbox for safe human-centered robotics.

Frontiers in robotics and AI·2026

Same journal

The implications of robot navigation in social space: perceptual effects of socially aware and baseline navigation.

Frontiers in robotics and AI·2026

Same journal

DPTG: diffusion policy with tactile feasibility guidance.

Frontiers in robotics and AI·2026

See all related articles

Search research articles

Related Experiment Video

Updated: Oct 28, 2025

A Tactile Automated Passive-Finger Stimulator TAPS

A Tactile Automated Passive-Finger Stimulator TAPS

Published on: June 3, 2009

Using Tactile Sensing to Improve the Sample Efficiency and Performance of Deep Deterministic Policy Gradients for

Andrew Melnik¹, Luca Lach^1,2, Matthias Plappert³

¹CITEC, Bielefeld University, Bielefeld, Germany.

Frontiers in Robotics and AI

|July 16, 2021

Summary

This summary is machine-generated.

Adding tactile information to deep reinforcement learning significantly improves robot training efficiency and performance in object manipulation tasks. This advancement reduces the need for extensive interaction samples, making robotic learning more accessible.

Keywords:

deep learning in-hand manipulation reinforcement learning robotics sample-efficiency shadow dexterous hand tactile sensing

More Related Videos

Automated Rat Single-Pellet Reaching with 3-Dimensional Reconstruction of Paw and Digit Trajectories

Automated Rat Single-Pellet Reaching with 3-Dimensional Reconstruction of Paw and Digit Trajectories

Published on: July 10, 2019

Estimation of Contact Regions Between Hands and Objects During Human Multi-Digit Grasping

Estimation of Contact Regions Between Hands and Objects During Human Multi-Digit Grasping

Published on: April 21, 2023

Related Experiment Videos

Last Updated: Oct 28, 2025

A Tactile Automated Passive-Finger Stimulator TAPS

A Tactile Automated Passive-Finger Stimulator TAPS

Published on: June 3, 2009

Automated Rat Single-Pellet Reaching with 3-Dimensional Reconstruction of Paw and Digit Trajectories

Automated Rat Single-Pellet Reaching with 3-Dimensional Reconstruction of Paw and Digit Trajectories

Published on: July 10, 2019

Estimation of Contact Regions Between Hands and Objects During Human Multi-Digit Grasping

Estimation of Contact Regions Between Hands and Objects During Human Multi-Digit Grasping

Published on: April 21, 2023

Area of Science:

Robotics
Machine Learning
Artificial Intelligence

Background:

Deep Reinforcement Learning (DRL) has advanced robotics.
Training DRL models requires numerous interaction samples, limiting efficiency in simulated and real-world environments.

Purpose of the Study:

To investigate the impact of tactile information on sample efficiency and performance in dexterous in-hand object manipulation tasks.
To analyze the influence of tactile sensor parameters (accuracy and resolution) on DRL training.

Main Methods:

Utilized simulated dexterous in-hand object manipulation tasks within the OpenAI Gym Shadow-Dexterous-Hand environment.
Integrated tactile information with varied sensor-measurement accuracy (ground truth, noisy, Boolean) and spatial resolutions.
Trained DRL agents with and without tactile feedback.

Main Results:

Tactile information substantially increased sample efficiency by up to threefold.
Performance improved by up to 46% with the addition of tactile data.
Investigated the effects of sensor accuracy and resolution on observed improvements.

Conclusions:

Tactile sensing is a crucial factor for enhancing sample efficiency and performance in robotic manipulation tasks trained with DRL.
The study provides valuable insights into optimizing tactile sensor design for robotic learning.
Released enhanced robotics environments to facilitate future research.