I guess I haven’t thought of having a robot that “learns”. Really all I’ve thought of is an adapting code based on boolean values of success (accomplished goal/didn’t accomplish goal). I don’t have any idea how humanizing a robot would be possible. Theirs so many factors in making a robot that can observe its environment for no reason. I’m not even sure if a bot could be made to function without any kind of goal.(maybe it would just lay on the floor uncontrolably jittering)
Oh and Defwheezer I am sorry we have REALLY gotten off topic. Your rover just looked so alive it got me thinking about AIs.
I have absolutely know idea what your signature means. I’d like to find out though.
Copy the human brain, I didn’t mean laterally, I meant conceptually, and as far as intelligence goes, that’s a very controversial topic because its more or less based on personal opinions and ideas. Depending on the application, the term intelligence is going to vary. My programming skills don’t go much beyond IF…THEN…ELSE…, so I really can’t contribute to the topic beyond my personal opinions.
I think we are in the baby stages with A.I. and robots, but I think the day will come when the two will merge together and provide real practical applications while also opening a variety of new problems. When that day comes, I think it will be like the age of the PCs, it will be the age of PRs (Personal Robots).
Hard to say how far we can get without some breakthroughs. Behaviors can get us started. A robot that can simply scan the floor and pickup and move small toys into a toy box might be a boon for young parents. Some simple tasks might be programmed, such as taking out trash on a regular basis. Lawn care shouldn’t be too hard, vacuuming either, for that matter. These don’t require “intelligence” per se.
When does intelligence come in? Above the general household chores level perhaps. But if a robot can understand simple commands and carry them out, it is perhaps starting to demonstrate limited intelligence. The more parameters or variables in the command and task, the more the robot must modify the task, then perhaps the more intelligence is being demonstrated.
When does a robot “think”? As mentioned before simple behaviors can be construed as intelligence or thinking; the line between the two is a bit gray.
As we develop behaviors, we’ll see emergent behaviors that we hadn’t planned on. At what point does Jonny 5 start to rewire himself? The future should be interesting!
Alan KM6VV
P.S. You can program behaviors with if-then-else statements!
I recently got the latest floor vacuuming robot. It was advertised as being able to
“learn” from its environment and apply the “learning” to improve its performance.
The first time I used it to vacuum the floor, it got stuck on a towel that had fallen on the floor (the towel got caught up in the brushes and stalled the robots motors). I was disappointed because that is exactly what my old Rooba was notorious for- getting stuck on small toys, a piece of twist tie, or a towel or rag left on the floor. I had to spend a lot of time before turning on the Roomba going around and picking up everything on the floor that Roomba might get snagged on.
Well, I decided to give my new “learning†robot (edit: this one is not an iRobot model, but a prototype based on work by Hod Lipson’s group at Cornell) another chance, and left a towel on the floor on purpose to see if the robot had “learned†from its first encounter with this obstacle. Much to my delight, the little bugger had indeed seemed to learn to avoid the towel- if fact, it actually used a little “beak†I had not previously noticed to push the towel into a corner where it would not be a nuisance! Needless to say I was mightily impressed. For several weeks now the robot has been cleaning the floor with increasingly improved efficiency!!
But yesterday I came home from work and the robot had not cleaned the floor at all. I thought it was a malfunction of some sort- a dead battery or fried board. Before sending it back for repair, I charged the battery up and did a “hard†reset using the button on the bottom of the robot (which, when pressed for 5 seconds, clears all memory and resets everything to the original defaults). I was a bit worried because when I pressed the reset button, the robot made a pathetic sound like a small trapped animal. The manual was no help at all in deciphering the meaning of this sound- it only said somewhat obliquely that the robot will “make a variety of sounds depending on its current stateâ€. Well, the next day the robot working great again. But after a few more days it failed to clean AGAIN. I tried to do another hard reset, but this time the button on the bottom appeared to be broken because no matter how long I held it down, the robot failed to reset and in fact, made a new sound- “nee ner nee ner nee nerâ€. Was this what the manual referred to as “emergent behaviorâ€??
LOL! now that’s funny. Perhaps you can set up a hidden camera and find out what the bot is doing while you are at work. Perhaps it’s having a bot party with the neighborhood bots shooting up negative ions instead of cleaning the floor.
I don’t know if that could be an emergent behavior. It seems a Roomba wouldn’t be complicated enough to start having unexpected things happen to it.(unless if something simply breaks) If it really did solve that towel problem then maybe it is “learning” though. Maybe it even became self-aware and decided it didn’t want to be your slave anymore.
