[CoLoCo] CU CS Colloquium; Dec. 13 on robots and wifi in tight places

[Neal McBurnett]

The weekly colloquia of the Computer Science department at CU Boulder
are great fun.  Here is one example which I'll send you in case you
want to come or subscribe.  I was inspired to forward this based on
Andrew's expertise in robots.

 "Wireless Video Sensory Networks Over Bluetooth For Small Search and
 Rescue Robots"

Neal McBurnett                 http://mcburnett.org/neal/

----- Forwarded message from Pat Warrick <Patricia.Warrick at Colorado.edu> -----

To: Colloquium ListServe <cs-colloquia at lists.colorado.edu>
From: Pat Warrick <Patricia.Warrick at Colorado.edu>
Subject: CS Colloquium; Dec. 13, 2007:  Richard M. Voyles

University of Colorado at Boulder 
Department of Computer Science
ECOT 717 Engineering Center
Campus Box 430
Boulder, Colorado 80309?0430
(303) 492?7514, FAX: (303) 492?2844
CS COLLOQUIUM
Wireless Video Sensory Networks Over Bluetooth
For Small Search and Rescue Robots
 
Richard M. Voyles
University of Denver
Thursday, December 13, 2007
3:30-4:30 p.m.
ECCR 265
 
Urban Search and Rescue (USAR), which generally implies collapsed
structures, presents a challenging environment for robotics research
because of profound difficulties in sensing, mobility, and
communication. Sensing is difficult due to poor lighting, high
clutter, the lack of engineered landmarks, and airborne
particulates. Mobility is difficult due to the quantity and variation
of rubble, the variation in surfaces and obstacles, and the need for
small size to access small spaces. Communication is difficult because
wireless is hindered by concrete, steel, and water; antennas are
de-tuned by close proximity to unpredictable ground planes; and
tethers get snagged in the environment.
 
In this talk I plan to touch on aspects of all three of these
difficulties. The core of the talk discusses our efforts to develop
multi-hop ad-hoc networking over Bluetooth for urban search and rescue
robots. We completely abandon the idea of single-hop communication as
unsuitable for the environment and embrace multi-robot
networks. Multi-robot teams can solve the communication problem, the
search coverage problem, and the search speed problem if an
appropriate infrastructure can be developed. (We'll ignore, for now,
the problems multi-robot teams create.) We choose Bluetooth over
Zigbee because its higher bandwidth is better suited to the high
resolution imagery required for USAR and the power budget and
complexity are better suited to small, resource-constrained robots
than WiFi. We have developed a hybrid routing protocol that combines
the latency benefits of a proactive protocol (cluster head gateway
switch routing) when a broken link is discovered with the overhead
benefits of a reactive protocol (similar to ad-hoc, on-demand distance
vector routing) in a highly volatile network. We show preliminary
results in simulation and implementation that this new hybrid protocol
outperforms other protocols for sparse, highly volatile networks
required for the USAR domain.
 
To put this work in context, I will also discuss the TerminatorBot,
the soda can size, two-limbed, crawling robot we use for exploring
voids and difficult-to-reach spaces.  Its small size means it must be
very agile to deal with the great variability of the environment. To
enhance its self-adaptability, we are developing a reconfigurable
computing platform that allows both hardware and software dynamic
reconfiguration in response to unknown terrain.
 
Computer Science Colloquia Mailing List
Subscribe/Unsubscribe at:
http://www.cs.colorado.edu/events/colloquia/mail/
Sign language interpreters are available upon request.
Please contact Pat Warrick at least five days prior to the colloquium.

----- End forwarded message -----