Military History | How To Make War | Wars Around the World Rules of Use How to Behave on an Internet Forum
Military Science Fiction Discussion Board
   Return to Topic Page
Subject: What about robotic infantry?
Treadgar    11/27/2006 10:03:54 AM
In the early 90s I used to play this game called F-19 Stealth Fighter. You flew missions in the Middle East and Russia. As I flew my simulated fighter I imagined my control inputs were being beamed to an unmanned fighter that was flying half a world away. Because my computer was very slow the game play was choppy. My imagination said this was an artifact of jamming or some type of enemy action along those lines. In the real world we see this happening with the predator. There are predators that are now armed. It seems to me this trend will continue. The question is, how far does anyone who reads this think it will go? Will there be robotic infantry, soon, or is that a hundred years in the future? Here are some more questions: 1. If humans control these robotic soldiers remotely, how would such a force be organized? 2. If the ability to jam transmissions is a function of distance do you think it would be best to have controllers near the FEBA to exert localized control? 3. What degree of autonomy would such units have? 4. What might robotic infantry look like? Would you have anthropomorphic designs or something else? 5. Does anyone think a science fiction novel with this kind of setup would be interesting? Treadgar
Quote    Reply

Show Only Poster Name and Title     Newest to Oldest
Pages: PREV  1 2 3 4   NEXT
Jeff_F_F       1/3/2007 2:24:57 PM
AI helps the bandwith problem because the more decisions that can be made onboard, the less information needs to be transmitted to a controller. If the robots have virtually no AI, the battle is like an FPS, and you need a lot of bandwith. If the robots have enough AI to determine a potential target but cannot be depended upon to distinguish enemy from non-combatants, the bandwith issue becomes easier, the human controller only needs to participate while identifying targets, and controlling the overall tactics of the unit. The effective limit on the controller to robot ratio would be determined primarily by the number of targets that could be identified by a single controller, since the more robots a controller was responsible for, the more targets would be encountered. Several controllers could be given queue similar to a cutomer service people phone system, so that those controller's whose robots were not in a target-rich area could help out the controllers whose robots were encountering more targets. With GPS onboard each robot and targets being identified at a range and bearing from the robot, all the robots would be able to use the identification information. The locations of the robots and their targets could be mapped against a 3D digital map of the terrain. 
If the robots were able to differentiate between enemy and non-combatants, the amount of bandwith would be further reduced, as only the locations of the robots and their targets would need to be transmitted. The controller would no longer be limited by the number of targets they were able to identify, and the only limit would now be the operator's effective span of control as they determine tactics. The general rule among game designers is that about a dozen units can be controlled in a RTS game with optimal efficiency so each controller would probably command about a squad. As AI gets better and better, small units of robots may be able to coordinate with each other, which would increase the span of control of each operator. For example, if a 4-robot team is able to coordinate through their mutual AI, and can thus be controlled as a single unit by the human operator, a human operator can now control an entire platoon. This process is likely to escalate with larget and larger units working together. Presumably humans would want to maintain control of operational size units, and at least of strategic command, otherwise you start getting into Skynet / VIKI (from I, Robot) / Matrix scenarios.
Quote    Reply

tomadog02       1/3/2007 3:04:39 PM
I think Treadgar has raised the most important question regarding robotic infantry that is controlled by human operators: reaction speed.  While the technology exists to run near-real time operations, it is still near not actual real time operations.  The SF book I refered to earlier solved this problem through some sort of full-sensory uplink that does not exist nor will in the forseable future.

Any robotic soldier who is remote controlled will face a moderate to extreme disadvantage when faced with a human opponent in terms of reaction time due to both the bandwidth issues and mechanical reaction time.  While this might not matter in certain situations, say rolling the robot around the corner to scout enemy positions or assualting a fixed enemy position, in a fluid battlefield this could be quite problematic.  Of course there are two ways around this: increase bandwidth or increase armor.  You can afford to be slower when you're sitting in a M1 then when you are running around on foot.

Quote    Reply

Jeff_F_F       1/4/2007 1:01:25 PM
Very good call. Ironically, AI would likely make the situation worse at first, if human operators had to sign off on targets before robots could engage them. This would be especially pronounced if the targets were encountered faster than the operators were able to evaluate them, causing a queue to build up. I agree that armor is the simplest way to offset this. Increasing firepower can have a similar effect - you can afford to react more slowly if when you do react it is with firepower able to bring about a more rapidly decisive decision.
On the other hand, the nature of one's opposition must be considered. It is entirely possible that even with the delayed response due to remote control, combat robots may still be able to react faster than human opponents under combat conditions. For one thing, a robot isn't going to suffer fatigue from running around all day in the sun wearing loads of armor and weapons and ammo and computers and batteries. Humans also do not necessarily respond ideally in combat, which is why soldier experience is so critical. Fear can exert a powerful impact on our ability to respond to dangerous situations, because the fear response operates completely seperately from the concious brain and tends to override the concious brain. Part of the advantage of realistic training is that is appears make this effect less pronounced. Since a robotic soldier's controller is not putting their personal hide on the line, this fear effect is going to be a non-issue. This fact also has important consequences for tactics, allowing controllers to make more aggressive assaults and to defend past the point when a human would be forced to retreat. Also consider that the law of avarages works to limit the experience that human infantrymen and other combatants can gain before being killed or wounded sufficiently that they can no longer fight. Obviously this does not apply to robot controllers allowing effectively unlimited levels of skill to be gained. Also for a human soldier there is always a difference between even the best simulation and actual combat, while for a robot controller simulation and combat could be made nearly identical.
Beyond these factors, consider some of what goes into reacting to a combat situation : one must perceive the situation with their senses, evaluate the situation, decide how to react, transmit the decision to the weapons one is using, and those weapons must put your decision into effect. There are two bottlenecks that remote control imposes : Perceiving via remote sensors and transmission of a decision to the remote weapons. However the perception of the situation is more complex than simply pointing a camera at a target and deciding to fire at it as one would in a typical FPS, especially on an increasingly net-centric battlefield. It has already been noted elsewhere on this site how much of an advantage net centric technologies give our forces. This is primarily because it effectively extends their ability to perceive their environment beyond the reach of a soldier's senses, and allows soldiers and their leadership to evaluate the situation before they make direct contact with the enemy. Soldiers and their leaders can then formulate tentative responses to the situation before the situation is encountered. Thus when contact is made the enemy, our soldiers are vastly more prepared to respond to that contact than their opponents.
Even if the actual reception of data from a robot's camera or other sensor is delayed, that data is only one part of the perception of the situation, and even if the order to respond is similarly delayed, because the controller would probably be able to formulate a response ahead of time, they would be send the order at the fastest rate that the data connection would allow, which might very well still be faster than an unprepared enemy might be able to respond. A robot would also likely have access to at least as many sensors as a human such as lowlight and thermal sights, and might well have many more, such as sophisticated accoustic sensors like the sniper detection sensors being mounted on some vehicles, FIDO - type IED sniffers, or the sensors being developed that allow vision through walls. Thus a combat robot because of its better access to data might allow a controller to engage targets in ways that a human would simply be unable to match. 
Another factor to consider is a combat robot may well have access to more effective weapons than a human would. A human is generally limited to a single weapon, along with weapons such as hand grenades. A combat robot might well have several weapons available to use as needed, such as grenade launchers-possibly with a variety
Quote    Reply

Treadgar       1/25/2007 11:51:43 AM
I like the idea about a human controling 12 units, or what might be the equivalent of a squad. Some operations might require more control/flexibility. Others might demand coordination of smaller units at higher operational levels. Movement to a phaseline might require more centralized control, and when bottlenecks occur you would have humans intervening to hopefully open up the bottleneck. Some situations would demand large scale coordination, others would not. So sometimes you'd have human controlers on standby while their units are under a more centralized control regime.
The customer service analogy is a good one, but clearly you wouldn't always want to respond to each crisis in the order in which the occur. You'd need to prioritize in accordance with the situation, and hopefully you'd prioritize correctly through the fog of battle.
Another point I liked was the one about experience. Someone wrote that the human controler, relatively safely removed from the battle could gain experience and improve control capabilities to a far greater degree than would be possible for someone directly involved in the fighting. What do they say? Green troops might have lots of enthusiasm, but little skill. As troops become more experienced the enthusiasm goes down a little bit. They start to be more and more concerned about their mortality. This reminds me of the Cylon Scar in Battlestar Galactica. Whenever Scar was "killed" Scar's personality would be beamed back to the Cylon base ship where  Scar could be resurrected to fight again. Even in "death" Scar was learning how to fight. Of course in that episode there was some discussion about Scar becoming disturbed about dying over and over again. That wouldn't seem to be a problem in the scenarios collectively imagined here.
Another poster (sorry I can't remember names right now as my access to the internet is limited lately) wrote something about bandwidth and how AI capability could reduce that. Target identification capabilities would indeed serve to limit the bandwidth problem. Doing that is easier said than done, or so it seems to me. I can imagine an IFF system, where targets are queried, but  that takes time. Then there are issues about when to allow units to freely fire upon targets without controler consent.
Quote    Reply

Treadgar       1/25/2007 11:58:22 AM
"Another factor to consider is a combat robot may well have access to more effective weapons than a human would. A human is generally limited to a single weapon, along with weapons such as hand grenades. A combat robot might well have several weapons available to use as needed, such as grenade launchers-possibly with a variety of grenades available(HE/DP, smoke, flash-bang, CS, etc) LMG or SAW, and perhaps rocket launchers or mortars. These weapons could be mounted on arms with independent sensors allowing them to be fired around corners or over obstacles as effectively as a human soldier would be able to make normal aimed fire."
This is another good point. I thought of this when I ply the first person shooter games. Doom is the first to come to mind. There our hero has different weapons you can cycle through: shotgun, pistol, chainsaw, BFG 9000, and others. Even in some of the newer games (like the Big Red One) you have choices of different weapons (like an M1 Garand, and Thompson MG, and my personal favorite the 1903 Springfield with a scope). Of course no human could possibly carry all these weapons in real life and actually be able to use them in combat. With a robot we can abandon the human form and design a platform that could indeed perform as described in the above quote. Of course you would probably have to have other units that come by and drop off those ammo resupply packets.
Quote    Reply

dll2000       1/25/2007 5:03:24 PM
Ignoring ethical considerations, what about a situation where you were using deceased men reanimated with electrical impulses and an AI computer chip or could even be remotely controlled?  It would eliminate most design problems as human beings are already equpped with opposable thumbs and the ablilty to navigate most terrain and obstacles. 
Now that I think about it, I think this was the premise for a terrible movie starring Jean Claude Van Dam sp? - Universal Soldier.
Didnt work out so well in the movie, because the soldiers retrieved memories from their former life.
Quote    Reply

dll2000       1/25/2007 5:10:48 PM
I think I also read a book a long while ago, 'Hong Kong' I think where kangaroo-like robots assisted in a democratic revolution in China.  They jumped around killng Chinese soldiers.
Not a military man, but I believe I have read that AC-130's boast technology that can differentiate between friend and foe.
Quote    Reply

tomadog02       1/25/2007 8:13:00 PM

Thus a combat robot because of its better access to data might allow a controller to engage targets in ways that a human would simply be unable to match. 


Expanding upon this idea, but referring back to my original concept of the “swarm” of robots under a singular human controller, imagaine a singular force of say, 3 robots.  All feed into the same relay system and all share data amongst each other.  However, the twist is this: 1 airborne, 1 armed and 1 dedicated sensor platform.  A design that was both armored and carried the array of sensors you’re depicting would grow larger and therefore less mobile.  Obviously the armored bot would retain some level of sensory perception so that it would not be blinded by the loss of the sensor bot, but by not compromising on design you allow more armor and more tactical vision. 


This is a scenario I envision:  Operator sets an aerial bot (say a modified Raven) on an aerial swing.  If you’ve ever played any Command-and-Conquer or Warcraft-esque strategy game, I’m envisioning a simple “patrol” command that simply sets the bot circling over a given piece of territory.  The operator now has an aerial view of the battlefield that he does not have to control, but can glance at on demand.  The sensor bot takes up a position slightly behing the armored bot, in a trail capability.  Armored bot moves in, takes fire, withdraws.  Sensor bot, using acoustic sensors, lowlight, thermal, etc…determines where armored bot has taken fire from.  Armored bot rolss back out and reengages. 

Quote    Reply

Treadgar       1/26/2007 12:06:02 PM
"This is a scenario I envision:  Operator sets an aerial bot (say a modified Raven) on an aerial swing.  If you’ve ever played any Command-and-Conquer or Warcraft-esque strategy game, I’m envisioning a simple “patrol” command that simply sets the bot circling over a given piece of territory."

Tomadog02, this is exactly the kind of scenario I envision, and it has been influenced by Blizzard games like Warcraft and Starcraft. Wherever you have a unit you have awareness, but I've often wondered how you could do that in real life? I've thought of aerial surveillance assets like you did. Different types of units, and combining them to offseat weaknesses and to augment strengths. I've played some of these games on line, and I thought that was interesting, esp. when I had human allies, and we would coordinate our activities against our enemy be they computer or other humans. This might be a window to future wars, but I still can't see how we can easily translate this into a realworld scenario without considerable technological development. That doesn't mean it's impossible. I think it's coming, and it might go faster than we now believe.

Quote    Reply

ens. jack       2/6/2007 2:47:19 PM
all you need to do is fit a Phalanx onto a tank chassis, it'll be taller, but it would be devastating against infantry, plus 20mm shells could be good against armor. a squad of about five of those could be commanded by a mobile hq in the battle line with a decent degree of efficiency. more armor would need to go into the hq, but thats a fair trade if you ask me. plus hq could be used to call in airstrikes, etc. that a drone wouldn't be able to do.
Quote    Reply
PREV  1 2 3 4   NEXT