Do military robots have a future in land warfare?

Do Military Robots Have a Future in Land Warfare?

Darth Vader Stalks the Battlefield

In essence, the military robots under development today represent the ideal soldier. They never tire as long as their power supply lasts; they are absolutely fearless; their reactions are many times faster than those of a human; they can lift or transport loads weighing many times their own weight; and last, but not least, they will eventually be able to mass-produce other robots in largely automated factories.

Robots are by no means new. The German Army had some 5 000 built for battlefield applications in World War 2, which weighed from 0.4 to 3.7 tonnes. These tracked and wire-or radio-guided miniature tanks were used to destroy fortifications with the explosive charges they carried, to eliminate weapon-protected obstacles, to carry ammunition across terrain under enemy observation and fire, and to breach minefields.

Tomorrow's Force Multipliers

Battlefield robots being designed today have essentially the same tasks as their W W2 predecessors, namely to go into action wherever the combat environment is too dangerous for humans.

The above-mentioned advantages are not however, the primary driving force behind the current efforts to develop battlefield robots in the industrial nations. The ratio of support personnel to combat troops is constantly increasing. It is currently estimated that each fighting soldier needs at least 10 men to support him, and the trend is accelerating.

This is being countered by introducing more capable weaponry and by increasing its automation. The process has come to be known as Force Multiplication. That this approach contains an inherent fallacy is obvious. While such measures indeed enhance the combat capability of the soldier, they also increase the need for qualified personnel to service and repair the advanced equipment. It is by no means always possible or cost-effective for this to done by simply replacing defective modules. Consequently, the manpower ratio is shifting progressively even more to the logistical and support side.

Since the number of men available to serve in the armed forces is finite (and, in some cases, declining) and, above all, the qualified personnel required to maintain today's sophisticated weapons is everywhere in short supply, the idea of using robots as force multipliers was born.

In the United States, the Army has recently drafted an "operational and organizational plan" which lays out a requirement for a robotic vehicle to perform dangerous missions. Besides their ability to minimize the soldier's exposure to hazardous situations, robots also offer the means to offset an enemy's possible numerical superiority in men and vehicles. The draft reveals a new and somewhat unexpected attitude of the Army towards fighting robots as it recommends that these should be able to carry modular payloads and thus, for the first time, opens up the prospect of lethally armed robotized vehicles.

Military Robots Today

The military want to see robots in service as soon as possible. In some fields of application this has been achieved already. Smart air-to-ground missiles and modern Unmanned Air Vehicles (UAVs) possess considerable intelligence. The same can be claimed for sea mines and torpedoes which are equipped to search, identify and attack targets with their own processing systems.

For land warfare, weapon systems of comparable capability are not yet operational. This is because of the enormous difficulties posed by the necessity to move cross-country, in a two-dimensional space which abounds with obstacles.

This has prompted the initiation of studies into the possibility of giving land robot systems a "hovercraft" capability for moving rapidly and safely over impassable terrain. But translating such plans into reality is still in the distant future, since the computer capacity needed cannot yet be sufficiently miniaturized to fit into a viable combat robot.

Two-pronged Approach

This limitation has resulted in a two-pronged approach for the design of tomorrow's combat robots.

The first is a long-term effort. This has the goal of developing the advanced technology necessary to construct highly intelligent robots which will be capable of operating autonomously on the battlefield. Such an advanced robot would have to be able to take the place of a soldier, to move intelligently in difficult terrain, to recognize dangers, distinguish between friend and foe, and above all, fight efficiently with minimum human intervention. This effort cannot be expected to bear fruit before the end of the century.

The second approach aims at the development and early field testing of sophisticated, but remotely operated, machines which are essentially extensions of a human soldier's fighting skills. This extension is limited in distance by the length of the remote-control wire or the range of the data-link channel. A machine thus controlled cannot be described as a "robot", in the true sense of the word, since a human remains the decision-making element in the man/machine loop.

Technically viable concepts for remotely-controlled robots were formulated by military planners in the late 1970s, and industry rose to meet the challenge in most industrialized countries. As in so many other commercial developments with potential military spin-offs, industry and the universities are often reluctant to admit that their robotic research has military applications. Many of them are even more reluctant to acknowledge that their products can become key components in weapons systems. It is, however, no secret that the development of military robotics is impossible without the expertise gained by the manufacturers of industrial robots.

Industry's Role in the US...

In robotic technologies, Japanese firms and/or their overseas subsidiaries are among the most noted developers. In the United States, for example, Seiko Instruments USA Inc. and Hitachi America Ltd. are deeply involved in this field alongside their all-American counterparts. The latter include Cincinnati Milacron's Robot Division, Microbot Inc., Unimation Inc., Odetics Inc., John Deere, Caterpillar and Intelledex. Companies with experience in the armament field like Martin Marietta, Hughes, Gould, Grumman, FMC, General Dynamics, Cadillac Gage, General Motors or Robot Defense Systems Inc. (a newcomer to this established group), are also engaged in robotics.

None of these companies would be able to design robots without the help of the many software and computer hardware producers involved in artificial intelligence development, many of whom remain relatively anonymous. The not-so-shy include Denelcor Inc., the supercomputer producer Cray Research Inc. and IBM.

The US Army and the US Marine Corps are currently attempting to harmonize their requirements for a common vehicle. Martin Marietta and Grumman are both developing systems which are being examined by the Army Missile Command. These optical fibre and radio-controlled all-purpose robots are designed to be carried on an HMMWV and would have a speed of 10 km/h. As envisionned by the Army, the system (which has already been designated Caleb) would include a transporter, an operator control unit and the actual robot.

... and in Europe

In Europe the picture is very similar. In West Germany, the armed forces have reportedly ordered from an unnamed source two experimental battlefield transport vehicles. One of those is wheeled, the other tracked. They are fitted with manipulator arms for moving containerized ammunition pallets. Dornier, together with AEG, ESG, MBB, Krupp-Atlas Elektronik, Siemens and Thyssen-Henschel recently received a sizeable contract from the German Ministry of Defense for research into the options offered by battlefield robots. The tasks envisioned include transport, vehicle recovery, mine-clearing, decoy services, reconnaissance, command and control, and finally combat.

In Italy considerable activity can be noted. Ansaldo, Elsag, Selenia and Vitro Selenia have created the consortium Italrobot in which Fiar, Officine Galileo and Oto-Melara are also participating. The goal is collaboration with other European companies in the Advanced Mobile Robot (AMR) Project.

The AMR will be designed to move at high speeds over all types of terrain, have the capability to evade obstacles, to follow a chosen track and serve as a transport vehicle. Such capabilities will make the AMR an ideal platform for anything from weapons to reconnaissance equipment.

In the UK, a collaborative effort between the government and industry has resulted in the formation of the Mobile Advanced Robotics Defence Initiative (MARDI). On the industrial side, this involves Alvis, British Aerospace, CAP Scientific, GKN Defence, Lucas, Marconi, Royal Ordnance, SD Scicon, STC Defence Systems, and Westland Dynamics. On the government side of MARDI are the Royal Armaments Research and Development Establishment (RARDE) and two universities. The long-range goal is identical to that of Italrobot, but in addition an autonomous robot is being studied. Robotic mine-clearing tanks have already been delivered to meet an urgent requirement on the Falkland islands, where the unmapped mines left after the war prohibit agricultural use of valuable terrain.

Most currently known efforts to develop combat robots center primarily on equipment which is remote-controlled by human operators. Today's robots are equipped with sensors which permit the operator to observe on his command monitor everything the machine's infrared or TV sensors perceive. A built-in sight permits the aiming of the weapons carried on the platform.

This technology was pioneered for paramilitary purposes almost 20 years ago in the UK, with the tracked "Wheelbarrow". This, and similar equipment has been in service for some time with police and security forces to inspect suspicious packages and to destroy or disarm already identified bombs. Such robots can be made highly mobile and some designs can even climb staircases. They can also be armed and may be used to dislodge criminals using shotguns or incapacitating gases.

Identical technology is employed in mobile robots needed to service equipment in the radioactive zones of nuclear powerplants. After the Chernobyl disaster, for example, West Germany supplied a number of remote-controlled, tracked earth-movers to speed up the work in the contaminated area.

Robots for Combat Support Missions

Similar equipment is envisioned for combat engineer missions. * Runway Repair Robot. Filling craters and repairing runways on airbases that have been attacked are becoming increasingly hazardous. Modern airfield attack weapons often include delayed-action anti-personnel and anti-vehicle mines designed to hinder repairs and keep the base out of action for as long as possible. Such jobs can be performed with remotely operated and heavily armoured earth-moving equipment. John Deere has developed experimental robotic equipment in this category for the US Air Force which has successfully tested it. * Cargo Handler. The Universal Self-Deployable Cargo Handler (USDCH) promises to become another valuable robot combat support device. It is the successor to the Battlefield Robotic Ammunition Supply System (BRASS), which entered development in 1984. The BRASS was to be an autonomous vehicle for handling ammunition under fire on the battlefield. It soon emerged that the computer and artificial intelligence technology then available was not sufficient to complete the project.

In mid-1988, however, the technology had evolved sufficiently for the US Army to let a contract to FMC for the design, manufacture and technical integration of the prototype, re-designated USDCH. This robotic vehicle will be an extended-reach fork lift for work in hazardous environments. Next to operating in ammunition depots, the tasks could include moving dangerous materials and handling radioactive or chemically contaminated objects. The USDCH also belongs in the remotely-controlled category of robots, and will primarily be used as a basic experimental platform which may be fitted with more sophisticated equipment than a fork lift.

FMC simultaneously received a separate contract as project leader for an advanced robotics manipulator. This will be an electrically-driven robot with two arms that reproduce human arm, wrist, and finger movements, including tactile sensors which let the system handle eggs as well as heavy objects. Cooperating with FMC are Odetics Inc., Bonneville Scientific and Stanford University.

The USDCH fitted with such a robotic system would indeed become a universal machine, capable of tasks ranging from the removal of a jet engine from an aircraft, to recharging the external ammunition magazine of a battle tank under combat conditions.

Remote-controlled Combat Robots

* Robotic Ranger. Also in the remote-controlled category of military robots is Grumman's Robotic Ranger program. Basically designed as a small, multi-purpose cross-country platform, it can carry any type of portable infantry weapon. In tests, the Robotic Ranger has been fitted with M240C machine-guns and infantry anti-tank rockets such as the Matra Manurhin Apilas and FFV AT-4. Operation by randomly-selected infantry soldiers turned out to be a complete success.

The 160 kg vehicle is powered by electric motors driving four wheels mounted at each corner of a diamond-shaped frame. Extensive studies by Grumman had indicated that this unusual wheel location provided the best compromise between all-terrain performance and sufficient platform stability for weapon firing.

The Ranger platform is only 63.5 cm high, to which must be added the height of the sensors and weapons. As

a machine-gun carrier, overall height is little more than one meter. Control is via a console which fits into an attache case containing a video display. The console is also fitted with instruments to read off distance travelled, energy remaining, current vehicle attitude, etc. Contact with the Ranger is via a two-way data-link, based on a thin, single-strand fiber-optic cable paid out by the vehicle. Maximum speed is about 16 km/h in fairly rough terrain.

An important aspect of the Ranger and similar concepts is its ease of maintenance. According to Grumman, the vehicle can be maintained in the field with readily available tools common to any automotive system. * ROBAT. A similar project designated the Robotic Obstacle-Breaching Assault Tank (ROBAT) is being designed in-house by the US Army's Tank Automotive Command (TACOM). By modifying M-60 tanks, a number of prototypes have been produced and tested. The primary task of the ROBAT is mine-clearing, though it may be adapted to alternative roles, such as that of a missile-armed tank destroyer or supply vehicle.

With the tank's turret removed and replaced by a launcher for mine-clearing rockets towing explosive cables, and the addition of a 10-tonne mine-clearing roller, the vehicle can indeed clear a reasonably safe path. The ROBAT is remotely operated by a dual-channel system.

Fiber-optics and robot-mounted sensors (a combination not unlike that used on the Ranger), as well as a back-up two-way microwave data-link are available. Fiber-optics are used if the ROBAT is handled beyond the line-of-sight of the operator. If direct eye contact can be maintained, radio-guidance is used. This dual guidance system was developed by Gould Electronics and Robot Defense Systems Inc, the latter having gained a vast experience in military robotic technology during development of the Prowler, the first prototype of an armed robot sentry ever to have been tested. * Wiesel. The US Army's Tank Automotive Command is also investigating the possibility of using the Porsche/Krupp MaK Wiesel armored vehicle as a base for the development of a robotized vehicle. Four vehicles will be involved in the programme to test remote control systems, communications equipment, computer-aided remote driving, vision sensors (stereo forward-looking and peripherical cameras and image intensifiers), reconnaissance sensors (colour video camera, forward-looking infrared devices and acoustic systems) and a navigation system (route planning and autonomous road following). * Prowler. The Prowler is an acronym for Programmable Robot Observer With Logical Enemy Response. It is a six-wheeled cross-country vehicle the size of a pick-up truck. The vehicle itself is produced by Standard Manufacturing Corp. The Prowler can be armored and fitted with a large variety of turreted weapons, ranging from shotguns to anti-tank missiles.

The Prowler family which has now grown to three members, with more under development, is intended for service as an automated guard, to watch perimeter fences of airfields, ammunition depots, or industrial zones. Such robots can also be used to guard critical border areas and even pipelines.

By using its on-board computer, the vehicle can follow a preprogrammed path, maintaining a set distance from the fence (measured by a laser range-finder), and stopping to investigate an area where a break in the fence marks an intrusion. The damaged fence is detected by the onboard sensors, which may consist of passive infrared equipment or TV cameras.

The Prowler's computers can also be used in conjunction with existing intrusion warning devices installed as a permanent feature in the protected perimeter.

Though the newest Prowler models are capable of operating independently and, with appropriate software and sensors fitted, could aim automatic fire at any intruder, they are at present kept on a tight data-link leash. The concept is radically new and unproven, and nobody in charge wants to run the risk of the machine running wild and shooting up everything its robotic brain might define as hostile.

The remote operator can see on his monitor whatever the TV or infrared "eyes" of the robot have discovered, and command the appropriate action from the Prowler. This may range from use of a spotlight with a vocal warning issued via loudspeaker, to firing the twin M-60 machine-guns.

More advanced members of the Prowler family can already find their way with the help of a digitized map loaded into the computer's memory. The input of the various sensors is continuously compared with this and any deviation can be instantly discovered and corrected.

The target acquisition and gun-laying systems work on the same principle. The computer holds in its memory images of potential enemies which would be automatically engaged upon discovery. If faced with a larger scale attack, the robot can summon the help of other Prowlers to contain the intrusion until a human intervention takes place. A robot system like the Prowler indeed represents an ideal substitution for costly, tedious and manpower-intensive guard duty, but also has the potential to evolve eventually into a fighting machine for the battlefield. * MedTec. The US Army MedTec is another fascinating project. The rapid evacuation of casualties from the battlefield has always been a difficult and hazardous task. The MedTec remotely operated robot is designed to assist in this task. It is envisioned as a vehicle of the size of the Prowler fitted with the appropriate remote-controlled manipulators to pick up the wounded and load them into its hold. * Robot tank. It is obvious that a country with an acute shortage of soldiers must mount special efforts to automatize its combat equipment. Israel is such a country, and Israel Aircraft Industries has announced that it is developing a remote-controlled main battle tank designated Pele (Miracle). All the driving and fighting functions of a tank are carried out by remote control, based on sensor and control elements not unlike those used for the remote control of unmanned aircraft. The Pele is currently in the initial development stage.

Towards the Autonomous Robot

Once the far more advanced second generation of combat robots becomes a reality in the next century, it promises to represent a totally different breed. Second-generation robots will be able to see, hear, touch, and even have a sense of smell. They will feature adequate artificial intelligence to hunt, identify and destroy hostile targets, and to become highly efficient and tireless logistical handlers. The first tentative steps towards this goal have already been taken, and considerable progress in mating artificial intelligence and robotics is becoming evident.

While the development of remote-controlled robots is solving the basic problems of structures and materials, tactile sensing and manipulator technology, there still remain numerous artificial intelligence hurdles before autonomous robots become a reality. For example: if a robot is ever to recognize depth, edges and curves, and then utilize the scanned shapes for comparison with digitized models stored in its computer memory, in order for example to avoid obstacles and chart a course, it must have a visual system that can see in three dimensions. Though engineers have developed numerous types of vision systems, robots are still limited to a basically two-dimensional view, where the depth perception is lacking. This considerably limits their usefulness at present, and intense research is being conducted to correct the deficiency. Stereoscopic vision might be achieved by using two sensors and two computers, with their outputs merged after processing, or the use of ultrasonic, laser or microwave rangefinders to provide some measure of depth. This problem seems to be close to a satisfactory solution, and it may well be that robot "eyes" will not only have a vision comparable to that of a human, but will be far superior in being able to utilize the infrared and ultra-violet bands. This will allow them to see in the dark and through fog or haze.

In May 1989, a European project named Vidimus got under way. Its goal is to achieve three-dimensional vision for robots for industrial assembly systems. To be viable, a system of this type must contain viewing technology of a type which will also be applicable to mobile robots. Project leader of the program is British Aerospace, collaborating with AEG in Germany, CEA-IRDI of France, Ibermatica of Spain, Philips subsidiaries in West Germany and the Netherlands, and Thomson-CSF.

Three-dimensional vision demands extremely fast computers and a vast storage capacity with very short retrieval time. There is little doubt that the presently employed computer generation is not adequate for handling the task. What are required are a number of supercomputers working in parallel or else other advanced architectures. An adapted version of the very fast German Suprenum computer architecture (a collaborative effort of Krupp-Atlas, Siemens, Dornier, the German Mathematics Association and a number of universities) might be a candidate for achieving the real-time computing needed in a mobile autonomous robot. Suprenum recently has a processing speed of 5 Gflops (5 billion operations per second) but promises to attain 100 Gflops eventually.

Wheels and Tracks...

Once a robot possesses rudimentary 3-D sight and a fast, powerful computer run on a top-grade artificial intelligence program, it will be able to apprehend its environment and navigate in it with precision. For testing and assessing the various proposals aimed at that target, Martin Marietta constructed the Autonomous Land Vehicle (ALV) for the US Defense Advanced Research Projects Agency (DARPA). The ALV was first tested in 1985. It currently serves as the US national test bed for developments by industry and universities in artificial intelligence software and advanced computer architectures under field conditions.

The ALV is an eight-wheeled, 6 800 kg self-contained laboratory suitable for human as well as autonomous operation. The automotive parts were designed by Standard Manufacturing and are almost identical to those of the Prowler.

In late 1984, during the first autonomous experiments performed on a smooth test track, the ALV reached the somewhat discouraging average speed of 4.8 km/h. Every 2.4 seconds the vision assembly sent an image to the computer, which interpreted the data and instructed the ALV's guidance system which way to steer to keep the vehicle on the road's centerline.

Subsequently, this performance has been gradually enhanced, in keeping with the advances made in computer and software technology. It is predicted that, by 1992, a speed of 96 km/h over any passable type of terrain will be achieved, and that the machine will be able to select the best possible path and avoid obstacles.

The first results of the ALV led the US Army to launch the Advanced Ground Vehicle Technology (AGVT) program. Contracts were let to two teams, one headed by FMC, the other by General Dynamics. FMC selected as basic platform a modified M113 which could be operated either remotely or autonomously on its own computer system. In 1986 the vehicle demonstrated its ability to perform battlefield missions, such as artillery fire direction, forward observation and reconnaissance for the discovery and identification of hostile field positions. Under remote control, it moved at speeds of up to 40 km/h, and in the autonomous mode it reached an astonishing 19 km/h.

The General Dynamics team, including Cadillac Gage and Hughes, integrated its AGVT concept into a Cadillac Gage Scout wheeled vehicle which was tested with similar results.

In 1986, however, the US Army came to the conclusion "that the available technology is not sufficiently mature to support the development of a Robotic Combat Vehicle (RCV)". One year later, both teams presented refined versions of their prototype, both of which showed notable progress during a field demonstration.

As a result of this effort, a multi-purpose RCV platform is now envisioned that can be equipped with a variety of mission modules. These can be straightforward combat or combat support mission-oriented. It was also discovered that a special robot command center (RCC) would have to be developed to control or keep track of several robots simultaneously. In late 1987 FMC was awarded a contract for the development of an RCC which would be able to handle two remote-controlled robots and two autonomous robots simultaneously. The control center would be installed on a derivative of the M-109 SP-howitzer chassis, so as to provide protection coupled with mobility. For the future, such RCCs might also be used to control other smart weapons such as UAVS and stand-off air-to-ground weapons.

Numerous other programs have been initiated in the USA. On the basis of the success with Grumman's Ranger, for example, the Tele-operated Mobile Anti-Armor Program (TMAP) was launched in which Martin Marietta is working with Grumman on advanced concepts. Robot Defense System's Prowler is undergoing further refinements to develop it into an effective anti-tank robot.

... Versus Legs

In the meantime, it has been discovered that use of wheeled vehicles increases the artificial intelligence requirements over those needed for legged robots - also referred to as mechanical walkers. The reason is that a wheeled type must generate and execute instructions to navigate its way around an obstruction, versus simply climbing over it. The armed forces as potential users, however, are not too convinced that the latter is feasible, because legged robots appear to be too vulnerable in a hostile environment. However, this is an assumption which is currently under intense discussion.

The American firm Odetics Inc. specializes in legged robots and several years ago constructed a compact eight-legged experimental robot which resembles a giant spider. The vision sensors are housed under a perspex cupola and the system is remote-controlled. More advanced robots of this Odex family are designed to operate autonomously. The early Odetics robot weighs 180 kg but is capable of lifting a one-ton weight.

Ohio State University has constructed the Adaptive Suspension Vehicle (ASV), an experimental semi-autonomous behemoth robot weighing some 2300 kg. It features six computer-commanded legs which are capable of automatic foothold selection, while the direction of travel is commanded by a driver. The payload is expected to be around 225 kg. The goal is to provide fast transport over terrain which is impassable for wheeled and tracked vehicles.

Conclusion

It seems though that, for the time being, the emphasis in all projects is being placed on remote operation. This appears to be for two reasons. First, the current state-of-the-art in remote-controlled robots is such that it will allow weapon systems to be fielded at short notice, if necessary, but only under positive human control.

This leads to the second reason. Among the military, as well as budget authorities, there remain grave reservations about fielding armed robots which might escape human control once they are turned loose. The psychological barriers which have to be crossed before full trust can be placed in the performance of fragile processors, and possibly corrupted software, in charge of lethal weaponry is understandably high.

These barriers may, quite possibly, never be crossed.

PHOTO : Based on Odetics' ODEX 1, the SRL Walking Robot is variable in height and profile. It has

PHOTO : a lifting power of 300 lbs.

PHOTO : Introduced by Odetics in 1983, ODEX 1 was the first "functionoid", a walking machine

PHOTO : capable of performing a multiplicity of tasks too hazardous for Man.

PHOTO : On the lines of the British "Wheelbarrow", this Security-Explosive Ordnance Disposal Robot

PHOTO : from Standard Manufacturing Company was shown at AUSA '89.

PHOTO : This robotic "hand", developed by the University of Southampton, can lift a 100-kg load

PHOTO : and also unscrew a light bulb.

PHOTO : An early Hunting Engineering robot - a mobile remote - controlled robot weapon platform,

PHOTO : the Hunter - seen climbing stairs.

PHOTO : Grumman's experimental TMAP (Tele-operated Mobile Anti-Armor Program), like its Robotic

PHOTO : Ranger, has four wheels in a diamond-shaped configuration.

PHOTO : The US Army's Tank Automotive Command is examining the possibility of using the

PHOTO : Porsche/Krupp MaK Whezel AWC as a base for a robotic combat vehicle.

PHOTO : A team headed by General Dynamics incl. Hughes and Cadillac Gage came up with this concept

PHOTO : in response to the Army's AGVT Program, based on CG's Scout Car.

PHOTO : Schematic illustration of Martin Marietta's proposed Field Material Handling Robot

PHOTO : (FMR) - a vision of the future?