fbpx

DARPA’s Next Grand Challenge – Humanoid Robots

Asimo – the most famous humanoid robot – will not attend the Challange
The U.S. Defense Advanced Research Projects Agency (DARPA) is about to announce a new and extremely challenging robotic competition where – for the first time humanoid robots will compete to perform diverse human tasks (semi) autonomously. DARPA (which has yet to announce the competition officially) will be hoping to give the humanoid robot research the push it gave the driving robotic research with its challenges in the past decade.
According to the Hizook robotics website, DARPA will soon announce details of its 4’th robotic "Grand Challenge" this time involving Humanoid robots. Although details are sketchy at this time (at this point no official word was given by DARPA), some info was apparently given by Dr. Gill Pratt – a former professor at MIT who is currently employed by DARPA in the field of robotics. Pratt gave a talk during a recent Defense Threat Reduction Agency"Industry Day", saying that the next Grand Challenge will involve humanoid robot that will have to handle rough terrain.
Pratt described the mission along the following lines: the humanoid robot will have to maneuver into and drive some sort of open-frame vehicle such as a tractor to a given target. After reaching the location he will get down from the vehicle and open a locked door using a key, walk about 330 feet along a corridor full of rubble, climb a ladder, locate a pipe and close it using a nearby valve. All this will need to be accomplished more or less autonomously (what this mean is still not 100% clear).
If this will indeed be the general outline of the upcoming DARPA challenge, analysts are predicting that it will be extremely difficult (maybe even impossible) to achieve given the state of current robotic and Artificial intelligence technology. However the first DARPA challenge launched in 2004 looked almost as hard back than – and indeed out of the 150 mile course, the "winner" – Carnegie Mellon University’s Red Team and car Sandstorm was only able to complete a small fraction – 7.3 miles. However in the next 2 DARPA challenges in 2005 and 2007 many robots were able to finish the course (although both  challenges where far shorter and had other degrees of difficulty built into them). It is possible that DARPA is looking to create an unrealistic first humanoid challenge in order to push the robotic industry forward just like it did a decade ago.
According to Hizook there are going to be six hardware teams which will develop new robots, and twelve software teams using a common platform – the way hardware and software teams will work together is not clear at this point as is the exact announcement date of the upcoming challenge, but we shall keep our eyes open and report when official info will be released.
During 2007 TFOT reported extensively on DARPA’s Urban Challenge including an article covering the full history of DARPA’s challenge as well as some background on the development of driving robots. We have also covered (in a separate article) the results of the 2007 challenge where – Carnegie Mellon University BOSS robot, was the overall winner after completing all three main missions and 19 submissions successfully scooping gold for Carnegie Mellon.

Update: more information regrading the upcoming challange was just pubslihed (much of it is quite similar to the information found on this article).

Related Posts