For military service members in high-risk occupations, such as
explosives detection and disposal, robots can be welcomed relief.
With planned investments of at least $200 million on robotics-related
research during the next five years, the Defense Department has
more than just financial motivation to push robotics technology
into mainstream programs. Robots can help put fewer human lives
at risk and, in some cases, take over mundane jobs that troops find
boring and unfulfilling.
But even though the technology is advancing in the laboratories,
and systems are being used successfully in fields such as mine-detection,
there are non-technical issues that the Defense Department should
address before robots are taken to combat, said Army Lt. Col. John
G. Blitch, program manager for tactical mobile robots.
The TMR program, which started in 1998, is sponsored by the Defense
Advanced Research Projects Agency (DARPA). The goal is to field
small, agile robots that can assist human forces on the battlefield,
Blitch explained during a briefing to the Association of Unmanned
Vehicle Systems, in Arlington, Va.
The Army plans to spend $120 million on ground robotic vehicles
during the next five years. About $20 million will come from DARPA.
One significant challenge for the Army, said Blitch, is how to
teach soldiers to treat robots as weapons, not as toys. That is
an important consideration, he said, “if we require the soldiers
to maintain proficiency in their field and get additional robotics
skills.” Research indicates that, once a soldier starts to
learn how to operate a robot, “it’s likely that the
primary skill will go downhill,” Blitch said.
Many people wonder “what the heck is so tough” about
operating robots, Blitch noted. The Air Force, for example, has
been doing it for many years. But ground platforms are entirely
different, he explained, because they require skills at “obstacle
negotiation.”
“Flying a UAV [unmanned air vehicle] around is a relative
easy task. It’s a benign environment [with] no obstacles,
except for the occasional ski cable car,” said Blitch. “You
don’t have people trying to pick you up and throw you away.
Controlling a ground platform is infinitely more difficult.”
The ground robots currently in development still require high levels
of human intervention—they are not autonomous. The upshot
is that the operators must be well trained and knowledgeable, said
Blitch. “Right now, we have pretty dumb robots. So we have
to compensate for robot stupidity with skilled operators.”
Twenty years from now, “we may have more sophisticated vehicles
that will not need as much operator skill.”
The robots in the TMR project are not “entertainment-type
systems,” said Blitch. “Our platforms have to deal with
challenging tactical situations.”
Nonetheless, Blitch said, the last thing the Army should do is
“degrade combat units” when robots are introduced to
the force. “We should not be messing around with our front-line
combat units,” he asserted. For that reason, robot operations
should be assigned to National Guard units. “The Guard has
more freedom to fail,” he said. In active-duty units, conversely,
if the technology fails, a commander’s career is over and
the unit gets “blacklisted.”
The TMR program is not designed to develop robots that can replace
people, but rather machines that can do what humans cannot, Blitch
said. The current focus of the program is to teach the robots how
to deal with obstacles, how to figure out their location and how
to right themselves after tumbling or flipping over.
In the TMR program, he said, robots must not only avoid obstacles,
but also have to perform better than manned vehicles in rugged terrain.
“If the robot can’t handle slopes steeper than what
the Humvee can climb, [it is of] no use to me.”
The next phase of the TMR testing will focus on urban combat and
trying to get the robot to penetrate denied areas, such as city
sewers. These robots could be used, for example, to search through
rubble and find victims of earthquakes or bombings.
Most robots in development today are not intended for combat in
the trenches. The TMR, however, “will be designed to be shot
at,” Blitch said. It also may be able to fire small arms to
distract the enemies.
In the long term, he said, robots will acquire some level of on-board
intelligence. “We want the robots to alert us when they want
a human to step in and control. ... It has to know whether it has
completed the task well enough.”
One idea that the TMR office is evaluating is the so-called “marsupial”
concept of operations, in which a larger combat vehicle carries
the robot to battle, and the robot, in turn, can serve as a refueling
platform.
Asked whether today’s young soldiers, many of whom grew up
playing Nintendo games, are more apt as robot operators, Blitch
said the answer is “surprising.” During early experiments
with TMR, the “Nintendo factor” came into play, “but
nowhere near what we thought.
“I figured that kids would be walking straight into the robot
control,” Blitch said. But operating robots is much more complex
than playing video games, because it requires an understanding of
“sensor fusion.” Every video game comes with a lot of
noise and visual action. But the data provided by robots come from
multiple sensors, so interpreting that information is much more
difficult than just hearing and seeing.
Even though Blitch is optimistic about the technological achievements
in the TMR program, he said it would be unrealistic to assume that
these systems will work until more testing is done with large numbers
of robots.
“I would recommend buying 300,000 of these immediately,”
he said. “That is the best thing you could do for robotics.
Buy a large quantity [to] flush out the human-interface problem.
... We need to be able to play around with tens or hundreds of these
things.”
Robots are performing quite successfully in the field of explosive
ordnance disposal (EOD), said Paul Milcetic, a representative from
the Defense Department’s EOD program office, which has 3,300
qualified EOD technicians from all the military services.
Unmanned ground systems also are becoming popular in civil engineering
and force-protection applications, said Capt. David E. Shahady,
chief of robotics research at the Air Force Research Laboratory.
The service so far has fielded about 200 robotic systems.
Among those systems is a 350-pound robot (the Andros Mark V-A1)
that can remove small explosives from aircraft and buses. A 700-pound
vehicle targets medium-size, improvised explosives, Shahady said.
The newest robot is a 10,000-pound all-purpose system, designed
to detect and remove large explosives, such as a truck bomb. At
least 20 have been fielded, Shahady said, and there are plans to
buy 30 more.
EOD units are considering buying a laser weapon for ordnance neutralization.
It would be a 1-kilowatt chemical laser that would burn unexploded
ordnance, said Shahady. He told potential vendors that the price
has to be no higher than $300,000, in order to make it available
to all EOD units. Additionally, the Defense Department wants new
robotics tools to remove explosives from dead bodies, as well as
large robots that can travel fast and rescue victims from an aircraft
crash site, where it would be unsafe for humans to operate.
The Andros robots employed by the Air Force also are used by the
Army in Kuwait and Saudi Arabia, because they can maneuver well
in the sand, said Shawn Farrow, vice president of Remotec, in Oak
Ridge, Tenn. The company is a subsidiary of Northrop Grumman Corp.
The company makes the Andros line of robotic vehicles, which can
maneuver in sand, gravel, snow and mud, Farrow told reporters during
an Army conference in Fort Lauderdale, Fla. The company’s
largest military program is the remote ordnance neutralization system,
which includes 178 robots. The program began 11 years ago. Deliveries
should be complete by 2002, Farrow said.