Find and compare the best Artificial Intelligence software for your business.
Welcome Guest | Sign In
TechNewsWorld.com

Robots Meet DARPA Challenge Despite Pratfalls

By Richard Adhikari
Jun 8, 2015 3:08 PM PT

Robots from Korea and the United States staggered off with the top three prizes at the DARPA Robotics Challenge, held in Pomona, California, over the weekend.

Robots Meet DARPA Challenge Despite Pratfalls

In all, robots from 23 teams participated in the challenge, which consisted of an obstacle course simulating conditions similar to those following the Fukushima nuclear plant disaster.

The U.S. fielded 12 teams; the other 11 were from the Republic of Korea, Japan, Germany, Italy and Hong Kong.

The robots had to complete eight tasks relevant to disaster response, including driving on their own, walking through rubble, tripping circuit breakers, turning valves, using a tool to cut a hole in a wall, and climbing stairs.

They had to perform with degraded communications between their operators and themselves.

DRC-Hubo, the entry from South Korea's Team Kaist, took home the first prize of US$2 million; second prize went to Running Man, from Florida-based Team IHMC Robotics; and CHIMP, from Carnegie-Mellon University's Tartan Rescue Team, came in third.

DRC-Hubo completed the course in 44.5 minutes; Running Man in 50.5 minutes; and CHIMP in just over 55 minutes.

Stumbling Toward Sophistication

There were falls aplenty during the challenge.

"Falling down is not necessarily a bad thing, especially for events of this type," said Dan Kara, practice director for robotics at ABI Research, because it forces developers to come up with different types of fail-over approaches and make their systems more stable, lighter and more robust.

"What you're witnessing with the disaster-response robots is very similar to the first DARPA Grand Challenge (in 2004), which was designed to speed the development of autonomous driving technologies," he told TechNewsWorld.

None of the self-driving vehicles finished the 150-mile route. The greatest distance traveled was 11 miles, and the technology was "extremely expensive," Kara said, but "today, self-driving cars are logging hundreds of thousands of miles per year, using relatively inexpensive technologies."

Similarly, the Robotics Challenge "fosters the development of new technologies that let robotics systems operate autonomously in unstructured environments like the real world," he pointed out. "It also excites the imagination of the participants and the world, [which] will spur further innovation."

Saving the World?

Robots could be very useful in disaster rescue because they can go into hazardous places, reducing the risk to rescuers' lives.

The DARPA Robotics Challenge's winner, DRC-Hubo, "utilized an adaptable approach where it could use either wheels or legs for locomotion," ABI's Kara said.

There was talk last year of using roots to help combat Ebola, and the White House reportedly has discussed the possibility with experts.

Robots can be equipped to sense things not perceptible to humans, and they may be able to go places and, with their greater strength, do things humans cannot.

Miles to Go Before Robots Sleep

However, "much work" remains to be done, especially for systems with legs, Kara remarked. Manipulation "still requires much work. None of the robotics platforms exhibited [at the DARPA Robotics Challenge] have anything close to human-level capabilities."

The hardest thing for robots to do is navigate in an unknown environment, pointed out Jim McGregor, principal at Tirias Research.

"Even having a robot walk is a major challenge," he told TechNewsWorld.

Honda has developed one -- Asimo -- but "combining the ability to walk with the ability to avoid obstacles is exponentially more difficult."

The best potential for robotics may be to combine it with drone or aerial vehicle technology, as well as machine-learning algorithms.

"It takes a great deal just to train a robot to do specific actions, especially when some of the operating conditions are unknown," said McGregor.

Looking to the Future

"We are still evolving robotics around materials, design, programming, machine learning" and other areas, McGregor said. "We will improve the technology over time."

Many have voiced fears about artificial intelligence and self-directed robots -- the so-called Terminator scenario -- "but these concerns are akin to the same fears that we have," suggested McGregor, "and which continue to arise with other technologies, which comes from ignorance of the unknown."


Richard Adhikari has written about high-tech for leading industry publications since the 1990s and wonders where it's all leading to. Will implanted RFID chips in humans be the Mark of the Beast? Will nanotech solve our coming food crisis? Does Sturgeon's Law still hold true? You can connect with Richard on Google+.


Which type of online advertising is most likely to attract your favorable attention?
Straightforward display ads
Ads based on my interests
Informational articles on products/services
Video ads
Ads designed to grab my attention, e.g. pop-ups, autoplay
None -- I avoid all online ads