Undergrad Research Project - Mobot without vision

Fall 2007

Kyunghwan Choi
Philip Koopman
Project description

Introduction As the popularity of GPS and navigation grows, so does the interest of self-navigating robots. Carnegie Mellon University has been sponsoring mobile robot (mobot) races since 1995 and has been actively involved with DARPA challenge through the red team. The objective of this research is to build a mobot without using vision.

Problem Most of the mobots are camera-driven. They try to identify the course through the camera and navigate through the white line. As this may work within a small environment, under harsh weather conditions or in a larger course where there is no path to follow, it will most likely fail. In case of a Sandstorm or Highlander vehicles built by the red team, they use other technologies such as GPS, laser sensors, radar, etc.

Purpose of the Study If we design a mobot without vision, we may encounter problems such as less accurate acknowledgement of the surrounding environment at the runtime. But with other alternatives such as using accurate map information and solve navigation problem, using LEDs to verify the course, or using tiltmeters to find the exact position, the mobot can be less vulnerable to the weather conditions or unexpected loopholes that computer vision can introduce.

Return to project list