Hardly a day seems to pass without a new report of a soldier or civilian being killed or maimed by an improvised explosive device (IED) in Afghanistan or Iraq. Just such a weapon killed two coalition members on Monday in Iraq's volatile southern region, according to NATO. Meanwhile, data published October 22 by Wikileaks indicates that IEDs are the biggest killers of British and US troops in Afghanistan, accounting for more than half of all fatalities.

To protect people from these weapons, the U.S. military increasingly relies on robotic bomb detection and disposal units (in addition to the scores of unmanned aerial vehicles (UAVs) overhead with reconnaissance and strike capabilities), creating life-and-death combat relationships between man and machine that will only deepen and proliferate over time.

U.S. and coalition military operating in Afghanistan have experienced about 10,500 roadside bomb incidents so far this year, up from 8,994 in 2009 and 2,677 in 2007, Army Lt. Gen. Michael Oates, director of the Joint Improvised Explosive Device Defeat Organization (JIEDDO), reported October 21. A significant portion of these are IED-related incidents.

The success of such attacks ensures that they will continue for at least the short-term, which means the military is searching for new ways of addressing the problem. In Afghanistan U.S. soldiers conduct month-long Explosive Hazard Reduction courses to train Afghan soldiers about how IEDs work, the threats they pose and techniques for finding them. Coalition forces have spent about $2 billion since 2006 to train soldiers stationed worldwide about IEDs, according to Oates.

A lot of money is also being spent on anti-IED robots. Earlier this month the U.S. Army TACOM Contracting Center in Warren, Mich., ordered $14 million worth of robot intelligence software and spare parts for its PackBot tactical mobile robots, made by iRobot in Bedford, Mass. This was TACOM's 20th such order for iRobot technology and part of the unit's larger $286-million "xBot" contract for purchasing bomb-detection robots.

There are more than 3,500 iRobot devices serving in the U.S. military as well as the armed forces of another 20 allied countries, says Joseph Dyer, iRobot's chief operating officer and a retired U.S. Navy vice admiral. Scientific American spoke with Dyer about the role that robots are playing in modern warfare and where the technology is headed.

[An edited transcript of the interview follows.]

The military's reluctance to rely on robots in combat situations has changed dramatically in recent years. Why is that?
Interest in robotic devices really started with unmanned aerial vehicles in the early 1980s and the RQ-2 Pioneer UAV [developed by AAI, which is now part of Textron Systems Corp., and Israel Aerospace Industries]. The change necessary to adopt, much less embrace, unmanned aircraft first encountered resistance from the white-scarf system of fighter pilots who believed that an unmanned system couldn't do the things a human pilot could do. While there's still some resistance to unmanned vehicles, the realities of war are changing that. Whereas there was a standoffish attitude with the white scarves regarding robotic vehicles in the air, with the Army the soldiers realized robots could do the things the soldiers themselves really didn't want to do. The Army's mission is up close and personal, dangerous and dirty. Consider this: it took 20 years for the unmanned aircraft market to become a half-billion-dollar industry, but it took the newer market for unmanned ground systems only half as long to reach a half-billion dollars.

What is it about the combat in Iraq and Afghanistan that has created a demand for land-based robotic devices?
In urban and desert terrains robots contribute anywhere you need distance between your body and bad things happening. Urban warfare, even for an Army and Marine Corps as accomplished as ours, is scary stuff and fraught with issues of collateral damage and risk to noncombatants.

How are robots typically used in combat situations today?
Of the 3,500 iRobot devices serving the military, the majority of them are helping with explosive ordinance disposal (EOD). Normally, troops will be told about an explosive device or actually find an IED. These soldiers will then call in the threat to specialists who investigate and possibly engage the IED with the aid of a robot, which can be used to send video of the device to an EOD team standing out of the device's range and/or to dispose of the device. In both Iraq and Afghanistan, because of the proliferation of IEDs and the fact that you don't have enough EOD troops to be everywhere they're needed, there's also a strong desire for non-EOD teams to use robots for what's known as route clearance, where the robots will help identify and possibly clear roadside IEDs. During the next phase of use, robots will not just be focused on IEDs or road clearance but rather on becoming a real player in the infantry.

A player in what sense? How is the military's use of robots changing?
Let me tell you a story. In a battlefield exercise about three years ago at Fort Benning, Georgia, one where many new technologies including robots and new UAVs and wearable computers and all of those things were being tested in a war game, an Army captain leading one of the teams was asked which of these technologies he would want to take with him to combat today. He said he would take iRobot's small unmanned ground vehicle (SUGV) [used for surveillance, reconnaissance and/or bomb disposal] and AeroVironment's Raven UAV. The captain explained, "I want situational awareness on the battlefield, and I want it from God's view, and I want it from up close and personal. If I have foreknowledge of the battlefield, it ensures the survivability of my troops and the success of my mission."

The next phase of this, which is just beginning, is robot autonomy. This is probably less exciting than most people would believe. Whereas today you have a soldier using a joystick to manipulate a robot much like a video game, what's next are a few pieces of autonomy that enable the robot to take over some of its own navigation, like cruise control. I don't have to drive the robot every step of the way, I can tell it to maintain a certain vector while traveling. Another area of autonomy is to program robots to, when they lose their communication signal with the troops, automatically navigate back to the last position where it could send and receive a signal. Today, if a robot loses communication, you have to go out and get it, which is not a popular thing. If you wanted to be out there in the field, you wouldn't have sent a robot in the first place. Little by little, robots will be able to autonomously complete more and more complex assignments to the point where you can program them with actual missions.

What types of robotic missions are we talking about?
Today, the ratio of robots to operators is one to one. You're going to start to have one operator with multiple robots that can be operated in a swarm, carrying out coordinated tasks. For an example of how this might work go back to the 9/11 time frame when we had anthrax in the Capitol Building offices. We couldn't even get in there for days. Now you have robots that could go in there and do chemical and biological testing and even create maps that would give you an understanding of the dangers and where they are located within the building.

How are robots themselves changing to meet the demands of ground combat?
If you go back and look at the early days of airplanes and even hot-air balloons, they began as tactical reconnaissance technologies. Then, once you got frustrated at being able to see the action but not do anything about it, the technology was given new features such as the ability to fire weapons. Robots likewise started out this way. We're not there in terms of giving robots strike capabilities yet, but in time we will see armed robots.

Would armed robots pose a danger to both sides in a combat situation, especially if there is a malfunction?
In terms of lethality it's our position that you're always going to have a man in the loop, even if a lot of the press and all of Hollywood don't see it that way. Will we have autonomous lethal systems? I don't think so. I don't think it's ethically right, and I don't foresee a world that hands over life-and-death decisions to a machine, even if it's a very capable system.