Subject:
|
Re: Robot intelligence
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Tue, 16 Mar 1999 20:22:00 GMT
|
Original-From:
|
hancu <HANCU@stopspamSCF-FS.USC.EDU>
|
Viewed:
|
1007 times
|
| |
|
|
>
> It's all a matter of framing it in the correct reward/punishment context.
> E.g., if you want it to learn its environment, punish it for not running
> into things, reward it for actively avoiding obstacles (which opens the
> question: How're you mapping the environment, but...)
I just got my mindstorms kit a week ago, and I was thinking of doing this
sort of thing using a stack or some other data type in c++... Does the
rcx have enough memory to make this possible?
Dan
--
Did you check the web site first?: http://www.crynwr.com/lego-robotics
|
|
Message has 1 Reply:
 | | Re: Robot intelligence
|
| (...) If you use GCC to program your RCX, you have somewhere around 28K, which is enough memory for a lot of things. If you use anything based on the standard firmware, you have 32 registers and no way of doing indirect addressing. Things will be (...) (26 years ago, 16-Mar-99, to lugnet.robotics)
|
Message is in Reply To:
| | Re: Robot intelligence
|
| (...) It's all a matter of framing it in the correct reward/punishment context. E.g., if you want it to learn its environment, punish it for not running into things, reward it for actively avoiding obstacles (which opens the question: How're you (...) (26 years ago, 15-Mar-99, to lugnet.robotics)
|
5 Messages in This Thread:
 
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
