Document Type

Honors Project

Abstract

Scientists have been working to create robots that perform manual work for years. However, creating machines that can navigate themselves and respond to their environment has proven to be difficult. One integral task to such research is to estimate the position of objects in the robot's visual field.

In this project we examine an implementation of computer vision depth perception. Our application uses color-based object tracking combined with model-based pose estimation to estimate the depth of specific objects in the view of our Pioneer 2 and Power Wheels robots. We use the Camshift algorithm for color-based object tracking, which uses a color-histogram and gradient density climbing to find the object. We then use the POSIT algorithm, or Pose from Orthographic Scaling with ITerations, to estimate the pose of the object based on a model. Although pose consists of a translation vector and rotation matrix, we are only concerned with the z-component of the translation vector as it is an estimation of depth.

Our results show that distortions caused by the environment result in inaccurate distance estimations. These distortions include reflections of the object off surfaces such as a linoleum floor. However, when no such distortions are present, the application is rather well behaved and returns consistent, accurate depth estimations.

An advantage of the application is that it can be run in real-time for both standard web cameras and more expensive higher quality cameras. It can also be used with stored image files.

Share

COinS
 
 

© Copyright is owned by author of this document