cjh (06/29/82)
In response to your message of Mon Jun 28 22:49:47 1982: In the latest issue of DISCOVER (general-audience science magazine put out by Time-Life Co.) there is a short take on a combination computer/manipulator that will solve a given cube disarray and then implement its solution. (I don't recall that it has sensors; I think you have to enter the contents of each face, then put it in the press with the white face-cubelet up, but it then does all the necessary turns itself.) So far it's no threat to the whiz kids; the manipulators can only shift faces on one axis, so it takes up time rotating the cube for the next face shift. The story said it took about 5 minutes to solve a typical mess. Of course, a much more sophisticated program could solve it in less practical time (but might take too long thinking) by determining the actual moves which scrambled the cube and reversing them. It has been hypothesized that the maximum scrambling takes no more than 12 moves, since that is suffi- cient to exchange most of the edge cubelets (UDFBLR, or any other sequence of all three opposing pairs, done twice), but I don't know whether anyone has gone further with this.
kolstad (07/01/82)
#R:csin:-15200:uiucdcs:9200001:000:482 uiucdcs!kolstad Jun 30 19:29:00 1982 The University of Illinois chapter of Tau Beta Pi did indeed construct "Robbie Rubik" (sic) (sick) -- a machine which had manipulators and computer power enough to solve a Rubik's cube. Its original LED's and sensors to determine initial edge placement were flaky, but the machine performed admirably all through engineering open house. It is really a miracle of modern technology -- took about 12 minutes to solve a scrambled cube. 0.8 seconds of that was CPU time for the Z80.