Building A Robotic Beckham

LabVIEW and Apple Technology at Virginia Tech

Robots

At RoMeLa, the Robotics and Mechanisms Laboratory at Virginia Tech, graduate and undergraduate students working under Director Dr. Dennis Hong are using Macintosh systems and National Instruments LabVIEW (Laboratory Virtual Instrumentation Engineering Workbench) to turn out robotic creations that think and move by themselves. Among them are STriDER, which is six feet tall and walks around on three legs; MARS, a hexapod that strolls around on six; CLIMBeR, a cable-suspended robot that climbs unstructured cliffs; and IMPASS, a highly mobile actuated robot with spoked wheels. And then there’s DARwIn.

DARwIn (Dynamic Anthropomorphic Robot with Intelligence) is humanoid, autonomous, and made of precision-machined aluminum. He is only three years old, but moves and thinks independently. He comes in two versions, 16 and 24 inches tall, and is currently training to be a soccer player. He competed in the 2007 RoboCup competitions at Atlanta, Georgia, where teams from around the world put autonomous soccer-playing robots through their paces. RoboCup’s underlying goal is the development of humanoid robots for research purposes. But it bubbles with excitement just the same.

“This is a big deal,” says Hong. “The competition is intense and packed with energy.” The stated goal of RoboCup is to field a team of 11 robots capable of beating the human World Cup soccer champions by 2050.

TEAM:SPRInt, the student team that prepared DARwIn for RoboCup, did well in Atlanta. Of the dozens of teams that applied for the competition from all around the world, only 24 qualified, and TEAM:SPRInt (Soccer Playing Robot with Intelligence) was among them – the first and only team from the United States ever to qualify for the RoboCup humanoid division. DARwIn did not make it to the finals, but the somersaults and backflips of DARwIn 2A (the goalkeeper version) and the kicking power of DARwIn 2B (the striker version) made them clear crowd favorites as they played other teams. The team is honing DARwIn’s soccer-playing skills for the 2008 RoboCup in China.

LabVIEW: The Mind of DARwIn

DARwIn thinks with a 4-inch-square Pentium-powered PC/104 computer running the realtime version of LabVIEW, a graphical engineering software environment that RoMeLa uses for all robotics development. The other TEAM:SPRInt favorite is the Mac. The team uses a 20-inch and a 24-inch iMac, a pair of MacBooks, and a pair of MacBook Pros, all linked by Airport Extreme, for code development in LabVIEW. An Apple Xserve server with Xserve RAID provides a terabyte of data storage and helps keep track of the revisions of the developed codes for the robots.

“We started developing DARwIn in Windows,” says graduate student Sean Egger. “But you can port LabVIEW code from PCs to Macs and vice versa. Many of us had Macs and prefer Mac OS X to Windows, so we ported most of our DARwIn development software to the Mac version of LabVIEW. The Macs give us a stable operating system, and more importantly, a UNIX core, which makes things easier – such as TCP connections and serial port communication.”

LabVIEW code, written on the Macs and uploaded to the PC/104, is DARwIn’s brain. It receives inputs from his sensors and decides what to do next in terms of soccer-playing behavior. His eyes, for example, are two Firewire cameras mounted one above the other. The lower one is fixed, pointed at his feet to look for the ball. If that camera can’t find it, the top camera pans and tilts to search for it. The top camera also looks for the goal and for chalk lines to tell LabVIEW where DARwIn is on the field – a process known in robotics as localization. It also identifies teammates and opponents. DARwIn’s inertial measurement units (IMUs) measure and report acceleration in three directions, including the angular acceleration.

“One of the reasons we really like LabVIEW is that it will connect so easily and seamlessly to any kind of sensor or motor controller. We don’t get bogged down writing drivers. LabVIEW’s got packages that communicate with just about anything we’ve thrown at it.”

LabVIEW programming has given DARwIn some remarkable soccer skills. He can identify a ball, know where he is on the field, find the goal, and kick a ball through it. If he falls down, he can get up by himself. But he is not a one-dimensional personality. The lab uses him as a research platform for a variety of other robotics projects using LabVIEW. He can, for example, read handwritten notes and follow their instructions, interacting with people by playing dice, dancing, and shaking hands – all part of rounding out his humanoid capabilities.

1 2

LabVIEW Seminar

This seminar introduces you to the fundamentals of creating computer-based measurement and automation applications using the Mac and LabVIEW graphical programming. Learn more.

A Car with A Mind of Its Own

Dennis Hong’s student engineers, as part of Virginia Tech’s team Victor Tango, took their robotics expertise to the streets in the finals of the third DARPA Grand Challenge driverless vehicle race in November, 2007. This was the Urban Challenge – a race through city streets by driverless robotic vehicles.

“The vehicle had to obey all the traffic laws and respond to traffic lights and signs,” says Dennis Hong. “It had to drive in moving traffic and be able to park itself. First prize was $2 million. We took third place and won a $500,000 prize.”

Team Victor Tango’s driverless vehicle, developed with LabVIEW, relied on sensor data, including laser range finders, cameras, and IMU and GPS inputs, had to navigate through traffic, intersections, and multilane streets obeying traffic lights and directional signs.

“The previous Grand Challenge race was through the desert,” says Sean Egger. “This one was through town. The vehicles had to be completely autonomous – no driver, no remote control, no external communication at all. They had to think and act on their own.”

The Urban Challenge is part of DARPA’s response to an act of the U.S. Congress that requires that a third of all U.S. ground-based military vehicles be autonomous by 2015. The technology has applications in agriculture and other industries.

The Graphical World of LabVIEW

The NI LabVIEW environment that simplifies development of RoMeLa robots is based on a visual language called G. Student engineers use G to program graphically instead of writing lines of code. They build block diagrams on the screen, creating virtual instruments and tweaking values graphically instead of writing lines of code. LabVIEW on the Mac is incredibly intuitive; using LabVIEW’s visual library, a student with no prior experience in robot vision programmed the vision response of the DARwIn humanoid robot in just two hours.

LabVIEW, developed on the Mac 20 years ago, keeps growing in terms of its capability and easy-to-use interface. You’ll find LabVIEW wherever scientists and engineers need to take measurements.

“Most scientists and engineers are not computer scientists,” says Mike Neal, LabVIEW Product Manager for National Instruments. “They're used to expressing themselves in the flow-chart environment that LabVIEW provides, so it’s a natural fit for the way they think. It’s also oriented to collecting data, unlike general-purpose tools like C++.

“Labs used to look like science-fiction movie sets. Desktop computers are now so powerful that they’re replacing specialized equipment for development, and a lot of people prefer to use the Mac for that. It’s well shielded – which matters when you're taking measurements where other things are going on. And these people don’t want to trade reliability for performance. The Mac offers both, and is a great platform for that reason.”