This is the first in what will, I hope, be an ongoing series that picks the best from among stories that cross the juggling computer network that was described in the Winter 1990-91 issue.
What sort of topics do people discuss on the juggling discussion group? Here is a sampling from the past few months...
The electronic discussion can be a way for jugglers to get feedback that's unavailable to them otherwise. For example, Duane Starcher posts his messages from St John's, Newfoundland. He has sought (and given!) tips on numbers juggling which otherwise would have had to suffer a long delay through the Juggler's World publishing cycle, or until he got to meet others at the St. Louis festival.
There was a lot of talk about the festival. Some tried to link up to share rides to St. Louis, and to organize pre-festival get-togethers on the way there. Network jugglers received an early edition of the workshop schedule from Mike Vondruska. In addition, those unable to attend were treated to early postings about what happened during the festival.
One of the most active discussions prior to the festival was how to hook up with other "electronic" jugglers in St. Louis. People tossed around different ideas for designs to go onto a T-shirt or a button or sticker that would instantly identify members of the discussion list to each other. As with many such discussions, it came to a close when one ambitious juggler, in this case Lindsay Bagnall, stepped up and volunteered to produce buttons and bring them to St. Louis. Though she brought twice the number that people requested over the network, the supply ran dry and she will have to fill orders through the mail.
Because the members of the discussion groups are more inclined to be in one field or another related to computers, the messages often concern juggling and computers. Allen Knutson posted a program he wrote for a personal computer that can display juggling patterns known as "site swaps," which was described in the Summer issue of Juggler's World. Others soon adapted this program to run on other types of computers, and before long computers were tossing site swap patterns all over the network.
Knutson also posted an intriguing note about a lecture he attended about a pair of juggling robots. Prof. Dan Koditschek and his group at Yale University have built two juggling robots [1, 2, 3]. In a talk he gave at Caltech, Koditschek described their work and showed a two-part videotape. Here's what Allen had to tell us:
"The first part has a board at 30 degrees or so from the vertical with sensors to locate a puck to within a square inch or so (the puck sends out a signal and is battery powered). Since the puck never leaves the board, this has already removed a degree of freedom. Attached to the board is a lever, which can pivot around one point. (According to the way these roboticists appear to count, the arm rotating provides one degree of freedom, and the fact that it's one-dimensional, not a point, provides another, so the numbers do add up right.)
"They demonstrate the arm moving to tap the puck, and keep it bouncing to a certain height from a certain point on the arm indefinitely, much like bouncing a tennis ball on a racket. The main emphasis is on robustness, so a human comes in and disturbs things occasionally to keep it interesting.
"Then the puck is bounced from one side of the pivot to the other, back and forth. Then two pucks are put in, one on each side. Imagine a tennis racket with two heads, trying to bounce two tennis balls asynchronously. The robot is very good. The human comes in to annoy it, and the robot makes an amazing save, better than a human. Like I said, the main emphasis is on robustness.
"The second part is in three dimensions. The robot now has a shoulder, waist, and wrist, and a flat board to hit ping pong balls with. Everything is painted black, except for the ping pong ball, as the researchers are not interested in the vision problems. Two TV cameras watch the ping pong ball, triangulating its position. The first demonstration is the robot just holding the ball, not that easy a thing on a smooth, flat board. The second is it bouncing it to a certain place, every time.
"What's most interesting about the theory they used, besides that they don't fully understand it yet, is no planning was involved (if I understand correctly). The robot was given instructions that were completely based on what the computer believed the current position and velocity to be, with nothing set up for the future. Three of us net jugglers (Bruce Tiemann, Jack Boyce, and myself) went to a talk given by this guy."
I contacted Allen about using this note for this column, and he gave me a
pointer (that is, the electronic mailing address) to Prof. Koditschek. I sent
him a message asking whether there were any problems with my printing this
note, and he wrote back with an offer to mail me a set of reprints of papers
his group have published. So now I have a nice stack of papers describing
their work. Moreover, I can forward him a copy of this column before it is
published. Perhaps this work on the network will lead to a feature story on
these juggling robots in some future Jugglers' World issue!