Date of Award
Santa Clara : Santa Clara University, 2017.
Christopher A. Kitts
Robotic systems, when used in real world situations, often have many control modes that need to be switched between in real time. These control modes may be addressing the robots in many different ways and may be executing in different locations. Conventional robot control software platforms do not provide a flexible way of switching between all of the different types of control modes in real time. In this paper, we present a system architecture that seeks to solve this problem of real time controller switching using industry standard software. The architecture implemented is one that allows for robot control regardless of hardware, and ensures continuity of control when undergoing arbitrary controller switching. Our experiments show that this software can be used to switch between types of controllers, location of controller execution, and single and multi-robot controllers. They also show that switching controllers takes 10ms, which is well within our requirements. We conclude that our system is useful in its present state for smaller projects at research institutions, or as a stepping off point for future work. To improve this system for general use, improvements beyond solely sending messages through Wi-Fi will be necessary. It is also recommended that this system be transferred from Indigo, the current distribution of ROS used, to Kinetic the newer distribution of ROS, because Indigo is nearing end of life. Despite these recommendations, the success of this project shows how important this type of system will be in continuing robotic development.
Head, Ethan and Fattor, Addison, "Multi-Robot Cluster Control" (2017). Electrical and Computer Engineering Senior Theses. 39.