Despite a rapid development in computers and sensor technologies, surprisingly few autonomous robot systems have successfully made it to the consumer market and into people's homes. Robotics is a popular topic in research circles, but focus is often on ground-breaking technologies, and not on putting the robots on the commercial market. At the time when this research project was started in May 2010, the amount of successful commercial applications based on mobile robots was very limited. The most known applications were vacuum cleaners, lawn mowers, and few examples of specialized transport robots used in warehouses and hospitals. At the same time, despite attempting to solve the same tasks and applications, the resulting software and products of research groups was very fragmented. Even if being open source, the software was based on self-made frameworks and often only used internally by the individual groups and perhaps a few close industrial partners. This research project addresses the problem of increasing the potential for more commercial applications based on mobile wheeled robots. Therefore the main focus is not on inventing new ground-breaking robotics technology, but instead understanding why the existing technology and algorithms are not ready for production. One focus area of this project is an analysis of these existing Technologies and algorithms for mobile robots. The most fundamental task for a mobile robot is navigation, yet no generic and ready to use implementation for solving this exists. This project includes an effort towards such a generic navigation system. It should provide a stable and easy-to-setup experience for robotics researcher and industry integrators who needs navigation capability for a specic mobile robot. At the same time a common package for navigation will provide a base for many researchers to contribute to and mature over time. The second focus area was to close the gap between research and industry by providing the necessary tools and motivation for researchers to create more robust prototype applications. During the time of the project period, a signicant research community was created around one specific robot control framework called ROS. From the very beginning,this research project acknowledged the value of such a community, and put a significant eort into in uencing the ROS framework to become usable also for industry and commercial applications. Based on a requirement analysis for such a framework, a prototype implementation of an industry ready component based ROS compatible middleware was created. The project also includes work towards a smart parameter framework, assisting in conguring the individual components in a component based control framework. The smart parameters adapt to the respective robot, and makes it possible to reuse advanced software components, without Expert knowledge about the underlying algorithms. The smart parameters also assists in building a robot system, that can autonomously calibrate and optimize itself.
Main Research Area:
Koed, Kakob, Andersen, Nils Axel, Ravn, Ole
Technical University of Denmark, Department of Electrical Engineering, 2013