ICRA 2015: Using the Hubo Platform to Advance Humanoids Research

From HuboWiki

(Redirected from ICRA 2015)
Jump to: navigation, search


Program for Saturday, May 30, 2015

Full Day Tutorial Session: 8:30 AM - 5:00 PM


8:30 AM: Youngmoo Kim, ExCITe Center, Drexel University: Recent Advances in the Hubo Platform

9:00-11:00 AM: Hands-on tutorial: MAESTOR, a ROS-based software platform for Hubo control and simulation

10:00-10:30 AM: Coffee Break

11:00 AM: C. S. George Lee, Assistive Robotics Technology (ART) Lab, Purdue University: Ladder climbing with Hubo and DRC-Hubo

11:30 AM: Jeremy Fishel, SynTouch: BioTac sensor integration with Hubo


12:00-1:30 PM: Hubo live demonstrations


1:30 PM: Paul Oh, Drones & Autonomous Systems Lab (DASL), UNLV: Hubo platforms and the DARPA Robotics Challenge

2:00-3:00 PM: Hands-on tutorial and demos: Motion scripting and sensor integration using MAESTOR

3:00-3:30 PM: Coffee Break

3:30 PM: Stefan Schaal, Computational Learning and Motor Control Lab, USC: Machine Learning for Hubo and other humanoids

4:00-5:00 PM: Tutorial wrap-up

Information for hands-on tutorial sessions

During the hands-on tutorials, participants will have the opportunity to script gestures to run on a Hubo2+ humanoid, live at the workshop. To participate, you will need the following:

  • A laptop (Mac, Linux, Windows) with approximately 20GB of free storage space.
  • VirtualBox virtualization software. Feel free to download and install this in advance of the session.
  • A virtual machine image that we will provide and distribute at the session. The updated image can also be downloaded here.

Links to source code

Tutorial Overview

The Hubo platform is a series of adult-sized humanoids developed at the Korea Advanced Institute of Science and Technology. Hubo was one of the first large humanoids developed in academia, and has been the centerpiece of several large international collaborative projects, including the DARPA Robotics Challenge (involving a team of 10 institutions) and an NSF Major Research Instrumentation project (bringing Hubo research to 7 US Universities). This tutorial will bring together a broad range of collaborators on these and other projects to survey the research being performed with Hubo and to detail the systems and software that have been developed using Hubo that have broad application to other humanoids and the general robotics community. In addition to presentations from leading researchers, the session will feature hands-on tutorials using newly developed open source software for Hubo simulation and control. Participants will also have an opportunity to test their simulations on a Hubo robot at the session. This tutorial will provide a forum for advancing collaboration by highlighting integrated humanoids research, bridging a diversity of efforts from mechatronics and hardware design to machine learning for sensing, perception, and motion planning.

The purpose of this session is to bring together the broad range of researchers working with the various Hubo platform robots to present the latest advances in the hardware platform and systems and research enabled by the platform. Much of the recent software developed for Hubo has been designed to scale across multiple platforms, so this session will be highly relevant to all Humanoids researchers, from those working with small to adult-size humanoids. The session will also emphasize new Hubo simulation platforms so that those without access to hardware can still engage with humanoids research. Highlights of the tutorial program include the following:

  • Current state-of-the-art of research activities using the Hubo platform
  • Live demonstrations throughout the session on a Hubo2+ robot
  • Sessions for hands-on learning in open source software simulation environments
  • Providing an opportunity of demonstrating and discussing recent results of integration activities;
  • Advancing collaboration in humanoid robotics by bringing together researchers pursuing a wide variety of efforts, from humanoid hardware design to motion planning and machine learning for sensing and perception.


We thank the IEEE RAS Technical Committee on Humanoid Robotics for their support.

Personal tools