Development of an Open Source Software Package for Autonomous Robotic Docking Using Stereoscopic Imaging

The goal of our project, Development of Machine Vision Algorithms for Cooperation in a Multi-Agent Robotic System, is to create and implement algorithms to enable a robot to find and dock itself with another vehicle system.The intent of this system is to create a framework to allow for robotic missions to be decentralized by separating the various sensors from the rover itself.Those sensors can be placed on an inexpensive chassis without control or movement systems, leaving the actual rover system to be a much simpler device only needing equipment to move itself and a sensor package.Currently a matching pair of IEEE1394 digital video cameras are used on a PC running a combination of open source image processing software and our target tracking and control C code.Our code can currently detect a target in a cluttered scene, and then find the direction to the target.The code to plot the course to the target will be ready shortly, and then will be combined with our motor control code.Once the framework has been implemented and tested on our current hardware, work will then shift to making the software easier to adapt to various robot and camera systems, and, once it is available, allowing our code to work with NASA's CLARAty software.This will allow us to offer a package of free software that can be integrated into any robotic system to provide a ready to use docking and control mechanism.