[robocup-humanoid] Invited Session: Complex Motion Planning for Humanoid Robots (Code LQx83), Due 9th April 2007

[Jacky Baltes]

Hi,

it is not too late to submit a paper to the special session at IROS 2007.
The deadline has been extended to April 9th. 2007.

CU,
   Jacky

On 3/17/07, Jacky Baltes <jacky at cs.umanitoba.ca> wrote:
>
> Hi,
>
> the session "Complex Motion Planning for Humanoid Robots" has been
> accepted as an invited session at <a href="www.iros2007.org"> IROS2007</a>.
>
> Even though we have four submssions to this session already, we can
> accommodate more if so desired.
>
> If you are interested in taking part in this session, then please submit
> your paper by the deadline via the IROS conference management system.
>
> The conference uses the paperplaza system for handling submissions, so you
> will need an account and PIN on the system to make your submission.  If
> you
> do not have a pin for the system already, you can create one here (click
> on
> the "Register a New Pin" choice):
>
> https://ras.papercept.net/conferences/scripts/pinwizard.pl
>
> If you have a pin already, the main page is here:
>
> https://ras.papercept.net/conferences/scripts/start.pl
>
> When your paper is ready, upload it to the iros conference by following
> the
> instructions on the site.  There is a separate choice for submitting
> papers
> that have already been tied to an invited session, and that page will ask
> you for a session code.  The code for our session is LQx83.
>
> Again, please remember that IROS's deadline is the 28th of March.
>
> Session Proposal:
> Complex Motion Planning for Humanoid Robots
>
> Humanoid robots are an active research area within the robotics
> community.  To date, however, most work in this area has focused on
> the generation of basic walking gaits for flat terrain. Walking is
> obviously a core competency for a humanoid robot, and a skill on which
> more challenging behaviors such as running are based, but it is still
> a single task with a much more limited control scope than the range of
> motions a human can easily generate and control. While a mastery of
> walking control can form a basis for more sophisticated behaviors, it
> is also important to more directly attempt broader forms of complex
> motion planning without necessarily basing this control directly on
> walking behaviors. A human can crawl through a complex series of
> randomly placed hazards on an unstructured obstacle course, for
> example: scaling walls, moving to a kneeling position, crawling under
> overhead structures, and planning a series of these motions
> together. Assuming that the control used in a basic stable walking
> gate can scale to these more difficult problems is very similar to
> scaling problems that other areas of artificial intelligence have had
> to face in the past. While a true human obstacle course is an extreme
> goal for a humanoid robot, these same skills are demonstrated in
> everyday simple tasks such as picking up a ball that has rolled under
> a table. In order to develop sophisticated control mechanisms, these
> more complex forms of motion planning must be considered beyond
> attempts at scaling walking behaviors. A range of such complex motion
> planning should also be expected of any reasonably adaptive humanoid
> robot.
>
> The reason that basic walking gaits have been the major focus thus far
> is not solely a narrowness of focus, but has also been dictated by
> hardware limitations.  Until recently, mobile processing units have
> not had the power to adequately drive sensing and control for
> humanoids performing in complex, unstructured domains. To some degree
> this is still true for high-bandwidth senses such as real-time vision,
> but current mobile processors can now be expected to support a broader
> array of sensory and effectory equipment, and correspondingly support
> more sophisticated motion control. Similar developments in affordable
> sensory equipment have also occurred: accelerometers, gyroscopes, and
> force-feedback sensors suitable for small humanoid robots are now
> reasonably affordable. At the same time, better algorithms are being
> produced to allow vision and other forms of sensing to further support
> more complex motion planning. Together, these factors raise the
> fields' expectations of the motion planning capabilities of modern
> humanoid robots.  There are also a number of humanoid robot
> competitions (the most well-known of which are FIRA and RoboCup),
> which encourage the development of completely operational systems that
> concretely demonstrate approaches to motion control. FIRA, especially,
> has worked hard in recent years to encourage work in adaptive motion
> and complex combinations of tasks. These include adaptive
> lift-and-carry tasks over uneven terrain, requiring compensation for
> both terrain change and shifting weight, as well as obstacle runs that
> include local navigation, obstacle avoidance, stepping over obstacles,
> moving around obstacles, and crawling under obstacles.
>
> This session will consist of papers that explore beyond simple walking
> gaits to demonstrate fully-implemented complex forms of motion
> planning and feedback control for humanoids, such as actively
> controlled climbing, lifting, and crawling motions in unstructured
> domains. The intent of this session is to bring solid research work
> typical of that expected of IROS, with a concrete grounding in
> complex, unstructured domains, and an emphasis on generality and
> adaptivity in complex motion planning tasks.
>
> --
> Jacky Baltes, EITC E2-402 Department of Computer Science, University of
> Manitoba Winnipeg, Manitoba, Canada R3T 2N2
> Phone: +1 (204) 474-8838, Fax: +1 (204) 474-7609
> Email: jacky (AT) cs.umanitoba.ca
> http://avocet.cs.umanitoba.ca




-- 
Jacky Baltes, EITC E2-402 Department of Computer Science, University of
Manitoba Winnipeg, Manitoba, Canada R3T 2N2
Phone: +1 (204) 474-8838, Fax: +1 (204) 474-7609
Email: jacky (AT) cs.umanitoba.ca
http://avocet.cs.umanitoba.ca
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.cc.gatech.edu/pipermail/robocup-humanoid/attachments/20070329/8f465186/attachment.html