THIS IS A DRAFT SPEC; IT MAY CHANGE SOMEWHAT DURING THE TERM.
Robots today have been used successfully in many
domains, from exploring Mars and finding evidence of
water, to mapping the health of coral reefs, to
assisting long-distance drivers, to assembling cars.
In our course we will pursue a Grand Challenge approach
to robotics and create new robot bodies and brains. We
are motivated by tasks at the frontier of today's
robotic capabilities. We will develop solutions for
these tasks that are grounded in state-of-the-art
algorithms and systems science for robots. We will
implement these solutions and test them using a
challenge format.
Our robots will employ sophisticated techniques for
perception, navigation, and manipulation to cope with
unknown environments, negotiating intricate paths,
adapting their next move to obstacles, finding useful
objects in the environment, and using them to build a
structure. This work will provide our students with the
foundations for creating computer systems that interact
with the physical world, leading the way from PCs to PRs
(personal robots).
The grand challenge for this course is Build a Shelter
on Mars. Imagine a robot delivered to an uncertain
location in a remote and unknown environment such as the
surface of Mars, and given an uncertain prior map of the
local terrain. Imagine further that construction
materials, in the form of distinctively colored blocks
in a few discrete sizes, have been similarly delivered
and are scattered around the landscape. Some blocks
have ended up where intended (i.e., in known locations),
whereas others have ended up in unknown locations or may
have been lost or destroyed.
Your goal is to design and implement a robot (both its
body and its code) that can move about within its new
domain, collect blocks, transport them (all at once, in
small batches, or even one at a time) to some
autonomously-determined construction location, and
assemble them into a primitive shelter. The shelter may
range from a simple low wall, to a multi-level (stacked)
wall, to an "L" or "V" shape, to a room-like
structure.
The elements needed to solve the Challenge would
also enable other robot applications, ranging from
autonomous navigation with dynamic obstacles, searching
and rescuing victims at a disaster area, tidying up a
room, clearing the dishes in a cafeteria, delivering
packages in an office environment, and fetching
meal service items from a kitchen.
CAD models of the arm and claw can be found here: http://courses.csail.mit.edu/6.141/spring2012/pub/labs/CAD & here: http://courses.csail.mit.edu/6.141/static/pub/labs/CAD
|