The workshop takes places on Saturyday, July 12, 2014 in room 210 Wheeler. It is structured around 3 pairs of invited presentations, each pair consisting of a human and a robot side. The topics for the three pairs will be a) hands, b) grasping, c) perception. Following each pair of invited talks, there will be a discussion session on the corresponding topic. After the presentation/discussion sessions, there will be a panel discussion. The workshop will close with a poster session under the theme “Novel Directions in Grasping and Manipulation”.


Topic

Humans

Robots

Intro (15 mins)

Organizers

Hands (9:00-10:00)

Francisco Valero-Cuevas

slides as PDF

Aaron Dollar

slides as PDF

Discussion 1 (10:00-10:30)

Towards the right hardware

 

Break (10:30-11:00)

 Coffee break

Grasping (11:00-12:00)

Marco Santello

 

slides as PDF

Oliver Brock
(presenting a talk co-authored with Antonio Bicchi)

slides as PDF

Discussion 2 (12:00-12:30)

Towards effective grasp planning

 Break (12:30-14:00)

 Lunch break

Perception (14:00-15:00)

Michael Arbib

 

slides as PDF

Justus Piater

 

slides as PDF

Discussion 3 (15:00-15:30)

Towards appropriate perception

Break (15:30-16:00)

Coffee break


Panel Discussion (16:00-17:00)

All invited speakers + one moderator from among the organizers

Novel directions in grasping and manipulation (17:00-18:00)

Poster session with contributed and invited posters


Closure (15 mins)

 Organizers


List of questions to be discussed:

  • What kind of knowledge representation do we need to efficiently capture the understanding of the world?

  • How can we exploit human knowledge in an efficient way as a teaching tool?

  • Are we currently exploiting all the capabilities of the robotic hardware that we have, or the limitations that we experience come more from the sensing/ planning/ control/ execution/ supervision side?

  • How much should we rely on sensorial vs a-priori information?

  • Is machine learning the new paradigm for robotic manipulation?

  • Do we really need grasp/manipulation.task planning, or we can create a more natural robot-environment interaction?

  • What type of benchmarking do we need to compare advances in robotic manipulation?