Published October 23, 2012
Who wouldn’t want a MacGyver-bot, right? Count U.S. military in.
A research team led by Georgia Institute of Technology professor Mike Stilman, a rock star in robotics, says its "MacGyver" robot can use the everyday objects it comes across to accomplish high-level tasks. A new promotional video for the school exclusively revealed by FoxNews.com shows the potential for this robot and other next-generation bots.
The team hopes this MacGyver-like bot will perform creatively in intense, high-risk situations such as natural disasters or combat.
“Our goal is to develop a robot that behaves like MacGyver, the television character from the 1980s who solved complex problems and escaped dangerous situations by using everyday objects and materials he found at hand,” said Stilman, an assistant professor in the School of Interactive Computing at Georgia Tech.
The Navy hopes so too. Stilman’s team has received a three-year, $900,000 grant from the Office of Naval Research to continue its work on this robot, which currently goes by the unusual name Golem Krang.
The secret to MacGyver skills
So how can a robot learn to use tools? An algorithm Stilman is honing breaks such behavior into three parts. The robot identifies objects in its vicinity, works out what that thing could potentially do, and finally uses it to solve a problem or complete an objective.
For reasoning, perception, inference, performance and learning, Stilman is collaborating with researchers Pat Langley and Dongkyu Choi who will further develop their ICARUS cognitive architecture.
With their help and his new algorithms, MacGyver will be an evolutionary leap past Stilman’s current Krang, a humanoid designed and built by the Humanoid Robotics Lab at Georgia Tech with Schunk Inc.
Krang is a ”massive beast” in Stilman’s words: Weighing in at 300 pounds, each of his arms weighs more than 100 pounds.
Typically, these immense humanoid robots can only move soda cans.
Stilman and team have built a mini 11-pound version of the robot that can push 33 pounds -- a whopping three times its body weight.
If they can design a human-sized bot, it should be capable of astonishing physical feats, including a goal of moving one ton.
Golem Krang has broken ground by being able to stand by itself and then balance while interacting with its environment.
Currently, robots don’t really do that; instead people tell them what to do. MacGyver-bot would instead be given a basic knowledge of mechanics and simple, machine-like tools so that it can work things out on its own.
It could approach a door, realize that it wasn’t strong enough to move an obstruction blocking the way, and seek out a lever in its environment to get through.
A fully developed MacGyver-bot could stack boxes to climb over a wall, build bridges to get across gaps or make a ladder to reach something high – all cobbled together from random things in its environment.
“[It will use its] whole body and accomplish whatever the task requires. My goal is to make a robot that’s really useful,” Stilman said.
Previously, his work focused on navigation among movable obstacles, robots that could identify impediments and independently shove them out of the way. The MacGyver-bot would take this a step further, considering how he could use the obstruction rather than just get rid of it.
“Once a robot gets there, how does it use its body, use the environment to make things happen -- that’s exactly where MacGyver comes in,” Stilman said. “The concept behind MacGyver… is the idea behind how we think about objects. Do we think of objects as obstacles or tools?”
Ultimately, Stilman’s intelligent, resourceful robot could work side by side with soldiers on challenging missions.
But that’s just the vision. MacGyver-bot can’t yet use a paperclip and bubblegum to escape dangerous situations.
But just wait until next season.
Ballet dancer turned defense specialist Allison Barrie has traveled around the world covering the military, terrorism, weapons advancements and life on the front line. You can reach her at firstname.lastname@example.org or follow her on Twitter @Allison_Barrie.