Robotics on a roll: Researchers unveil Ballbot

Legs may now be obsolete — at least for robots. After a team of Carnegie Mellon researchers recently unveiled a new robot that requires neither the use of legs nor wheels.

Instead, the mobile robot, dubbed Ballbot by the team, moves by balancing on an aluminum and polyurethane sphere.

Ballbot is the creation of the Robotics Institute research professor Ralph Hollis, who built his first robot in 1957.

“When I was in high school I read an article on them in Scientific American,” said Hollis. “I thought they were the coolest things.”

But there is very little resemblance between Hollis’s new robot and the robots he originally read about.

Inverted mouse-ball drive

Ballbot uses an “inverted mouse-ball drive mechanism,” as coined by Hollis and his team.

This is because the mechanism is the exact opposite of how a computer mouse ball works. “It’s not necessarily new, but it’s novel,” said Tom Lauwers, a doctoral student in robotics, who was part of the team that developed the robot.

Instead of the mouse ball driving rollers inside the mouse and relaying the information to the computer, the onboard computer relays information to a set of rollers around the ball that allow for Ballbot to begin to move.

Ballbot uses a 200-MHz Pentium processor, with which it can perform onboard computations, and the team can communicate with Ballbot via an 802.11b wireless link. This is an upgrade from Hollis’s first robot built in 1957, which used telephone relays as a brain.

Timing belts connect the ball’s driver rollers to two high-torque servomotors. Opposite the drive rollers are two passive rollers that apply force to the ball’s equator in order to make sure that the ball remains in contact with the drive rollers.

Hollis designed Ballbot to be tall and slender. The robot’s physical appearance mimics that of a human. As a result, its center of gravity is much higher than traditional mobile robots, which often have wider bases.

The robot’s size and shape are not by accident. “Ballbot is roughly the size of a person because we want it to work in places where there tend to be a lot of people,” said Hollis.

Team Ballbot

Hollis first came up with the concept of Ballbot approximately 10 years ago in his garage, but did not start work on the robot until five years later.

“I decided that I would just do it. So I sat down in my garage by myself and machined out the parts,” said Hollis.

Hollis then looked into financing his new idea. “After that, I wrote some proposals to the National Science Foundation, and one finally got accepted.”

With the money granted from the NSF, Hollis was able to recruit a team to work on the robot.

The team included Robotics Institute project professor George Kantor, as well as graduate students Tom Lauwers, Anish Mampetta, and Eric Schearer.

“The first year was relatively frustrating because there was a lot of conceptual work to be done,” said Lauwers, who was first to work with Hollis on Ballbot.

The team said that an area of particular frustration was the construction of the actual ball.

“The ball itself was not already available,” said Hollis. “It’s made of aluminum and polyurethane, and it required a lot of trial and error. It’s taken us several tries to get a ball that works well.”

Even so, Lauwers, Mampetta, and Schearer all felt that getting the opportunity to work on Ballbot has been a good experience.

“It takes a long time to make progress when you start a project,” said Schearer, who joined the team last year. “At least for me, there was a lot of figuring out how things work.”

Anish Mampetta also joined the team a year ago after leaving India to attend Carnegie Mellon.

“When I first heard about what Ralph was doing, there were no pictures,” said Mampetta, “but when I first saw it balance I was completely taken off my feet.”

Ballbot’s future

Ballbot is expected to undergo a series of changes in the future. The team plans to add a head and arms to the robot down the line.

Before the appendages can be added, however, the team must run simulations to determine how attaching the arms or head would affect Ballbot’s balance.

Hollis said that the team also wants to add another motor to the robot so that it can turn toward the direction that it moves. As for now, the robot can face only one direction.

When asked if he hoped that Ballbot would eventually be able to serve a particular purpose to the public, Hollis said, “We’ll probably never get it to that stage by ourselves.”

The team does hope to understand more about how the robot will interact with people and furniture in its environment.

With their research, the team may find that the inverted mouse-ball drive is a big advancement in robotics. But for now, there is no plan to mass-market a line of robots.

“At this point building robots is more of an art than a science. It’s really hard to generalize the process,” said Hollis.