17
$\begingroup$

What system do they use to distinguish robots from inanimate parts?

I previously asked how a robot would know what metals to build itself with in this question. That made me think about an other important question. Otherwise they might rip each other to shreds whenever they need materials to build... other robots. Or they might confuse a shop of spare parts with a graveyard. Well I could just circumvent this by having the robots networked so they know at all times which unit is where. But I want my machines to be autonomous and most importantly with a sense of self.

Robby sits in a pile of scrap metal. Is Robby made of scrap? Or is the scrap made of Robby? He beeps, for he does not know.

Robots in my setting come in all shapes and sizes. There is no one universal shape that defines machines as individuals or part of a group. They can freely switch parts to optimize themselves or replace every part of themselves with new ones. A robot might even change their function from factory overseer to farming bot if their consciousness program decided to do so. This incredible variety means that they need a system to distinguish what is a robot and what is a spare part or piece of scrap.

They could check their CPUs to make sure they are active. However when they are turned off (or resting) this does not work. Checking if there is a power signature of some kind could also be an option but again it could be turned off. What is your take on the issue?

$\endgroup$
16
  • 8
    $\begingroup$ A frame challenge - robots can't. So this is a cutthroat robotic world. $\endgroup$
    – Alexander
    Nov 7, 2021 at 18:31
  • 7
    $\begingroup$ Riff on old-biker saying: "If it moves, don't scrap it, if it doesn't move - kick it 'till it moves then don't scrap it." $\endgroup$ Nov 7, 2021 at 18:57
  • 10
    $\begingroup$ I like the method used in Alita: Battle Angel. They determined her scrapped head was still alive because it was very cute. $\endgroup$
    – Willk
    Nov 7, 2021 at 20:06
  • 4
    $\begingroup$ I think if you're going to treat these robots like life, you need to have the rules for life. If you don't recognize it as your kind of life, or value it as life, it's scrap (living or no). A refrigeration unit with no self-awareness may be valued to you, while a rival group of robots might be actively pursued. How do ants know when not to eat another ant? There's a giant war in California, and an ant can be moved a hundred miles and seamlessly be recognized as "in," while an ant can be moved two feet and be ripped apart instantly. $\endgroup$
    – DWKraus
    Nov 7, 2021 at 20:17
  • 2
    $\begingroup$ Have you read the prologue chapter of Code Of The Lifemaker (Hogan, 1983)? (The answer there is: They don't, instead they evolve mechanisms to flee one another or to defend themselves from other scrap-seeking robots ...) $\endgroup$
    – davidbak
    Nov 8, 2021 at 17:53

7 Answers 7

17
$\begingroup$

Give robots a life signal. Something, anything by which they can be recognized as a robot and not lifeless scrap, even when they are powered off.
Your computer and phone still have some functionalities while they're powered off. (Ever tried to set an alarm then turn your phone off? On my phone alarms rings even while it's off...) Why should your robots be any different?
Some ideas:

  • A heartbeat
  • A faint internal fan to cool it off
  • A laser light coming out of their eyes
  • An internal clock that makes a faint noise that others robots can pick up on
  • An internal GPS
  • Or anything really, so long as it can be recognized by other robots
    Based on the question, it sounds reasonable to assume that your robots could turn themselves on. This means that some part of them still has power. This part can also power their life signal.
$\endgroup$
8
  • 5
    $\begingroup$ "Better turn my phone off to not accidentally disturb everyone else in the theatre and thoroughly embarrass myself" [...] BEEP BEEP BEEP BEEP... — "Thanks a lot, designer. Thanks. A. Lot." $\endgroup$
    – Andreas
    Nov 8, 2021 at 13:40
  • 2
    $\begingroup$ It could be as simple as a basic EM field, like a wireless/BlueTooth signal that acts as a notification that a robot is still viable, even if it is powered off currently. In fact, most electronics don't fully power off anyway. Things like TVs, cars, optical disk players, VCRs, coffee maker, and anything else that uses a remote or otherwise turns itself on doesn't fully shut down, since it still needs to listen for the remote or timer setting. And that amount of power is all that's really needed for a small EM field to exist. $\endgroup$ Nov 8, 2021 at 16:10
  • 5
    $\begingroup$ You could even use a zero-power display to indicate when a completely powered down robot was last active. Even if a robot has absolutely zero electrical or mechanical activity, knowing that it was last active 10 minutes ago may help you differentiate it from a scrap robot that hasn't been active for 10 years. Not a perfect system, but perhaps useful for differentiating among machines that are truly inactive. $\endgroup$ Nov 8, 2021 at 16:59
  • 4
    $\begingroup$ "Your computer and phone still have some functionalities while they're powered off." No, they don't. You're thinking of devices that are in a low-power state. You'd have to move to something like a self-powering RFID responder to truly have "functionalities while [...] powered off". $\endgroup$
    – Matthew
    Nov 8, 2021 at 21:00
  • 2
    $\begingroup$ IIRC computers have small, round, rechargeable, button-like batteries (CMOS battery may be the name), primarily to keep track of time and make sure scheduled maintenance tasks will be carried out when the computer is turned back on. A robot may try to scan the signatures for this specific battery to determine the probability of an offline robot actually alive. $\endgroup$ Nov 9, 2021 at 18:47
20
$\begingroup$

Try to power it up

A powered-up robot has signs of life. It might move out of the way to avoid being eaten--and if it doesn't, it'll have distinctive magnetic fields and incidental radio and thermal emissions.

Now, yes, robots can be completely powered down without being really dead, unlike humans; but humans haven't always been great at identifying when other humans are really dead, either! And if you're not sure, what do you do? You try to wake them up! You try to revive them! If they don't wake up right away, maybe you check their batteries and try to repair them yourself... but if surgery fails, you slate for "organ donation".

If, that is, you care. If you don't care... a powered-down robot is easy prey! And thus, you end up with evolutionary pressures acting on prey robots and predator robots, and eventually a whole robotic ecosystem a la Code of the Lifemaker.

$\endgroup$
4
  • $\begingroup$ But if it died because of power issues, then it'd be like charging a dead battery. Sure, it will get started and be fine while it runs on alternator power, but the second it's off, it's dead again. How do you differentiate between a dead robot that has been "jumped" and one that's still alive? $\endgroup$
    – Anoplexian
    Nov 8, 2021 at 16:54
  • 1
    $\begingroup$ @Anoplexian a dead robot that has been "jumped" would have corrupted memory or other issues that prevent it from functioning correctly. If it's fine but currently requires an external power source it's not a dead robot, just one in need of an "organ" transplant. $\endgroup$
    – Rob Watts
    Nov 8, 2021 at 19:42
  • $\begingroup$ @Anoplexian As you have just discovered, the line between "life" and "death" is not any more clear-cut for robots than it is for humans. I'd say that if it has been jumped is currently functioning, it is alive. And it's not really dead until it cannot be revived again by any means. If you disagree, then what should happen when it fails again depends on how your robots define "death". If you can actually provide a precise operational definition of death, then how to detect it will be obvious. $\endgroup$ Nov 8, 2021 at 20:27
  • $\begingroup$ @Anoplexian Any decent robot would have a full suite of self-diagnostics software to check all its systems and report any issues. So long as the robot does not value its own existence, it should self-report any critically failing systems. So you jump-start the robot and order it to perform a self-diagnostic to determine if it's worth repairing, otherwise, time for the scrap-heap. $\endgroup$ Nov 9, 2021 at 18:42
9
$\begingroup$

How do we tell a living human from a dead one? The living human has vital traces: the heart will let an electrocardiogram record some signal, the brain will let an electroencephalogram do the same.

And for a robot, search for traces of electrical activity that distinguish it from junk metal.

A simple RFID can be used even when the robot is in sleep mode to wake it up on call. The same RFID can be used to probe its robotiness.

$\endgroup$
3
  • 2
    $\begingroup$ Electronics can be completely powered down, meaning that there would be no electrical activity. Still, once power is supplied, they can turn on. $\endgroup$
    – Alexander
    Nov 7, 2021 at 18:29
  • 3
    $\begingroup$ @Alexander passive RFID is powered by the reader, not by the tag. A completely depowered robot with a passive RFID tag would still respond to a RFID read. $\endgroup$ Nov 7, 2021 at 18:38
  • 2
    $\begingroup$ @TheSquare-CubeLaw Though if it’s being used to do more than just get data back from the tag, it’s arguably no longer RFID. It would probably be something like NFC, which still allows for drawing power from the transmitter if designed correctly (for example, as is done with NFC-capable U2F keys or chip-cards). $\endgroup$ Nov 8, 2021 at 3:06
7
$\begingroup$

Only things that have actively been marked as scrap are scrap.

This is a "fail robot" situation. We can always scrap robots. We can't unscrap them. Therefore there are special markers put on scrap, and if they fall off, the scrap has to be remarked.

$\endgroup$
2
  • 1
    $\begingroup$ Sure we can un-scrap them. The robot's friends just need to buy back a new set of parts. What we really need to do is track ownership of scrap. If an AI owns it's body, then there would be rules and registrations. Protocols for "condemning" broken robots and payments for scrapping. CPU's might need to be "validated" as disabled before being torn apart for scrap - or not, depending on cultural norms. $\endgroup$
    – DWKraus
    Nov 7, 2021 at 20:22
  • $\begingroup$ Whilst good as a concept (and allows for story arcs involving "live" robots incorrectly marked as scrap, and the battle to remove the marker by the robot itself if it's a victim, or by the union of scavenger robots which keep getting destroyed by an evil predator that puts a scrap marker on a heavily defended part of itself), this merely pushes the question back one level - what determines where to place the special markers, and how? $\endgroup$
    – Steve
    Nov 8, 2021 at 10:36
4
$\begingroup$

Logan and L.Dutch are on the right path, but...

A robot that's active, or even in a low-power state, isn't an issue. Such a robot can simply provide some signal when queried. Your problem is with robots that are completely depowered, either for long-term storage, or due to damage or because their power supply simply ran out.

As an aside: there's a cop-out here; whatever gives robots "consciousness" must be constantly maintained, as with biological life. Total loss of power equals "death"; therefore, a robot that can't reply to a simple query is scrap. But let's assume you don't want to go that way.

In that case, a robot must have a passive RFID tag that a) identifies it (doesn't have to identify an individual, depending on how you feel about privacy, but at least says "yes, I'm a robot") and b) provides instruction how to power it up if it is totally offline (either in the tag, or via some sort of "link" provided by the tag). You need the latter so that other robots can determine if they're dealing with a robot that is merely totally powered off (or has a dead/faulty power supply) or one that is no longer functional and could be recycled.

$\endgroup$
3
$\begingroup$

The standard answer seems to be to have some sort of active system that is detectable, and as the question was stated I'm all in favor of that as a simple first test. If you can detect emissions from the active systems then the robot is probably not scrap. Probably.

The real question for me is what defines a robot as being 'alive' or scrap? If a robot is powered down but just needs an initialisation to return to full power, it is clearly not scrap. But a robot whose CPU has been destroyed is scrap, regardless of whether or not the robot's body is still powered. Likewise if the robots programming is corrupted to the point that consciousness cannot be restored is clearly scrap.

This greatly complicates things since a scrap robot could be emitting detectable signals while a 'live' robot may not.

Assuming that a robot's consciousness can survive being completely powered down - which isn't necessarily true, depending on how that consciousness functions - then even ancient rusted piles of apparent scrap could in fact be live, just waiting for a new power supply to become fully conscious again.

The question is almost certainly not completely answerable under these conditions. Instead there will have to be some thresholds where even a powered down robot is considered non-viable and can be scavenged.

Robots should be fitted with a passive transponder that reacts to radio queries with a simple timestamp. The timestamp is updated periodically by the robot's main program as a watchdog process. When the robot powers off or the main program is otherwise not running - due to the CPU being destroyed for instance - then the timestamp will be out of date. This allows detection of end-of-life for otherwise functional or emitting robots.

To help with 'powered down' detection the timestamp should also have a flag that is only set when the robot is in the process of powering down. During normal operation the flag is cleared.

Now let's say Robbie (our active, damaged robot) is out prospecting for parts to repair itself. It happens across a robot that looks like a good source of those parts and proceeds to query the transponder. It gets back a timestamp from 6 months ago, but the 'powered down normally' flag is set. This is not a viable source of spare parts. Continuing on, Robbie spots another robot that is emitting all sorts of interesting EM but doesn't seem to be doing anything. On query the transponder reports a timestamp from 3 days ago with the 'powered down normally' flag not set. Robbie attempts to talk to the robot but gets no response. It then proceeds to pull the robot's power supply and extract the parts it needs.

I can think of a few ways that this could fail, but there have to be limits or nothing could ever be reasonably presumed to be scrap.

$\endgroup$
2
  • 1
    $\begingroup$ Good complementary answer. I’ll use parts of it. But I wonder what would happens if a corrupted robot sets the powered down flag? $\endgroup$ Nov 9, 2021 at 5:40
  • 1
    $\begingroup$ @LiveInAmbeR No system is perfect, and that's one of the possible failures I considered. Another is a corrupted system clock reporting an active robot as scrap, or faulty link between the main system and the watchdog unit. Computers are bad enough at this stuff, no reason robots won't be as well :P $\endgroup$
    – Corey
    Nov 9, 2021 at 8:47
2
$\begingroup$

You could add a sensory detection subsystem to the robot, teach it to distinguish its surroundings in terms of sensory measures that perhaps distinguish robot from robot. Robots could then be networked in a peer-to-peer fashion in order to help other robots continuously distinguish from other robots. This can open up possibilities in creating whole robot species and types that act as a sort of hive mind like from Rick and Morty Episode “Auto Erotic Assimilation”.

$\endgroup$
1
  • 2
    $\begingroup$ I never ever would have guessed that Unity would be mentioned under my question. Have a +1 friend. $\endgroup$ Nov 8, 2021 at 19:08

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .