Mercurial > cortex

comparison org/joint.org @ 352:d9128eb5f42e

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
added some stuff to joint.org about flowers.
author Robert McIntyre <rlm@mit.edu>
date Tue, 05 Mar 2013 18:54:52 +0000
parents 0596613e5a41
children
comparison
equal deleted inserted replaced
351:0596613e5a41 352:d9128eb5f42e
1 * Summary of Senses
1 2
2 3 vision -- list of functions which must each be called with
3 * A joint 4 the world as their argument, each of which returns [topology data]. Each
4
5 The point of this joint is that it uses exploratory motor movements to
6 learn how to move to any particular position.
7
8 visual-information -- list of functions which must each be called with
9 the world the argument, each of which returns [topology data]. Each
10 element of data is a number between 0 and 255 representing the 5 element of data is a number between 0 and 255 representing the
11 intensity of the light recieved at that sensor. 6 intensity of the light recieved at that sensor. Each element of
7 topology is a pair of numbers [x, y] such that numbers whose pairs
8 have a short euclidean distance are generally physically close on the
9 actual sensor.
12 10
13 proprioception -- list of nullary functions, one for each joint, which 11 proprioception -- list of nullary functions, one for each joint, which
14 return [heding pitch roll]. 12 return [heding pitch roll].
15 13
16 movement -- list of functions, one for each muscle, which must be 14 movement -- list of functions, one for each muscle, which must be
18 in the muscle. Each function returns a float which is (current-force/ 16 in the muscle. Each function returns a float which is (current-force/
19 total-possible-force). 17 total-possible-force).
20 18
21 touch -- list of functions which must each be called with a Node 19 touch -- list of functions which must each be called with a Node
22 (normally the root node of the simulation) the argument, each of which 20 (normally the root node of the simulation) the argument, each of which
23 returns [topology data]. Each element of data is a posive float 21 returns [topology data]. Each element of data is [length limit] where
24 between 0.0 and the max length of the "hairs" of which the touch sense 22 limit is the length of that particular "hair" and length is the amount
25 is composed. 23 of the hair that has been activated so far. (= limit length) means that
24 nothing is touching the hair.
25
26
27 * A Flower
28
29 A flower is a basic creature that tries to maximize the amount of
30 light that it sees. It can have one or more eyes, with one eye being
31 "special" in that it is this eye which must recieve maximum light. It
32 can have multiple articulated joints and mulcles.
33
34 Want an algorithm that uses the sense data of =vision=
35 =proprioception=, and =movement= to maximum benefit in order to look
36 at the light source.
37
38 The light source will move from place to place and the flower will
39 have to follow it.
40
41 The algorithm should be generalize to any number of eyes and muscles,
42 and should become /more/ preformant the more sensory data is
43 available.
44
45 I will punt on working out an elegant model of motivation for the
46 flower which makes it want to go to the light.
47
48 Maybe I need a motivationless entity first, which just learns how its
49 own body works? But then, wouldn't that just be a motivation itself?
50
51
52
53
26 54
27 #+name: load-creature 55 #+name: load-creature
28 #+begin_src clojure 56 #+begin_src clojure
29 (in-ns 'cortex.joint) 57 (in-ns 'cortex.joint)
30 58