US Army pleads for help making robots safe
Tons of gear – few answers
RoboBiz The US military plans to have a vast robot army in the near future but has yet to come up with a concrete method for ensuring that mindless devices can operate safely alongside brained troops. So, the military has been forced to beg for help.
General Charles Cartwright urged attendees here at the Robo Business conference to pursue research projects that address "how you put ground troops and robots together at the same time." The need for technology that can guarantee the smooth behavior of robotic systems proves dramatic given that the Army plans to start deploying a new fleet of robotic gear in Iraq by 2008. Shortly thereafter, the Army intends to push out myriad robotic devices, stretching from automated weapons systems to autonomous tanks.
"By the end of September, we'll have over 10,000 different ground robots in Iraq and Afghanistan," Cartwright said, championing how much progress the Army has already made with its robotic agenda.
At the moment, the majority of the robotic systems are fairly small. One of the most often cited military devices is the PackBot made by iRobot. The Army can change the payload that sits on the PackBot Scout base to handle different tasks such as disarming bombs with an arm mechanism or finding snipers via a sophisticated sound sensor.
A much more complex set of robots will arrive courtesy of the military's Future Combat System (FCS) program. At the heart of FCS will be a brand new network designed to connect air and land vehicles, soldiers, robotic systems and myriad sensors.
The FCS agenda will require the Army to create a new operating system dubbed SOSCOE - System-of-Systems Common Operating Environment. (Cartwright described this software as "Windows on steroids," which didn't seem terribly reassuring.) The OS will link together the myriad hardware components and attempt to transfer information in real-time between all the different devices. A plane, for example, could fly over a site and instantly send information about what it sees to approaching ground vehicles and troops. In addition, robotic convoys could use the information for improved navigation.
Under FCS, the Army will build 18 new vehicles with rather fantastic capabilities. The military will have robotic scouts that can crawl through sewers and tunnels, medical treatment vehicles, units that can carry soldiers' equipment for them, maintenance vehicles and, of course, a wide array of weapons systems.
Cartwright noted that in almost all cases the Army will concentrate on building and deploying unmanned FCS systems before it gets around to churning out manned versions of the vehicles and weapons systems.
All of this work comes as the military tries to meet a federal mandate to have one-third of ground vehicles operated unmanned by 2015.
Such an objective seems unrealistic given the current state of robotics. On the consumer front, the most impressive practical device to date is a vacuum cleaner. Meanwhile, the PackBots have proved their mettle on the battle field, but sophisticated multi-ton ground vehicles that can do more good than harm seem just a dream.
Beyond the practical issues of having a robotic ground vehicle work well, Cartwright raised a number of more perplexing problems. How can an autonomous weapons system distinguish between friend and foe in a combat situation? Even more importantly, can the military guarantee that it and not an enemy still maintains control of the robots?
"How do you know if the robot is still friendly when you start getting on the network," Cartwright asked.
For the first time, the Army has deployed a unit in Texas that will be experimenting with the military's robotic wares in the hopes of figuring out just how far the technology has come.
All of this military interest in robotics technology is, of course, just what the robotics industry needs. The US military machine has spent more than 50 years boosting American technology programs with its billions. Given the scope and ambitions of the FCS program, it seems clear that robotics companies will get more than their fair share of federal pork in the next decade.
For the US to come close to achieving its goals, however, the robotics field will have to demonstrate of level of imagination and competency that it has failed to show to this point.
With this context in mind, it's rather disconcerting to see a General plead for help to figure out how to make this multi-billion dollar robot army safe. Hopefully, someone or some group will step forward with a way to have troops, civilians and robots work together in a decent way before all the pork is dished out.®
Sponsored: Are DLP and DTP still an issue?