Students taking this course for credit will work on a computer vision project. Projects can be done independently or in groups, but group projects are strongly encouraged.
This year's projects will center around the Intelligent Ground Vehicle Competition (IGVC) to be held June 10-12 in Michigan. Financial awards are given to the best teams, and travel expenses are paid. However, participation in the competition is optional.
Computer vision resources from the web may be useful in your project. Please also look at and contribute to the project information page.
An electric-propulsion mini-hummer (a golf cart with a hummer-shaped body)
has been donated by SAIC to the ME
department, and will be the competing vehicle. ME students will equip this with
a vibration-damped platform for the vision sensors. Sensors will include a
four-camera stereo system made by Tyzx, also
coming from SAIC. Two cameras are for far
field, two for near field. These are black-and-white. Other sensors (e.g., a
color camera, sonars) may be added if useful.
The plan is to compete at least in the design and navigation challenges, but possibly in the autonomous challenge as well. Projects are designed for that contingency. The competition is annual, so we can compete next year in events where we are not able to compete this year.
Here are possible projects. Additional or alternative projects will be discussed in class.
Students will also present a selection of DARPA tech paper(s), describing systems that competed in the recent DARPA Grand Challenge for autonomous vehicles, and IGVC tech papers, which serve a similar role for the IGVC.
By the beginning of class on the following deadlines, each groups sends a single PDF file by email to tomasi@cs.duke.edu.