Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Learning reaching strategies through reinforcement for a sensor-based manipulator.

P Martín1, J del R Millán

  • 1Department of Computer Science, University of Jaume I, Campus de Penyeta Roja, 12071 Castellón, Spain.

Neural Networks : the Official Journal of the International Neural Network Society
|March 29, 2003
PubMed
Summary
This summary is machine-generated.

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

An online EEG BCI based on covert visuospatial attention in absence of exogenous stimulation.

Journal of neural engineering·2013
Same author

EEG-based Brain-Computer Interface to support post-stroke motor rehabilitation of the upper limb.

Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual International Conference·2013
Same author

Time-dependent approach for single trial classification of covert visuospatial attention.

Journal of neural engineering·2012
Same author

A brain-actuated wheelchair: asynchronous and non-invasive Brain-computer interfaces for continuous control of robots.

Clinical neurophysiology : official journal of the International Federation of Clinical Neurophysiology·2008
Same author

A local neural classifier for the recognition of EEG patterns associated to mental tasks.

IEEE transactions on neural networks·2008
Same author

Classification of EEG mental patterns by using two scalp electrodes and Mahalanobis distance-based classifiers.

Methods of information in medicine·2002

This study introduces a novel neural controller for robot arms, enabling them to learn obstacle avoidance and goal-seeking behaviors online using local sensory data. The combined actor-critic and differential inverse kinematics modules accelerate learning for robot arm control.

Area of Science:

  • Robotics
  • Artificial Intelligence
  • Control Systems

Background:

  • Multilink robot arms require sophisticated control strategies for complex tasks like obstacle avoidance.
  • On-line learning from local sensory data is crucial for adaptable and efficient robotic systems.
  • Traditional methods for calculating shortest paths in configuration space can be computationally intensive for manipulators with multiple links.

Purpose of the Study:

  • To develop a neural controller capable of learning goal-oriented, obstacle-avoiding strategies for multilink robot arms.
  • To enhance the efficiency of learning by integrating a differential inverse kinematics module to overcome computational challenges.
  • To leverage the inherent symmetry of robot arm kinematics for improved controller performance.

Main Methods:

Related Experiment Videos

  • A two-module neural controller architecture was implemented, comprising an actor-critic module and a differential inverse kinematics (DIV) module.
  • The actor-critic module utilizes the Shortest Path Vector (SPV) to guide actions towards the closest goal.
  • The DIV module approximates manipulator forward kinematics using a neural network and inverts it to provide a goal vector, bypassing complex SPV calculations for arms with more than two links.

Main Results:

  • The proposed neural controller successfully learns goal-oriented obstacle-avoiding reaction strategies on-line.
  • The integration of the DIV module addresses the computational burden of SPV calculation for multilink arms.
  • Experimental results with a two-link robot arm demonstrated a significant speed-up in the learning process due to the combined module approach.

Conclusions:

  • The presented neural controller effectively enables multilink robot arms to learn complex navigation and task-execution behaviors.
  • The DIV module offers a computationally efficient alternative for goal vector generation in robot arm control.
  • The combined architecture accelerates the learning process, paving the way for more adaptive and responsive robotic systems.