[robotics-worldwide] Ph.D. Position: Humanoid Walking on Deformable Terrains

Adrien Escande adrien.escande at gmail.com
Tue Sep 3 07:36:27 PDT 2013


Topic: Humanoid Walking on Deformable Terrains

Humanoid walk planning and control are studied both in robotics and in
computer graphics animation. Currently, virtual avatars and poly-articulated
robots such as humanoid robots can move on horizontal flat surfaces,
slightly sloped ground or climb stairs. The recently demonstrated walking
gait generation and footprints planning in humanoid robotic locomotion are
still primarily dealing with rigid (but not necessarily flat) supports in
the environment. Humans and animals however are capable to walk on
deformable terrains such as sand, mattress, etc. and even on highly
deformable ones. We distinguish two types of deformable terrains: the first
ones do not restitute any energy after deformation (e.g. sand or mud), the
second ones store energy when deforming and restitute (at least) part of it
when no load are applied on them (e.g. mattress or trampoline).

In the frame of the forthcoming KoroiBot FP7 EU project (starting in October
2013), the topic of the thesis is to make current humanoids capable to walk
on such deformable terrains, by developing adequate models and controllers.
One particularity of the KoroiBot project is to measure human strategies
that may possibly inspire the strategy of the controllers. This study will
be performed at one of the KoroiBot team and the interested candidate will
be involved in the recording and the interpretation of motion capture data
of human walking on different deformable terrains. These data can be
exploited to design the control strategy for humanoid walking. Once
implemented and experimented (under similar human measurement conditions),
we will compare the results and assess the robustness and the quality of the
humanoid robot walks.

These research and developments must be demonstrated in challenging
scenarios involving the HRP-2 and HRP-4 humanoid robots.

Locations: the thesis will be conducted at the CNRS-AIST Joint Robotics
Laboratory, UMI 3218/CRT at Tsukuba, Japan and also partly at the CNRS-UM2
LIRMM at Montpellier, France. 3 months shall be spent at one of the
partners’ laboratory to conduct human-motion capture data.

Interested candidates shall send a detailed CV with a short motivation
letter to the following advisory team:

Advisory team
Adrien Escande adrien.escande :at: gmail.com
Mitsuharu Morisawa m.morisawa :at: aist.go.jp
Shuuji Kajita s.kajita :at: aist.go.jp
Abderrahmane Kheddar kheddar :at: gmail.com

View this message in context: http://robotics-worldwide.1046236.n5.nabble.com/Ph-D-Position-Humanoid-Walking-on-Deformable-Terrains-tp5708489.html
Sent from the robotics-worldwide mailing list archive at Nabble.com.

More information about the robotics-worldwide mailing list