Evolution Robotics' ER1 Reviewed 94
Anonymous Coward writes "A useful review of Evolution Robotics' ER-1 by the boys at Techfocus. It covers: construction, customization, hardware requirements, best features, programming, durability of equipment -- and all that good stuff.
One interesting factoid is that the robot can recognize objects until the object is blocked - up to 40% - by something (like a piece of furniture). Techfocus aptly points out the Orwellian implications... Another thing that rocked my world is the notion that the robot is not as much of a drag on CPU as one might suspect. TF ran the unit on an NEC Versa VXi running Windows 2000, with a 900mhz CPU and 128mb of RAM, and encountered absolutely no problems. Encouragingly, if you want to further customize your robot, why not just write a script in C or Perl -- the manual even points users toward an app primarily based in Linux.
What's not surprising: it's pricey. Also some nice pictures of how the robot really looks right out of the box."
Re:Real world robots (Score:2, Interesting)
Robots will remain uncommon until much improved AI and sensory input is developed.
Re:I think I'll hold off... (Score:3, Interesting)
The others are modified to place that one in front.
Re:Real world robots (Score:3, Interesting)
In other words, this time saving marvel required more time spread out every day of the week than it took to mow it once per week on a single day. Then, you always had the concern about someone stealing it, breaking it, or running a toy over, if you were not there to watch it.
Now then, when they make the cold fusion model, with self defense laser turret, I'm there!
Another bit of the puzzle? (Score:3, Interesting)
I've been thinking for some time about awareness. After I read Creation: Life and How to Make it book by Steve Grand, I began thinking that perhaps awareness isn't the mystery it is sometime built up to be. What if we eventually discover that being "aware" doesn't require the phenomenal amount of computation that is often estimated? What if we discover that natures method of achieving it is actually highly inefficient (in terms of...physical complexity?) and easy to replicate using digital hardware?
At this point it is feasible to build a machine that can find you in a crowd (you, specifically, from among many others) talk to you, understand your commands and then travel where you tell it. This is already beyond the means of most animals.
If what I suspect is true, cognition is a relatively simple closed loop goal seeking (that seems to be a contradiction) parallel algorithm connected to a vast repository of highly lossy associative storage that ceaselessly works to achieve reproduction. Awareness is an emergent property of the process. You are a side effect, in the same way the useful work of a LISP function is often implemented as a side effect.
I'm not a professional AI researcher and it probably shows. I'll take it from someone who is. Martha Pollack, a professor at the Artificial Intelligence Laboratory at the University of Michigan and executive editor of the Journal of Artificial Intelligence Research recently said, "It's a crazy position to be in. As soon as we solve a problem, instead of looking at the solution as AI, we come to view it as just another computer system."
The significant progress made in AI to-date appears to be the result of reverse engineering nature until the core implementation of some basic function becomes clear. Just how many interconnected functions are necessary before you have a "who"?
Erector Set (Score:2, Interesting)
I always considered erector sets to be better for robotics experiments than legos, but I don't see erector set stuff much anymore. Did they go bankrupt? If so, why hasn't a Chinese toy firm resurrected the concept?
Maybe its the name