by Paul Scharre
The first successful PGM was the German G7e/T4 Falke (“Falcon”) torpedo, introduced in 1943. The Falcon torpedo incorporated a new technological innovation: an acoustic homing seeker. Unlike regular torpedoes that traveled in a straight line and could very well miss a passing ship, the Falcon used its homing seeker to account for aiming errors. After traveling 400 meters from the German U-boat (submarine) that launched it, the Falcon would activate its passive acoustic sensors, listening for any nearby merchant ships. It would then steer toward any ships, detonating once it reached them.
The Falcon was used by only three U-boats in combat before being replaced by the upgraded G7es/T5 Zaunkönig (“Wren”), which had a faster motor and therefore could hit faster moving Allied navy ships in addition to merchant vessels. Using a torpedo that could home in on targets rather than travel in a straight line had clear military advantages, but it also immediately created complications. Two U-boats were sunk in December 1943 (U-972) and January 1944 (U-377) when their torpedoes circled back on them, homing in on the sound of their own propeller. In response to this problem, Germany instituted a 400-meter safety limit before activating the homing mechanism. To more fully mitigate against the dangers of a homing torpedo turning back on oneself, German U-boats also began incorporating a tactic of diving immediately after launch and then going completely silent.
The Allies quickly developed a countermeasure to the Wren torpedo. The Foxer, an acoustic decoy towed behind Allied ships, was intended to lure away the Wren so that it detonated harmlessly against the decoy, not the ship itself. The Foxer introduced other problems; it loudly broadcast the Allied convoy’s position to other nearby U-boats, and it wasn’t long before the Germans introduced the Wren II with an improved acoustic seeker. Thus began the arms race in smart weapons and countermeasures against them.
PRECISION-GUIDED MUNITIONS
The latter half of the twentieth century saw the expansion of PGMs like the Wren into sea, air, and ground combat. Today, they are widely used by militaries around the world in a variety of forms. Sometimes called “smart missiles” or “smart bombs,” PGMs use automation to correct for aiming errors and help guide the munition (missile, bomb, or torpedo) onto the intended target. Depending on their guidance mechanism, PGMs can have varying degrees of autonomy.
Some guided munitions have very little autonomy at all, with the human controlling the aimpoint of the weapon throughout its flight. Command-guided weapons are manually controlled by a human remotely via a wire or radio link. For other weapons, a human operator “paints” the target with a laser or radar and the missile or bomb homes in on the laser or radar reflection. In these cases, the human doesn’t directly control the movements of the munition, but does control the weapon’s aimpoint in real time. This allows the human controller to redirect the munition in flight or potentially abort the attack.
Other PGMs are “autonomous” in the sense that they cannot be recalled once launched, but the munition’s flight path and target are predetermined. These munitions can use a variety of guidance mechanisms. Nuclear-tipped ballistic missiles use inertial navigation systems consisting of gyroscopes and accelerometers to guide the missile to its preselected target point. Submarine-launched nuclear ballistic missiles use star-tracking celestial navigation systems to orient the missile, since the undersea launching point varies. Many cruise missiles look down to earth rather than up to the stars for navigation, using radar or digital scene mapping to follow the contours of the Earth to their preselected target. GPS-guided weapons rely on signals from the constellation of U.S. global positioning system satellites to determine their position and guidance to their target. While many of these munitions cannot be recalled or redirected after launch, the munitions do not have any freedom to select their own targets or even their own navigational route. In terms of the task they are performing, they have very little autonomy, even if they are beyond human control once launched. Their movements are entirely predetermined. The guidance systems, whether internal such as inertial navigation or external such as GPS, are only designed to ensure the munition stays on path to its preprogrammed target. The limitation of these guidance systems, however, is that they are only useful against fixed targets.
Homing weapons are a type of PGM used to track onto moving targets. By necessity since the target is moving, homing munitions have the ability to sense the target and adapt to its movements. Some homing munitions use passive sensors to detect their targets, as the Wren did. Passive sensors listen to or observe the environment and wait for the target to indicate its position by making noise or emitting in the electromagnetic spectrum. Active seekers send out signals, such as radar, to sense a target. An early U.S. active homing munition was the Bat anti-ship glide bomb, which had an active radar seeker to target enemy ships.
Some homing munitions “lock” onto a target, their seeker sensing the target before launch. Other munitions “lock on” after launch; they are launched with the seeker turned off, then it activates to begin looking for the moving target.
An attack dog is a good metaphor for a fire-and-forget homing munition. U.S. pilots refer to the tactic of launching the AIM-120 AMRAAM air-to-air missile in “lock on after launch” mode as going “maddog.” After the weapon is released, it turns on its active radar seeker and begins looking for targets. Like a mad dog in a meat locker, it will go after the first target it sees. Similar to the problem German U-boats faced with the Wren, pilots need to take care to ensure that the missile doesn’t track onto friendly targets. Militaries around the world often use tactics, techniques, and procedures (“TTPs” in military parlance) to avoid homing munitions turning back on themselves or other friendlies, such as the U-boat tactic of diving immediately after firing.
HOMING MUNITIONS HAVE LIMITED AUTONOMY
Homing munitions have some autonomy, but they are not “autonomous weapons”—a human still decides which specific target to attack. It’s true that many homing munitions are “fire and forget.” Once launched, they cannot be recalled. But this is hardly a new development in war. Projectiles have always been “fire and forget” since the sling and stone. Rocks, arrows, and bullets can’t be recalled after being released either. What makes homing munitions different is their rudimentary onboard intelligence to guide their behavior. They can sense the environment (the target), determine the right course of action (which way to turn), and then act (maneuvering to hit the target). They are, in essence, a simple robot.
The autonomy given to a homing munition is tightly constrained, however. Homing munitions aren’t designed to search for and hunt potential targets on their own. The munition simply uses automation to ensure it hits the specific target the human intended. They are like an attack dog sent by police to run down a suspect, not like a wild dog roaming the streets deciding on its own whom to attack.
In some cases, automation is used to ensure the munition does not hit unintended targets. The Harpoon anti-ship missile has a mode where the seeker stays off while the missile uses inertial navigation to fly a zigzag pattern toward the target. Then, at the designated location, the seeker activates to search for the intended target. This allows the missile to fly past other ships in the environment without engaging them. Because the autonomy of homing munitions is tightly constrained, the human operator needs to be aware of a specific target in advance. There must be some kind of intelligence informing the human of that particular target at that specific time and place. This intelligence could come from radars based on ships or aircraft, a ping on a submarine’s sonar, information from satellites, or some other indicator. Homing munitions have a very limited ability in time and space to search for targets, and to launch one without knowledge of a specific target would be a waste. This means homing munitions must operate as part of a broader weapon system to be useful.
THE WEAPON SYSTEM
A weapon system consists of a sensor to search for and detect enemy targets, a decision-making element that decides whether to engage the target, a
nd a munition (or other effector, such as a laser) that engages the target. Sometimes the weapon system is contained on a single platform, such as an aircraft. In the case of an Advanced Medium-Range Air-to-Air Missile (or AMRAAM), for example, the weapon system consists of the aircraft, radar, pilot, and missile. The radar searches for and senses the target, the human decides whether to engage, and the missile carries out the engagement. All of these elements are necessary for the engagement to work.
Weapon System OODA Loop
In other cases, components of the weapon system may be distributed across multiple physical platforms. For example, a maritime patrol aircraft might detect an enemy ship and pass the location data to a nearby friendly ship, which launches a missile. Defense strategists refer to this larger, distributed system with multiple components as a battle network. Defense analyst Barry Watts described the essential role battle networks play in making precision-guided weapons effective:
Because “precision munitions” require detailed data on their intended targets or aim-points to be militarily useful—as opposed to wasteful—they require “precision information.” Indeed, the tight linkage between guided munitions and “battle networks,” whose primary reason for existence is to provide the necessary targeting information, was one of the major lessons that emerged from careful study of the US-led air campaign during Operation Desert Storm in 1991. . . . [It] is guided munitions together with the targeting networks that make these munitions “smart.” [emphasis in the original]
Automation is used for many engagement-related tasks in weapon systems and battle networks: finding, identifying, tracking, and prioritizing potential targets; timing when to fire; and maneuvering munitions to the target. For most weapon systems in use today, a human makes the decision whether to engage the target. If there is a human in the loop deciding which target(s) to engage, it is a semiautonomous weapon system.
Supervised Autonomous Weapon System (human on the loop)
In autonomous weapon systems, the entire engagement loop—searching, detecting, deciding to engage, and engaging—is automated. (For ease of use, I’ll often shorten “autonomous weapon system” to “autonomous weapon.” The terms should be treated as synonymous, with the understanding that “weapon” refers to the entire system: sensor, decision-making element, and munition.) Most weapon systems in use today are semiautonomous, but a few cross the line to autonomous weapons.
SUPERVISED AUTONOMOUS WEAPON SYSTEMS
Because homing munitions can precisely target ships, bases, and vehicles, they can overwhelm defenders through saturation attacks with waves, or “salvos” of missiles. In an era of unguided (“dumb”) munitions, defenders could simply ride out an enemy barrage, trusting that most of the incoming rounds would miss. With precision-guided (“smart”) weapons, however, the defender must find a way to actively intercept and defeat incoming munitions before they impact. More automation—this time for defensive purposes—is the logical response.
At least thirty nations currently employ supervised autonomous weapon systems of various types to defend ships, vehicles, and bases from attack. Once placed in automatic mode and activated, these systems will engage incoming rockets, missiles, or mortars all on their own without further human intervention. Humans are on the loop, however, supervising their operation in real time.
Supervised Autonomous Weapon System (human on the loop)
These supervised autonomous weapons are necessary for circumstances in which the speed of engagements could overwhelm human operators. Like in the Atari game Missile Command, saturation attacks from salvos of simultaneous incoming threats could overwhelm human operators. Automated defenses are a vital part of surviving attacks from precision-guided weapons. They include ship-based defenses, such as the U.S. Aegis combat system and Phalanx Close-In Weapon System (CIWS); land-based air and missile defense systems, such as the U.S. Patriot; counter-rocket, artillery, and mortar systems such as the German MANTIS; and active protection systems for ground vehicles, such as the Israeli Trophy or Russian Arena system.
While these weapon systems are used for a variety of different situations—to defend ships, land bases, and ground vehicles—they operate in similar ways. Humans set the parameters of the weapon, establishing which threats the system should target and which it should ignore. Depending on the system, different rules may be used for threats coming from different directions, angles, and speeds. Some systems may have multiple modes of operation, allowing human in-the-loop (semiautonomous) or on-the-loop (supervised autonomous) control.
These automated defensive systems are autonomous weapons, but they have been used to date in very narrow ways—for immediate defense of human-occupied vehicles and bases, and generally targeting objects (like missiles, rockets, or aircraft), not people. Humans supervise their operation in real time and can intervene, if necessary. And the humans supervising the system are physically colocated with it, which means in principle they could physically disable it if the system stopped responding to their commands.
FULLY AUTONOMOUS WEAPON SYSTEMS
Do any nations have fully autonomous weapons that operate with no human supervision? Generally speaking, fully autonomous weapons are not in wide use, but there are a few select systems that cross the line. These weapons can search for, decide to engage, and engage targets on their own and no human can intervene. Loitering munitions are one example.
Loitering munitions can circle overhead for extended periods of time, searching for potential targets over a wide area and, once they find one, destroy it. Unlike homing munitions, loitering munitions do not require precise intelligence on enemy targets before launch. Thus, a loitering munition is a complete “weapon system” all on its own. A human can launch a loitering munition into a “box” to search for enemy targets without knowledge of any specific targets beforehand. Some loitering munitions keep humans in the loop via a radio connection to approve targets before engagement, making them semiautonomous weapon systems. Some, however, are fully autonomous.
Fully Autonomous Weapon System (human out of the loop)
The Israeli Harpy is one such weapon. No human approves the specific target before engagement. The Harpy has been sold to several countries—Chile, China, India, South Korea, and Turkey—and the Chinese are reported to have reverse engineered their own variant.
HARM vs. Harpy
Type of weapon
Target
Time to search
Distance
Degree of autonomy
HARM
Homing missile
Radars
Approx. 4.5 minutes
90+ km
Semiautonomous weapon
Harpy
Loitering munition
Radars
2.5 hours
500 km
Fully autonomous weapon
The difference between a fully autonomous loitering munition and a semiautonomous homing munition can be illustrated by comparing the Harpy with the High-speed Anti-Radiation Missile (HARM). Both go after the same type of target (enemy radars), but their freedom to search for targets is massively different. The semiautonomous HARM has a range of 90-plus kilometers and a top speed of over 1,200 kilometers per hour, so it is only airborne for approximately four and a half minutes. Because it cannot loiter, the HARM has to be launched at a specific enemy radar in order to be useful. The Harpy can stay aloft for over two and a half hours covering up to 500 kilometers of ground. This allows the Harpy to operate independently of a broader battle network that gives the human targeting information before launch. The human launching the Harpy decides to destroy any enemy radars within a general area in space and time, but the Harpy itself chooses the specific radar it destroys.
Semiautonomous vs. Fully Autonomous Weapons For semiautonomous weapons, the human operator launches the weapon at a specific known target or group of targets. The human chooses the target and the weapon carries out the attack. Fully autonomous weapons can search for and find targets ov
er a wide area, allowing human operators to launch them without knowledge of specific targets in advance. The human decides to launch the fully autonomous weapon, but the weapon itself chooses the specific target to attack.
Tomahawk Anti-Ship Missile Mission Profile A typical mission for a Tomahawk Anti-Ship Missile (TASM). After being launched from a ship or submarine, the TASM would cruise to the target area. Once over the target area, it would fly a search pattern to look for targets and, if it found one, attack the target on its own.
Despite conventional thinking that fully autonomous weapons are yet to come, isolated cases of fully autonomous loitering munitions go back decades. In the 1980s, the U.S. Navy deployed a loitering anti-ship missile that could hunt for, detect, and engage Soviet ships on its own. The Tomahawk Anti-Ship Missile (TASM) was intended to be launched over the horizon at possible locations of Soviet ships, then fly a search pattern over a wide area looking for their radar signatures. If it found a Soviet ship, TASM would attack it. (Despite the name, the TASM was quite different from the Tomahawk Land Attack Missile [TLAM], which uses digital scene mapping to follow a preprogrammed route to its target.) The TASM was taken out of Navy service in the early 1990s. While it was never fired in anger, it has the distinction being the first operational fully autonomous weapon, a significance that was not recognized at the time.
In the 1990s, the United States began development on two experimental loitering munitions: Tacit Rainbow and the Low Cost Autonomous Attack System (LOCAAS). Tacit Rainbow was intended to be a persistent antiradiation weapon to target land-based radars, like the Harpy. LOCAAS had an even more ambitious goal: to search for and destroy enemy tanks, which are harder targets than radars because they are not emitting in the electromagnetic spectrum. Neither Tacit Rainbow nor LOCAAS were ever deployed; both were cancelled while still in development.