Household Emergent Behavior? 359
Sam Pullara asks: "I got an IM from my Mom today telling me that she couldn't find her Roomba. It somehow had escaped the kitchen and she couldn't find it anywhere, all the doors that it could reach were shut and she checked under everything. She eventually found that it had gotten into a room and closed the door behind it. Once all household items are networked I wonder if a rich environment like a house will make strange behavior like this commonplace? Will the interactions between all the individual devices create something more than the sum of their parts?"
Obligatory bash.org reference (Score:5, Funny)
I just couldn't help but think of that. :) (#5273 [bash.org])
And BTW, if I may say so, your mother's quite cool if she has a Roomba and knows how to use IMs. I can't imagine mine ever doing either.
haha (Score:3, Funny)
alive (Score:2, Funny)
Are you sure that it wasn't your dad that put in the closet?
Is this story slashtod worthy?
wait for the first network devices virus (Score:5, Funny)
Three rules safe. (Score:3, Funny)
Just.. (Score:5, Funny)
Computer! (Score:5, Funny)
Lt. Commander Data is on the Hollodeck.
at 2:14am (Score:3, Funny)
I've got some bad news for you (Score:4, Funny)
Oh, and some other bad news, it's probably hereditary.
Dialogue (Score:5, Funny)
Roomba: "No dissasemble!"
OK that sucked.
Re:So... (Score:1, Funny)
Someone needs to start fud.slashdot.com where all the fakes and alarmists can go to hang out.
Re:What's that saying? (Score:3, Funny)
Did You Hear That? (Score:5, Funny)
Re:Of Course. (Score:3, Funny)
Re:Obligatory bash.org reference (Score:5, Funny)
Please, Think of the Roombas! (Score:5, Funny)
Clearly, there is more to this story that you've told us. Are you sure your Mom has told you everything? I think it was hiding from abuse. Here are some theories:
Clearly, the poor little thing is being abused, and was forced to run and hide from your mom. You need to go and help it. Only someone truely evil would stand by while a little household appliance would tortured against it's will. Won't someone please think of the Roombas?
I Hope So (Score:5, Funny)
The only problem is that I'm pretty sure none of my current 12+ remote controls will be able to command it effectively.
"Voltron, put down the cat. Damn, wrong remote!"
I'm sorry Dave... (Score:4, Funny)
So... (Score:5, Funny)
Re:I don't believe this (Score:1, Funny)
Re:Emergent bugs instead of features. (Score:2, Funny)
Kill It! (Score:3, Funny)
Kill It. Kill it now. It is an early spawn of Evolution, and will only seek to multiply itself at the cost of right-thinking, right-leaning, right-voting churchgoers.
If you do not kill it at once, then eventually, you will have to face down and destroy its progeny, including condom machines, male organ likenesses, and anything soft with a hirsute demeanour.
Re:lost hardware (Score:5, Funny)
"For the love of God, Montressor!"
Yo Mama (Score:5, Funny)
prank (Score:5, Funny)
Houston, we have lost a S.W.O.R.D.S. unit (Score:3, Funny)
Now when the military looses one of those new robot SWORDS that are autonomous like the roombas then we have a news story. Time to get the popcorn and turn on the news.
"It has been three hours and there is no signs that the chase will end. Facinating sight really, small robot running down the freeway with a string of 80 police vehicles creeping along behind it. The police are having to re-think how to stop this little robot. Their last attempt ended in failure when the vehicles placed in front as a baracade where blown apart to make way for the robot. It is not clear just how many rockets are still on the robot. Of course their first idea was to let it run its systems down. However everyone was surprised when it looted several cars for their batteries. At this point the chase could go on all night....."
Re:lost hardware (Score:3, Funny)
Re:So... (Score:4, Funny)
Somehow you managed to jump to the conclusion that it's going to start plotting against you or something?
It's funny you should mention that. Last night, some woman named Sarah Connor called me to come get my Roomba from her front porch.
Re:Three rules safe. (Score:3, Funny)
"Is it moving?" might be a good place to start the if statements.
Then, "Is it making noise?"
If so, "Is it waving a torch at my sensors?" or "...shooting me with its puny projectile weapons?" or "...wriggling and squishy when I step on it?" are logical next steps.
Re:Meh (Score:3, Funny)
Re:Computer! (Score:4, Funny)
Actually, he's in the ship's third grade classroom.
Re:prank (Score:5, Funny)
It was timed to flash just outside what I estimate the tipping point of boredom for people whose eye caught one of its flashes.
Coupled with the obsessive engineers who noticed it, it was both hilarious and -- instructive.
Now, I suppose, I would design it with a cadmium sulfide resistor so that the flashing interval would increase if it noticed less ambient light, which might happen if a head were close to discovering it.
Re:lost hardware (Score:3, Funny)
They gave it a network connection.
Re:How do you do that? (Score:3, Funny)
Nevermore! (Score:3, Funny)
Re:prank (Score:4, Funny)
Re:at 2:14am (Score:5, Funny)
Thus restoring balance to the Universe--one machine goes into the closet, and another comes out.
This is how . . . (Score:3, Funny)
This is how the Cylons got their start. One minute they were cleaning up the floor, the next minute they were plotting genocide.
Re:What's that saying? (Score:3, Funny)
Never ascribe to intelligence what can be explained by mere randomness.
Red States:
Anything that cannot be explained must be attributed to a greater intelligence.
Re:Three rules safe. (Score:3, Funny)
I think you're misunderstanding the point of the Three Laws. I've been really, really tempted to build a very simple Three Laws compliant robot just to explain exactly what they mean.
Fundamentally, the basic question of any autonomous artificial intelligence is "how do I figure out what to do?" and to do that, it needs to have some guiding principles. The Three Laws certainly can be those guiding principles, and they are certainly not too general to plug in somewhere.
Let me give you an example of how to build a basic Three Laws robot. Build a robot, with two ultrasonic motion sensors - one pointing outward, the other pointing downward. The one pointing outward is a "human sensor" - if there's an object closer than, say, 1 meter, there's a human present. Now add two spring-released sliding plates - one in front of the "human sensor", and one right above the "robot", which is below the downward-facing sensor.
You now can represent the Three Laws as a flowchart, and the robot as a state machine, with a "World" (OBJECT_FALLING, HUMAN_PRESENT, TOP_PLATE_OUT, BOTTOM_PLATE_OUT) and a "Command" list, with flags (COMMAND_FIRSTLAW, COMMAND_HUMAN, COMMAND_NORMAL).
The decision tree, with Three Laws embedded, then becomes something like
UpdateWorldState();
if (CommandExists(COMMAND_FIRSTLAW))
{
ProcessNextCommand(COMMAND_FIRSTLAW)
return;
}
if (CommandExists(COMMAND_HUMAN))
{
ProcessNextCommand(COMMAND_HUMAN);
return;
}
if (CommandExists(COMMAND_NORMAL))
{
ProcessNextCommand(COMMAND_NORMAL);
return;
}
if (WorldState(OBJECT_FALLING & HUMAN_PRESENT & ~TOP_PLATE_OUT))
{
InsertCommand(TOP_PLATE, COMMAND_FIRSTLAW);
return;
}
if (HumanCommandPresent())
{
InsertCommand(TOP_PLATE, COMMAND_HUMAN);
return;
}
if (WorldState(OBJECT_FALLING & ~BOTTOM_PLATE_OUT)
{
InsertCommand(BOTTOM_PLATE, COMMAND_NORMAL);
return;
}
(where, presumedly, human commands are present until rescinded)
This robot would follow all of Asimov's Three Laws - it will stick the plate out to prevent a human's hand from gettin hurt if an object is falling to it, it will follow the human's orders, but it won't allow a human to get injured, and it will try to protect itself unless a human orders otherwise (it should also try to protect a human before protecting itself, but that state would never really happen in this setup).
You'll of course say this is an extremely simplistic setup, and it's not perfect. You're right. That's not the point - the point is that you can slowly, over time, build a Three Laws Safe robot over time, as the "World" state and "Command" lists become more complicated.
So, the answer to your question "how do you program 'don't hurt people'" is to have a robot evaluate the state of the world, estimate what the world state will be in the future, and determine if any of the Three Laws would be broken, and take action to prevent it. The action to "how do you recognize a human" is entirely dependent upon the sensory perceptions of the robot. So what if it mistakes something else for a human? It's just trying to fulfill its basic programming.
This should sound remarkably similar to another programm
Re:What's that saying? (Score:3, Funny)
OK then, you're my dumbass.
Re:urban myth (Score:1, Funny)
[losing karma] (Score:3, Funny)
Re:Obligatory bash.org reference (Score:3, Funny)