UCLA Engineering: UCLA Engineer Magazine

Students Advance Sensor Network Technology through Gaming System

Ragobot team members (from left): David Lee, Jonathan Friedman, Parixit Aghera and Advait Dixit.

“I need a robot. Build me one that can play games,” asked electrical engineering graduate student Aman Kansal, and with that, the Real Action Gaming Robots, or ragobot, project was born.

Kansal and an interdisciplinary team of graduate student researchers in electrical engineering professor Mani Srivastava's Networked Embedded Systems Lab designed and built a team of robots and reconfigurable terrain for a mobile gaming system that addresses many questions common to embedded sensor networks. They are the first team to apply advanced embedded network technology to a real-world version of strategy games like Warcraft or Age of Empires.

“There are three main purposes to the system we are building,” explained computer science graduate student Parixit Aghera, “entertainment, education and testbed research. It’s a more realistic deployment of autonomous robots than you could achieve in simulations, and we’re using the momentum from the project to interest more people in sensor networks.”

Ragobot is a novel testbed for embedded sensor networks and robotics for researchers concerned with actuation and controlled mobility issues. The team is constrained by real world parameters in trying to create a sensor network that will allow them to play games in real time.

“Almost all sensor network applications – monitoring forest fires, following bacteria contamination in the ocean, tracking vehicles in a field – involve the same command and control paradigm you see in strategy games,” explained electrical engineering graduate student researcher Jonathan Friedman. “You have limited resources under your control that you would like to deploy in such a way as to optimize the territory you can see and react to.”

A first generation ragobot.

Autonomous robots bring several advantages to sensor networks, including enhanced area coverage, redistribution of resources, improved communications and fault detection, and the ability to move closer to phenomena of interest.

Distributed motion planning, or SWARM mobility, is another area of interest to the research team, who are among the first to develop the necessary hardware and software. They are developing ways to help the robots determine their own location through interaction with one another, which will allow a team of robots to move from one area to another without having to rely on any reference points other than themselves.

The robots designed by the team are considerably less expensive than commercial robot models, but have greater mobility on inclines, uneven terrain, and vertical faces. The robots are able to climb over vertical obstacles up to an inch in height.

The second-generation ragobot, which has the assets of the original model, gains considerable functionality. Among the sensors the team has added is the newly available XYZ sensor node – designed by a UCLA alumnus now at Yale – an open source wireless sensing platform that offers improved processing capabilities.

The next-generation robot has more than 200 components, including a circuit board with audio and infrared sensors, a camera capable of recording stills and video, as well as additional processing power. The device has nearly 45 feet of wiring in just 17.5 square inches of space.

An ultrasound localization system, which operates like a GPS system, will help the second-generation robots determine their locations in relation to one another and the terrain.

“One of the things we’ve included in the ragobot, which most robotics and embedded systems don’t have, is an RFID reader that will allow it to retrieve information from RFID tags in the environment. Such a system will be helpful for object identification, and communication,” said David Lee, a graduate student in electrical engineering. “Sometimes a signal isn’t strong enough to allow for communication between robots, but with RFID tags, they could leave information in the environment for other robots to retrieve later.”

A first generation ragobot.

In addition to the robots, the team, aided by a group of undergraduate students, has built a reconfigurable three-dimensional terrain that challenges the robots’ navigation skills with obstacles such as those a mobile sensor might encounter in the real world. The trees and other objects are made from a nickel-titanium shape memory alloy that can respond to electronic signals from the robots, and were built last summer by Sophia Wong, a materials science and engineering senior.

The project is also being used in the classroom. A course project for graduate students was offered in computer science to help them learn embedded actuator platforms. And it’s also being used in an undergraduate course taught by Professor Bill Kaiser to introduce the students to sensor networks and robotics.

“We’re developing algorithms and a programming environment in which you can easily give a task to a set of robots and they’ll take care of how to do it,” said Aghera.

Added Lee, “It’s part of making it easy and readily available for people to use so they don’t have to learn a complex programming language.”

For additional information about the project, please visit http://www.ragobot.com.

- Marlys Amundson