Active self-landmarking to mobile robot camera calibration through wireless communications

Glenn Harden, Purdue University

Abstract

Mobile robotic platforms are becoming more and more popular, both in scientific research and in commercial settings. Robotic systems are useful for going places or performing tasks that are not suitable for humans to do. Robots are often able to precisely perform complicated or dangerous tasks with little or no human involvement. However, before a mobile robotic platform is able to be deployed, it must have a way of identifying where it is in relation to objects and obstacles around it. Often, this is being performed by using a visual system, such as a camera. However, just wiring a camera onto a robot is not sufficient. With many of the tasks that are given to a robotic system to perform, a great deal of precision is required to satisfactorily complete these tasks. This precision requires that the robot be given accurate information by the camera. Most cameras have minor imperfections, even though they are within the manufacturer's tolerances. Some of these imperfections can cause the aspect of the image to be slightly distorted, causing a perfectly square object to appear to be slightly rectangular. This aspect can be corrected with an appropriate scaling factor, reducing or enlarging the size of the image in the horizontal or vertical direction. Other imperfections in the camera manufacturing process can cause the image sensor to not be exactly aligned with the actual camera housing. This can cause the center point of the image to be slightly off of the center of the image sensor (e.g., the image could be shifted to the left by a few pixels, so the center of the image is actually at 317, 240 instead of 320,240). This project will implement a camera calibration algorithm that was developed in [1] on mobile robotic platforms to autonomously determine the center point and correct scaling factors for the platform's respective cameras. These mobile robotic platforms were chosen to be the National Instruments Robotics Starter Kit DaNI. They were programmed using National Instruments graphical programming environment LabVIEW. A robotic platform is to determine the center point and scaling factor of its own camera, using other identical robotic platforms to perform this calibration. The robots can then trade responsibilities, and a different robot can perform the required calibration. This project will also attempt to use wireless communication between the robots, allowing cooperative behavior. This project will also attempt to use the wireless signal strength to determine the locations of the mobile robotic platforms relative to one another.

Degree

M.S.

Advisors

Liu, Purdue University.

Subject Area

Computer Engineering|Electrical engineering|Robotics

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS