Assignments
All
assignments must be turned in electronically via email to jcchuang@stanford.edu. Please
check the FAQ page frequently for explanations,
additions, clarifications, etc.
There will be three types of assignments in this course.
Written Assignments
The Warm-Up Project Assignment
The goal of the warm-up project assignment is to provide you with
the opportunity to familiarize yourself with robot data and
(optionally) the robot hardware, to the extent necessary for the
research project.
You are given data sets of robot sensor and motion data, along
with a map of a building. You task will be to implement a probabilistic
localization algorithm of your choice that enables you to localize
the robot relative to the map. The starting position of the robot is
unknown; thus, you have to solve a global localization problem. We will test your software using
new robot data, not available to you during software develoment.
Optionally, you are encouraged to implement your localization
algorithm on a physical robot, so that we can all enjoy watching your
robot localize itself in an office building (a map will be provided).
Research Project
You have to propose a small research project using robot data and,
optionally, physical robots. The research project has to involve an
implementation using real-world data, and a statistical algorithm that
should be a variant of the algorithms covered in class. To successfully
pass this requirement, you have to submit a written proposal for
approval by the instructor. You also have to submit a final project
report, which has to address (1) the problem you solved, (2) the
statistical technique including its mathematical derivation, and (3)
experimental results you obtained using your approach. Ideally, the
project covers interesting new ground and might be the basis for a
future conference paper submission. Consult with the instructor for
identifying interesting and doable research projects.
The instructor prefers to run this course paper-free, that is, all
assignments should be turned in electronically via email to jcchuang@stanford.edu
|
