Lack of autonomy hampering progress of battlefield robots.

They scour caves in Afghanistan, drawing gasps of admiration from military and civilians alike. They scout enemy streets, keeping foot soldiers out of harm's way as much as possible. Years from now, the Pentagon hopes, they could pack a lethal punch equal to that of a combat vehicle.

The technological wonders of ground robotics have been sketched for years on Power Point presentations. But as these vehicles begin to transition from the lab to the battlefield, many questions remain about their future missions and capabilities.

The critical question about robots is how soldiers will fight with them and how they will impact everything from logistics to tactics, according to Maj. Gen. Joseph Yakovak, the Arm/s program executive officer for ground combat vehicles.

Ground robots have for years challenged scientists to turn them from toy-like remotely controlled vehicles to independently thinking killing machines.

Significant advances still need to be made in perception sensors--for the vehicles to be able to function in complex terrain and weather--and autonomous navigation. Many of the current unmanned vehicles are tele-operated, which means decisions are made by a human controller, rather than the robot.

Semi-autonomy has been tested. That allows the vehicle to respond to a command to go from point A to point B. A semi-autonomous system would have sensors for collision or obstacle avoidance, even when the link to its operator breaks down. Autonomy would require robots to execute human-like behaviors.

The fascination with this technology is reflected in the Army's requirement to include droves of unmanned ground vehicles in its Future Combat System, a network of manned and unmanned platforms that will replace existing tanks and fighting vehicles.

Out of a five-year budget (2004 to 2009) of about $13.7 billion, Yakovac said that $500 million will go to unmanned ground vehicles. "It's a pretty hefty sum of money," he said. "Acquisition is very aggressive."

The FCS includes three unmanned ground vehicle concepts: the small UGV--a soldier-portable reconnaissance and surveillance robot; the Mule UGV--a vehicle weighing up to 2.5 tons, suitable for reconnaissance or transport/supply missions--and the armed reconnaissance vehicle (ARV), a 6-ton robot equipped with missiles and guns.

While the requests for proposals (RFPs) came out about two months ago, the Army has not yet explained how robots will be employed. "I do nor think we have thought through the impact of UGVs, such as logistical support, training and man-machine interface as they go into a mission," said Yakovac.

Contractors will be responsible for providing solutions for the man-machine interface, logistics and maintenance, among other issues. Contract awards are anticipated for this month, or June at the latest, said Yakovac. Each of the three systems will go through a 24-month development phase, followed by a second 12-month phase. The Block I UGVs could be completed by 2007, although Yakovac said that is an ambitious schedule.

"It's a stretch, but that is how we have planned it," he said.

Small Unmanned Ground Vehicle 

Fielding the small UGVs, said Yakovac, "to me, this is the easy one." The SUGV consists of a mobility platform, which has to weigh less than 30 pounds; an operator control interface--which has to be lightweight and is being developed in collaboration with the Land Warrior program--and the modular payloads.

One of the main goals is to have a payload for the detection of chemical and biological agents, Yakovac noted.

He admitted that the program has yet to decide how the soldiers can carry the SUGV, where can the interface be provided for them and where the sensor package can be stored. "If [the interface] is on the vehicle, we better get to the vehicle guys, and talk about dividing space and room to recharge," he said. "If we are given a multiple sensor package, where do we put [the sensors]? Do we have a sensor van?"

There will be up to 1,800 SUGVs in the FCS. Each vehicle, without any sensor packages, will cost about $30,000, explained Yakovac.

At the Maneuver Support Center at Fort Leonard Wood, Mo., Dave Knichel wrote an SUGV operational requirements document, but he cautioned that this ORD is not yet part of the FCS and is being treated simply as suggestions that may or may not materialize.

Current SUGVs have an extremely short battery life, said Knichel at a conference organized by the Association for Unmanned Vehicle Systems International. When soldiers from the 10th Mountain Division worked with the robots, "they had to change the batteries every hour and a half, compared to the nine hours that would have been necessary," Knichel said.

The small robots also had a high failure rate, approximately 80 percent of the 96 hours required. They failed at less than 20 hours of use. They also have limited capability in cold weather. Humidity and rain cause problems. Any kind of vibration stops most of the system's components, while replacement parts are not readily available, said Knichel.

Commercial technologies are not always suitable, he said. "The military environment is harsher than the desktop environment, he noted." Knichel's recommendation is to adopt "automotive industry standards" for the UGVs. "My car starts in the cold, and it works in dust and snow," he said. That same reliability is needed in a UGV.

During the testing of small robots, it became apparent that the UGVs interfered with the soldier's primary missions, said Knichel. "We are going to put operator control gear into the system," he said. "Soldiers will nor drop their weapon to pick up an operator control unit." Also, the control units should be based on a common standard.

One standard the Army is evaluating at Fort Leonard Wood already has been accepted by NATO. The technology was developed by CDL Systems. "It not only operates small robots and UAVs, but it also does payload interfaces," he said.

Knichel said the goal is to have "plug-and-play" payloads. With current technology, sensors cannot simply be plugged into the UGVs, he said. Every payload connected differently and had different power needs. In recent tests, none of the payloads interfaced with the UGV control unit, Knichel noted.

A "plug-and-play?" architecture would be critical for the FCS systems, which are being designed with interchangeable payloads in mind, such as mine or NBC detection. The Maneuver Support Center partnered with the Navy robotics lab to develop a prototype standard for the small UGVs and will evaluate it in late 2003.

Autonomous Behavior

The Army and the Defense Advanced Research Projects Agency plan to demonstrate this month, at Fort Leonard Wood, the use of robots and Organic Air Vehicles (OAV) in counter-mine operations.

A reusable engine will allow the ground vehicle to detect, verify, report and mark the mines. The engine also allows the UGV to communicate with the OAV flying above a 10,000-foot-long airfield and spotting what it thinks maybe mines.

Knichel said that simulations are used to determine whether the engine can make decisions independently. "It's just like teaching a machine how to rake notes: 'when I find a mine I stop. Then, I locate the mine, verify it, and once I verify, I report, and after I report, I mark, and after I mark, I either bypass or neutralize.'"

DARPA'S I-Robot UGVs will be used in the demonstration, while the OAV was developed by DARPA and Honeywell, said Knichel. The vehicle's camera will don special lenses that will filter the light to make the mines easier to see. When it detects a mine it will send a report back and the GPS coordinates within 30 meters of the potential mine. This information is sent to the robots in waiting at the end of the runway.

Knichel believes that problems encountered in UGVs, such as the lack of common plug-and-play standards, are self-inflicted.

Small UGVs had poor or no communications in the buildings, while they had excellent communications in tunnels, city sewers and caves, Knichel said. Larger UGVS proved to have poor communications in urban operations, but had good, open, flat-terrain communications.

He said UGVs should have "switchable" higher and lower frequencies to be able to operate in various situations. Higher frequency (4400-5850 MHz, or 14250-15350 Mhz) would be needed for open terrain, tunnels, caves and sewers, while lower frequencies (100-400 Mhz) would support urban combat and operations in building interiors.

Frequencies were nor specified in the ORD, however. "We need to give the community the flexibility to solve the problem," he said.

The Mule

For the ECS, the Army is expected to procure up to 1,200 Mules, Yakovac said. It is going to rake three years to complete the development and testing.

The Mule operator interface is being developed in collaboration with the Land Warrior program. "Land Warrior has too many gizmos that they need to carry, so we have to think about an interface," Yakovac told the AUVSI conference. "We are thinking about how to maintain, sustain it and train it into tactics, techniques and procedures."

The mission equipment packages of the Mule include transport/communications relay, air assault and countermine.

Another component of the Mule is the Autonomous Navigation System, which Yakovac said will be a separate procurement. The ANS is going to integrate perception and navigation sensors. Technically, the ANS would enable the UGV to navigate in different kinds of terrain.

Yakovak expressed doubts about seeing the ANS fielded any time soon. "Most of the people I spoke to said that autonomous may be a little bit too much to handle," he said. "For a near-term solution...I am skeptical."

Armed Robotic Vehicle

The armed robotic vehicle is shaping up to become a very expensive program, with a price tag of about $5 million, nor including sensors, said Yakovak. "This is high-end stuff, and it costs a lot of money." The Army plans to buy up to 700 of these systems.

Skepticism about the ARV is running high. "I think that conceptually...and also technically this is a hard one," said Yakovac. "This is the one

the scares me the most."

The Army has developed separate ORDs for the reconnaissance and the attack versions of the ARV. Yakovak said that the two variants would need to have, if possible, one single configuration.

Even though the ARV has the same timeline of 36 months of development, like the other two UGVs, it is unlikely that it would be ready for Block 1 in fiscal 2007, Yakovak said.

With armed robots, the technology is not as complicated as the "tactics, techniques and procedures," he said, because the vehicle "has the ability to shoot and kill people." The Army is looking to build into the ARV a 30-mm gun, which has to be common with the guns on the manned platforms. Soldiers would decide when and where to shoot.

Yakovac said many questions remain unanswered. "How do you allocate in your planning fires to an unmanned vehicle? How do you ensure that, based on the rules of engagement, you have control over that vehicle? How reliable does it have to be? For example, if you give it a mission and the gun jams, what is your procedure? ... What happens if the vehicle in the mission breaks down, how do you go and get it?"

UGV developers have been enamoured with the technology, and have failed to focus on how these systems are going to be employed in combat, how they are going to be supported and sustained, and ultimately, how they are going to be integrated in the tactical maneuvers, said Yakovak.

DARPA UGCV 

Two years before the FCS ORD was completed, DARPA started developing the unmanned ground combat vehicle, or UGCV.

Contractors are developing two payload classes of vehicles. Team Retarius (Lockheed Martin Missiles and Fire Control, Sandia National Laboratories, BAE Systems and MSE) in January rolled out its 1,500-pound platform, designed for a payload of 150 kg. A month later, team Spinner presented its 7-ton version, with a payload of 2,000 kg. Carnegie Mellon University, Boeing Co., Timoney Technology and PEL Electronics team make up this team.

According to a study published last year by the National Research Council, the UGCV program is unrestrained by conventional design parameters associated with accommodating an onboard crew, such as shock and vibration, survivability and risk factors.

The Retarius UGCV is close to the size the FCS is considering for its Mule-type vehicle. The Spinner UGCV is closer in size to the ECS armed reconnaissance vehicle, according to a DARPA press release.

Endurance is a big factor in the UGCV program. The UGCV is expected to execute long re-supply missions, such as fuel drops. The objective is an endurance of 14-days and a range of at least 450 kin, according to Scott Fish, the UGCV program manager.

The UGCV will operate with imperfect knowledge of the environment and it may occasionally crash and roll over. The program has been looking at designs that can recover from crashes. The prototypes can climb 1-meter obstacles at slow speeds and .25-meter obstacles at moderate speeds, said Fish. A smaller robot is in testing. A larger one still needs a lot of work, said Fish. The smaller vehicle has all six wheels mounted on arms "that can rotate 360 degrees so you have power capability off the shoulder joints and in the wheels," he said. Both prototypes have hybrid-electric drives, for silent operations.

"We want the vehicle to be able to go into airplanes very easily, stay small, airdrop and move quickly," Fish said. Four vehicles of the 2,000 kg version can fit into a C-130.

Fish said that FCS designers may not like the way "these vehicles will come our." However, he said that valuable technology has been developed that would be useful for the FCS. Examples are the hybrid-electric drive and advanced lightweight composite materials.

DARPA developed a surrogate UGCV, called the XRSV, which will be used for field experimentation. "Before we go out and tell them that 'this is the greatest technology that will blow you away,' we would like to talk to the soldiers first-hand and see what they really like and what they don't," said Fish.

The surrogate program has two prototypes, based on the Omitech XRV model.

Fish said that early tests showed that the UGV allowed soldiers to carry less equipment. During a firefight, the vehicle can be used as a bullet shield. The vehicle also helped with the movement of ammunition and weapons forward into the field, while it was able to move casualties our of danger zones. After the fight, the XRSV was able to recover rucksacks.

FCS Schedule

The initial fielding of FCS, scheduled for 2010, is slated to use unmanned systems with limited autonomy. The unmanned vehicles would follow manned leaders. For the long term, the goal is to boost the onboard intelligence of unmanned systems.

A task force sponsored by the assistant secretary of the Army for acquisition, logistics and technology concluded a year ago that a UGV system in the semi-autonomous leader/follower class should be possible as a block upgrade to FCS by 2009. An autonomous capability class would only be possible by 2025, the panel said.

According to the NRC study, "clearly the highest priority for the Army should continue to be the development of perception technologies for autonomous mobility." Off-road mobility is limited and has not been evaluated in unknown terrain, at night, in bad weather or in the presence of obscurants, said the NRC. "The current level of perception capability cannot support an autonomous cross-country traverse of tactical significance, at tactical speeds under combat conditions."

Perception technologies need to be perfected to avoid turning the UGVs into liabilities on the battlefield.

A main reason for incorporating UGVs in the FCS is that they can save soldiers' lives by taking on life-threatening missions. However, the study found no "compelling arguments that UGVs will reduce force structure requirements."

Without advances in areas such as perception and tactical skills, four or more soldiers would be needed to operate each robot. Only beyond 2025 are UGVs likely to operate in a 1:1 ratio.

Each UGV class should be "specified and designed to do what robots can do better (or at lower risk) than humans, rather than trying to imitate what humans already do very well," said the study.

Skepticism persists among war fighters, said the report, because the benefits of the UGVs have been stated in terms of replacing soldier rather than aiding soldiers.