Game of Life with robots

[FLLCasts]

Hello everybody,

I’m Desislava Simova from the FLLCasts team (http://www.fllcasts.com). We have another great idea in mind and we want to ask you if you want to join our new scientific journey. It’s called Robotics Game of Life.

We want to organize a 14-sessions course (1 session weekly). The course is for students and classes of students. It is for Computer Science, Computer History, Robotics, Math.

We will develop a solution to the Conway Game of Life where the robots would actually play the game.

Students will construct the robots by instructions and lots of imagination. Initial programs will be provided by us and modified by students.

Would you join such a course? Please give us your answer. Your feedback is highly appreciated.

Different programming languages will be used for after class activities and solutions to problems. Videos of the solutions will be uploaded by students to FLLCasts.com and they will participate in our Global competition on Robotics Game Of Life. The best solutions will be awarded.

During the course you can ask live questions, post comments, listen to the recorded videos over and over again, experiment with the programs and get immediate and accurate feedback.

Topics

Computer Science, Programming Languages, Robotics, AI, Computer History, Computation & Turing Machines, Math

Level

Everybody from Beginner to Advanced.

You can find more information about the Robotics Game of Life in our publication in Tumblr - http://bit.ly/1QQBJGw

You can enrol in the course here - http://www.fllcasts.com/playlists/32

[dvgrn]

We want to organize a 14-sessions course (1 session weekly). The course is for students and classes of students. It is for Computer Science, Computer History, Robotics, Math.

We will develop a solution to the Conway Game of Life where the robots would actually play the game.

How many robots will you have available? You could certainly lay out a lot of them in a grid, and have each one just stay in place and raise or lower a flag (or send out a light signal, or whatever) to signify their state -- but that doesn't seem like what you're looking for.

There's no concept of a cell moving in Conway's Life, though, so it's hard to see how each robot could represent an ON cell. You might manage to get five robots to do a choreographed dance that looks like a glider moving, because there's no change in the number of robots needed, and each robot could move at most (1,1) per tick. But that's a long way from setting up an arbitrary Life pattern and having each robot do its part to simulate it correctly.

With even a moderate sized pattern, like an HWSS let's say --

Code: Select all

x = 5, y = 7, rule = B3/S23
2b3o$bo2bo$4bo$o3bo$o3bo$4bo$bobo!

-- where would the extra five robots come from, at T=1? Where would they hide again at T=2? If you look at the five isolated cells toward the lower right, the ones that die of loneliness... if those cells were represented by robots, how would they know where to move to next?

Maybe the idea is more that one or more robots will run around on a fixed grid, flipping over tiles or making similar changes according to Life rules? That could certainly be done, but it seems as if each robot would need to keep track of the entire board, rather than playing the part of a single cell and basing its activity only on what its immediate neighbors were doing.

If you did really want each robot to play the part of a single live cell, you might be better off picking a CA rule other than the Conway's Life rule -- maybe a reversible rule that conserves the number of ON cells?

[FLLCasts]

We assume that you can have any number of robots from 1 to N.
So that you can be in home alone or with a class of 5-6 robots. We do not expect more than 10-12 robots to join in a single game.

The actual game could a robot reading from a sheet of paper and another robot drawing on a sheet of paper.
Or even more interesting - the robot scanning a field of cells(cubes), calculating the new generations and sending it to a second robot. This second robot uses other cubes from which we have hundreds and places them on the field. And we could extend it to N robots scanning a generation here and N robots replicating a generation there.

Of course, not that complex generations, but the goal is to talk about computation, about programming, about robotics and communication and probably about AI.

What do you think about this?

[dvgrn]

Of course, not that complex generations, but the goal is to talk about computation, about programming, about robotics and communication and probably about AI.

What do you think about this?

I like the general idea of N robots with limited information, communicating in a way that allows them to complete a task that no one individual robot could perform.

The interesting thing about cellular automata is that all the cells are identical, and they can produce really impressively complex results even when they only know what their next-door neighbors are doing.

There must be a way to extend this idea to a project with robots, it's just a little tricky. How do you arrange it so that the robots are all equal participants, maybe operating on a set "clock" schedule" and following grid lines on a table, and they interact with each other in some interesting way to solve a problem?

Maybe some kind of CA particle simulation -- the Sand CA or various accretion models come to mind. Robots could carry blocks from a particle source and follow the rules to find out where to drop them, then go back for another trip. Could be done equally well with just one robot or with a larger number.

For robots simulating the B3/S23 Conway's Life rule, I haven't quite been able to imagine how to make it work well. Seems like it would be difficult to get multiple robots communicating effectively to do a Life simulation, unless you had enough to make a large grid and they all sat (boringly) in one place. One robot would be enough, of course, if it was just picking up and dropping blocks, but then you end up with an impressively inefficient Life simulator, something that's been done many thousands of times in the last 46 years on (almost) every conceivable platform.

I guess my thought is that it would be good if multiple robots could be set up to collaborate in some way that might produce unexpected patterns of behavior.

[FLLCasts]

Here is a short picture of the field that I am imagining.



http://media.fllcasts.com/assets/episod ... f_life.png

You can have one robot that first scan the field on the right and then constructs the field on the left.
You can also have two robots where one scans the field on the left and send information over bluetooth or wifi to a robot on the right that would construct it.

Or you can have NR reading robots on the left and NC constructing robots on the right where NR1 communicates with NC1. Each Reading robot is responsible for scanning exactly X number of rows and the corresponding Constructing robot is responsible for constructing the same number of rows. After we have constructed the generation on the right we then collect the boxes on the left and construct the new generation there. And this continues.

The speed depends on the size of the field that you want to have and the number of robots that you have, but generally the same algorithm is used for 4x4 and for 10x10 and NxN fields.

[FLLCasts]

Friends, we’re getting started with the Robotics Game of Life! Today we've received the cubes (our cells) for our game.

https://twitter.com/fllcasts/status/702442872482766848

We've also updated the description of our Robotics Game of Life playlist http://bit.ly/1PkGYOh

Can’t wait to see the result!

[FLLCasts]

Here is an update about our version of the Game of Life - with robots. Our robot successfully scans the Robotics Game of Life field and discovers the living cells - http://bit.ly/1QoZ5AZ

Also our we've released the Introduction episode - http://bit.ly/1RzXpHi

We hope you'll like it

[FLLCasts]

Here is the result of the magic touch of our almighty engineer, the master Yoda of the FLLCasts programming, the force that sometimes awakens - Dimitar! The Robotics Game of Life on the brick screen --> http://bit.ly/1QGePmx
What do you guys think

[FLLCasts]

We're glad to share with you the new Robotics Game of Life video - we have bigger field and more living cells.
--> bit.ly/253RyRT
We hope you'll enjoy it and share some of your thoughts.

[FLLCasts]

We have another great video for you https://www.youtube.com/watch?v=7GxdGr7vsQ4

The first brick says to the robot what is the field's size. The robot (the second brick) scans the field and sends the gathered information to the first brick via Bluetooth. And the first brick calculates the generations on an infinite field.

What do you think?

[FLLCasts]

So much fun! See what our robots do in our Game of Life in the next video https://www.youtube.com/watch?v=ZEPKMJvd0F0

The first robot – the Woodpecker – says to the second robot – the Beaver – what is the field's size. Beaver scans the field and sends the gathered information to the Woodpecker via Bluetooth. And the Woodpecker calculates the generations on an infinite field and shows us what would the next generation be. Then the Woodpecker arranges the cubes itself and we could see the next generation on the field.

[Scorbie]

Interesting to see manipulating Life in real life
I was puzzled at first and then realized that it was an infinite grid. My bad

Code: Select all

x = 6, y = 2, rule = b3/s23:T6,2
bo2bo$2b2o!

[FLLCasts]

Yes, exactly manipulating Life in real life

You should see our new video - whole room of Game of Life with robots. It's great! --> https://youtu.be/aeKWvdc4eII

[Gamedziner]

You might want to use a finite grid, though, in case someone gets the idea of making the robots simulate a Gosper glider gun:

Code: Select all

#N Gosper glider gun
#O Bill Gosper
#C A true period 30 glider gun.
#C The first known gun and the first known finite pattern with unbounded growth.
#C www.conwaylife.com/wiki/index.php?title=Gosper_glider_gun
x = 36, y = 9, rule = B3/S23
24bo11b$22bobo11b$12b2o6b2o12b2o$11bo3bo4b2o12b2o$2o8bo5bo3b2o14b$2o8b
o3bob2o4bobo11b$10bo5bo7bo11b$11bo3bo20b$12b2o!

(copied from http://www.conwaylife.com/patterns/gosperglidergun.rle)