Mild-Mannered Canadian Fury

Doug Stephen is Politely Peeved

DARPA Robotics Challenge Roundup


Thu, 25 Oct 2012 Ā«permalinkĀ»

Now that IROS is over and DARPA has officially kicked off the Robotics Challenge to the public and lifted the press embargo, there are some pretty cool articles floating around the internet containing some good information about competition.

The short version is that DARPA wants help in developing robots that can be utilized in disaster scenarios; the project was heavily influenced by the recent Fukushima disaster. The thinking is pretty simple; many modern disaster response robots aren’t actually very good at disaster response. First of all, none of them are humanoids, and in places like Fukushima the environment is built around the idea of humans being the entities that have to navigate the environment (in a non-disaster scenario). Even in a “degraded” environment, humanoids are the most suited to handle dangerous situations, but we can’t actually send humans in due to the danger and risk. Not little treaded tanks. In addition, most modern tools aren’t designed to be used by non-humans. They’re developed with human hands and ergonomics in mind. So instead of developing specialized tools for non-human robots, it would be nice if we could develop robots to utilize human tools. Lastly, many modern robots are hard to operate. Their UI’s are unintuitive, and they require a lot of training. The experts with domain knowledge in disaster management often don’t have the domain knowledge required to operate a robot. If we can simplify the operator UI, then the training barrier drops and people with expertise in disaster response can begin exercising their skills more readily.

There are four “tracks” in the DRC:

  • Track A participants are building their own robots.
  • Track B participants are only developing software at first; near the end of the first phase of the competition there will be a qualification round to see whose control algorithms perform the best in a simulated, virtual disaster. The top performing teams will then be awarded a robot known as the “GFE Platform” (Government Funded Equipment). The GFE is based on the Boston Dynamics PETMAN/ATLAS [1] [2] robot. The Track B contestants that make it this far will continue through the competition using the GFE as their robot for the rest of the competition. I’m part of a Track B team.
  • Track C participants will also be competing in the software stage, but will not receive any funding from DARPA up front. Track C teams will compete at the same virtual challenge as Track B teams in an effort to supplant one of them and receive a GFE robot and funding going forward.
  • Track D participants will supply their own hardware and software, but will not receiving any funding from DARPA at any point in time.

The team that performs the best with real hardware at the end of the whole challenge will receive a $2,000,000 USD prize.

For more information, here are some good articles:

And just a reminder, we’re looking to fill some job openings at the IHMC related to the DRC: