Putting Your Best Bot Forward

By Carl Weiss

You may have missed this item in the news recently, but on December 20 sixteen teams duked it out at the DARPA Robotics Challenge for a chance to win $2 million.  Sporting monikers like RoboSimian, Valkerie and Atlas, these aren’t your daddy’s droids.  These robots walked like a human and were required to perform a number of tasks such as driving a vehicle, climbing a ladder, open doors and turning valves.  While not able to replace their human counterparts as of yet, the competition was clearly designed to raise the bar in order to eventually create a rescue robot that could be sent into harm’s way in situations where humans fear to tread.  (Imagine how useful these bots could have been in Fukushima during the tsunami and reactor disaster.)
Not that you’d mistake these bots for your typical rescue worker.  Looking more like a Tinkertoy than the Terminator, Schaft the winner of the competition (recently acquired by Google), resembles a boom box sprouting arms and legs.  Standing 4 feet 8 inches tall and weighing in at 209 pounds, it is unlikely to win any beauty contests.  However, what it lacks in looks it more than makes up for in utility, if its performance at the DARPA competition is any indication. 

To begin with, all the robotic competitors were required to be able to drive a rescue vehicle and navigate their way around a series of objects.  Just as with most emergency situations that could require the services of a rescue robot, you can’t always assume that the location will be pristine.  Whether induced by tornado, hurricane, tsunami, earth tremor or act of war, being able to negotiate a rubble-filled roadway is job number one for any would be rescuer.   When disaster strikes what is required is a robot that can think on its feet.

While many people are familiar with bomb disposal bots being used in Iran and Afghanistan, most of these are operated remotely, necessitating that a human be at the controls.  While this kind of operational environment works in ordinance disposal, in the kinds of environments that rescue robots would be sent to work, the need for autonomous operation is imperative. 

In the first place, ordinance disposal droids are typically tracked and are usually limited to one articulated limb.  Rescue robots on the other hand are fully articulated and are required to perform tasks that would be difficult to accomplish in a telepresence-only mode.  Secondly, in high radiation areas, relying on remote controlled robots could prove difficult, since radio interference would be almost a certainty.

To make the competition more like a real disaster area, DARPA broke the playing field into 8 Tasks: Driving a Vehicle, Handling Rough Terrain, Climbing a Ladder, Clearing Debris, Opening Doors, Cutting Through a Wall, Opening a Valve & Using a Hose.  I know several humans that would have had a tough time accomplishing all of these tasks.  For semi-autonomous robots, the task was daunting, which is what separated the bots from the boys in the recent competition in Miami. 

Of the sixteen teams entered into December’s competition, only seven scored in the double digits. Several competitors such as NASA’s Valkerie failed to score even a single point, which shows how difficult the tasks were to accomplish.  Kind of like an Olympics for robots, the winners will advance onto the finals which are going to be held in the next 12-18 months.  Even those who did not make the cut will be able to learn from their mistakes as well as from more successful competitors. 
  • Team SCHAFT (SCHAFT Inc.): 27 points
  • Team IHMC Robotics (Florida Institute for Human & Machine Cognition): 20 points
  • Team Tartan Rescue (Carnegie Mellon University and National Robotics Engineering Center): 18 points
  • Team MIT (MIT): 16 points
  • Team RoboSimian (NASA Jet Propulsion Laboratory): 14 points
  • Team TRACLabs (TRACLabs, Inc.): 11 points
  • Team WRECS (Worcester Polytechnic Institute): 11 points
For those who couldn’t make the trip to Miami to see the competition live, fear not, because you can find the replay on YouTube. While the video shows the robots going through their paces, don’t expect to see the kind of performance that we have all come to know and love in Star Wars.  While some of the robots were able to score points by performing one or more of their assigned tasks, they did them with all the rapidity of a Galapagos tortoise.  That doesn’t mean that rescue robots are something only likely to save someone on the big screen.  What it means is that like aerial drones and autonomous vehicles, while these automatons are not yet ready for prime time, within several years these robots could be coming to a disaster area near you.  Until then it’s up to the ingenuity and persistence of robotics engineers to keep putting their best bot forward. 

Carl Weiss is co-host of the weekly radio show Working the Web to Win which is broadcast every Tuesday at 4 pm Eastern.  You can also see the show on YouTube.
Enhanced by Zemanta

1 comment: