The DARPA Robotics Challenge (DRC) is a competition established and funded by the Defense Advanced Research Projects Agency (DARPA) to encourage the development of semi-autonomous ground robots that can enter disaster and combat environments too dangerous for first responders and military personnel.
NetApp, an avid supporter of innovators, is sponsoring Virginia Tech’s Robotics team- Team Valor. Recently, the GovDataDownload editorial team had the opportunity to sit down with John “Semi” Seminatore, Graduate Research Assistant at Terrestrial Robotics Engineering and Controls (TREC) at Virginia Tech, to discuss some of the innovation that is coming out of Team Valor’s lab and Virginia Tech’s robotics program.
Continue reading to learn more about what Semi had to say about the DRC, DARPA and NetApp’s dedicated support of innovation in robotics:
GovDataDownload (GDD): Can you tell us about the DARPA Robotics Challenge, Team Valor and ESCHER?
John Seminatore (JS): In 2011, following the Great East Japan Earthquake, a tsunami disabled three reactors at the Fukushima Daiichi nuclear power plant, causing a power failure and loss of containment of the reactors. Because of the radiation levels, rescue teams wanted to send in a robot to contain the situation. One of the first things the robot came across was a flight of stairs covered in rubble. There was no way for the robot to maneuver through there, rendering it useless in that situation.
After the fact, DARPA asked- why can’t robots do this? Why can’t they climb stairs or move rubble? And so the inspiration was to hold a robotics competition in order to push the technology in the area of search, rescue, and recovery, so that we could more effectively send a robot into a disaster or combat zone to preserve human life. It would not have taken much to avoid those three subsequent melt downs after the tsunami if the could have had a hose connected to the proper valve and then pumped cool water into the reactor. Had the robots had the capability to climb stairs and really maneuver the environment, it would have saved a great deal of heartache.
Part of what makes ESCHER -Team Valor’s DRC robot- and the team unique is that we basically built everything from the ground-up. We made and cut the metal, built the circuit boards, programmed the computations and wrote the software ourselves. That is pretty rare amongst robotics labs.
Aesthetically, it’s a beautiful machine. With the equivalent of two desktop computers on board- and all of that compute power running at full tilt- there’s enough compute for the robot to balance itself, plant its feet and then send back images of the world around it to the operator’s station. The artificial intelligence (AI) is still in the beginning stages, yet we are extremely proud of ESCHER and what we have been able to achieve.
ESCHER is an upgrade from one of our previous robots that we had built for Navy’s SAFFiR project, a program established to build firefighting robots. We were already working on that project when the DRC was announced. Our SAFFIR robot was not up to the task. It could not walk untethered; it still needed to be plugged into a wall. But, it was a research platform. ESCHER represents a significant upgrade with far more computational patterns, the ability to run for two hours fully on battery power, much more precise vision systems and stronger limbs that have the ability to climb stairs and lift heavier loads.
The team is incredibly proud of ESCHER. We see DRC as ESCHER’s debut. For some competitors, DRC is a culmination, but for TREC’s Team Valor, we see DRC as merely the beginning. We are excited to see where ESCHER is two to three years from now. Team Valor is just getting started.
GDD: How did NetApp become involved in Team Valor?
JS: It was a bit serendipitous, really. Basically, with ESCHER we generate a great deal of data when we conduct an hour run. Our vision sensor alone will produce one gigabyte of data every two seconds, so you can imagine over a two hour run, which is what the DRC requires, we generate a huge amount of data, and so we were looking for a storage server.
Before then, we’d been using servers one of the guys had lying around. In addition to storage, we were also trying to cut down on the amount of time it took to extract data from the robot. If we did an hour run, it would take nearly three hours to get the data. We researched NetApp and knew the school already had a relationship with them. The school introduced us to our NetApp account rep, Lisa Lugar. and thus the partnership was born.
We ended up installing a NetApp E2712 storage system, and it’s great.
GDD: What kinds of support does a sponsor provide?
JS: We couldn’t do what we do without our sponsors.
NetApp, for example, we don’t just need their help with a system to store onboard data from the robot, but also the data from the footage we film and the tests we run. We use and generate huge amounts of data. Having a reliable solution- separate from the school network- is a necessity to what we are trying to achieve in the lab. NetApp provides us with the infrastructure. But, ESCHER is a precision machine, and so we have had a great deal of support from other sponsors as well. For example, our arms are from HDT Prosthetics. Arms are difficult to build, and we wanted to focus on the legs. With help from HDT, we have really strong, yet simply designed arms for ESCHER.
Again, we could not do what we do without our sponsors. We simply don’t have the manpower and time to handle every little aspect. Just last week, a group of third graders visited the lab and one little boy asked how long it took to build ESCHER. It took our team- with corporate help- 10 months, but it would take a single person decades to build a robot like ESCHER.
GDD: Why is supporting students in these types of competitions so important to the future of IT innovation?
JS: One of the things we really pride ourselves on is that we actually build whole practical systems and force students to get their hands dirty. There’s a huge difference between learning the theory behind something in a classroom, and actually building it and thinking your way through the practical, real-world design.
Competitions like DRC hold only one disadvantage. It can slow graduation for Ph.D. students because they can’t publish papers or research until the machine is built. But at the same time, the hands on experience with build and design will make them far better engineers. When they complete the program, they will understand the interactions of what’s going on. You’re not going to learn in a class how to integrate 4 programming languages like we have running in ESCHER with over 2 million lines of code. But, in competitions like DRC, we are teaching undergraduate students how to integrate this knowledge and these complex actions. Without corporate sponsors, like NetApp, there would be no way we could participate in these types of events and help our students hone their engineering skills before they even graduate.
Want to see ESCHER talk its first steps, check out this video:
To learn more about NetApp’s unwavering support of student and government programs, click here.