Environment Segmentation in Service Robotics
University of New Brunswick, Department of Computer Science, Fredericton, NB, Canada
Bonn-Rhein-Seig University of Applied Science, Technical Report, 2011
@article{roscoe2011environment,
title={Environment Segmentation in Service Robotics},
author={Roscoe, M.S. and Pl{"o}ger, P.G.},
year={2011}
}
In the field of robotics a common problem is attempting to understand the world or environment in which the robot is operating. This is a common issue, as robots do not have an "intuitive" sense about its environment. Environment segmentation is a technique that is used to allow for the isolation of different parts of an environment for identification or interaction. Beyond having to deal with perceiving an environment that a robot is embedded within, there is the problem of having to do so in a reasonable amount of time. This means that there is a real-time constraint placed on this system. With service robots operating in ever changing environments the system must be able to perceive these changes fast enough to ensure that the robot has enough time to appropriately react to the change. The need to perceive an environment and to do so in a time sensitive manner is demonstrated in the RoboCup Service Robotics Challenge [1]. This challenge requires robots to perform various tasks autonomously, such as fetching a drink from the kitchen. While this task sounds simple in nature it requires not only knowing what a kitchen and dink are; but how to find them and how to recognize that it has found them. While performing this task the robot must also be aware of people passing in front of it, or a new obstacle that may appear, creating a hazard for the robot or the people working around the system. This paper will look at the ability to discover planar surfaces in an environment, as most objects within the environment we are concerned with are located on planes. The ability to locate planes provides the system with a searchable area within the environment. The focus on finding planes will allow the evaluation of the feasibility of several possible improvements to environment segmentation. This project will have a scope that is limited to looking at the RANSAC algorithm, the Microsoft Kinect as well as parallel technologies in order to improve upon environment segmentation
October 31, 2011 by hgpu
