The DARPA Robotics Challenge is a competition funded by the US Defense Advanced Research Project Agency (DARPA). DARPA itself was started by President Eisenhower in 1958 as a response to the Soviet Union’s launch of Sputnik, and has since been instrumental in some huge tech advancements like virtual assistance and speech recognition technology.

In 2012, DARPA started a robotics competition aimed at developing autonomous or semi-autonomous robots capable of assisting humans in responding to natural and man-made disasters — one of the program’s key missions — and has since expanded the program to make robotics software and more accessible solutions long after the competition ends.

Next week, I’ll be covering the DARPA Finals in Pomona, California, for TechRepublic. To brush up on your DARPA and disaster robot knowledge, read this brief background and summary of what you should know about the competition, and what to expect next week.

1. It began with the desire to improve humanitarian assistance and disaster relief

The Fukushima disaster in Japan in 2011 was an inspiration for the competition, according to Dr. Gill Pratt, the DRC program manager. The team realized we never know what the next disaster will be, but we need technology to help us better address these types of disasters with better tools and techniques. And robots have massive potential.

“The particular part that we’ve chosen to focus on, here, is technology for responding during the emergency part of the disaster during the first day or two,” Pratt said in a media briefing several weeks before the competition. “So this is not about, for instance, robotics for doing the restoration of the environment many, many weeks, years after the disaster, but rather the emergency response at the beginning.”

2. The key challenge is communication degradation

One of the biggest issues focuses on the lack of communication immediately following a disaster. Robots, of course, can travel and work in areas that may be too dangerous or toxic for humans, and so there is a lot of potential for them as responders. The idea is to improve the way robots and humans interact when the communication is degraded. To simulate this, communication is turned off during part of the challenge. The robot goes inside a utility vehicle, it drives around to get to a site, has to get out of the vehicle on its own and go to a particular location. It cuts a hole in the wall and then has to perform a surprise task, Pratt said There are bstacles like climbing, moving debris, exiting, etc. There are eight total tasks worth one point each, according to the rule book:

1. Drive the vehicle

2. Egress from the vehicle (get out of the vehicle)

3. Open door and travel through opening

4. Open valve (similar to one of the three valves in Trials)

5. Use a cutting tool to cut a hole in a wall (similar to one of the two tools and the wall in


6. Surprise manipulation task (not disclosed until Finals)

7. Traverse rubble – Either cross debris field (by moving the debris or traversing it, similar

to Trials) or negotiate irregular terrain (similar to Trials)

8. Climb stairs (fewer steps and less steep than in Trials)

There’s no human help, and a degraded communication link between the robots and the human controllers, so the robots have to have some sort of autonomy.

3. The level of robot autonomy is still pretty low

“The state of autonomy in robotics these days is actually only beginning, and so the level of autonomy we’re talking about is at the task level, not at the mission level,” Pratt said. Robots that can open a door on their own without a human operator need to have humans that tell it what to do beforehand. The robot can then open the door and then wait for the next communication from humans.

That means a lot of preparation and work before the competition because teams need to develop really intelligent robots even for a fraction of a minute worth of autonomy. And most of the software development of the AI goes into the human interface, the computer that allows operators to visualize what is going on.

4. The competition has had three phases

The first, in June 2013, was virtual. There were 26 teams from all over the world. In December 2013, there was a physical challenge called DRC Trials with 16 teams. That was really the first run of the simulated events, which went so well, Pratt said, they decided to expand the program, and added some international teams. Now the total that will compete in the finals is 25 teams. About half are from the US. There are five teams from Japan, three from Korea, two from Germany, one from Italy, one from Hong Kong, and one from China.

The finals, free and open to the public and held on June 5-6, will be exceptionally harder than the previous competitions. Robots can’t be recharged and they have to run on their own power. Nothing can be used to help hold them up, and many are expected to fall and become injured. The entire mission lasts an hour, and the points are based on how many tasks each team completes. In total, there are $3.5 million of prizes. The first prize is $2 million, second is $1 million, and third place is $500,000.

5. DARPA doesn’t stop with robots

Basically, DARPA is the Defense Department’s “mad science” branch, and there’s some pretty crazy tech coming out of not only this competition, but from the program in general. But first, let’s talk robots: Some of those competing at the finals will be four-legged, some humanoid, some tracked. Atlas Unplugged, the robot created by Google-owned Boston Dynamics, runs on battery and stays upright, has a wide range of motion, and reportedly, seven of the teams that will compete in the finals are using a version of it.

NASA and DARPA just announced that they will collaborate to analyze spacecraft data on the Deep Web. DARPA also developed a web portal called Verigames in 2013 that crowdsources hackers and coders to work on things like cybersecurity and “cosmic puzzles,” and it’s expanding the gaming system this year. In other words, if you feel up to it, you can contribute to the rise of the machines.

Also see: