, , , , , , , , , , , , , , , , , ,

“If you had to associate Artificial Life with one artifact, either software, hardware or wetware, which one would it be and why?

These and many other questions were address last week in Paris at the 20th anniversary edition of the European Artificial Life Conference, August 8-12, 2011.  The theme for the conference was “Back to the origins of Alife”.

Some of us, who where part of the inception, or the early days, of the Artificial Life movement, were asked to answer questions about the good old days as well as give our perspectives on where the field is currently going.  This occurred at the Thursday night “Pioneers Panel”, and I’ll now discuss some of the panel responses to the above question.

I personally enjoy the self-replicating and evolving cellular automata loops by Hiroki Sayama’s lab. I think these replicator loops capture the gist of Alife at a glance, while they also beautifully show an open issue we have with simple, artificial, evolutionary systems both in software (simulations), wetware (chemical) and hardware (robotics). Evolution in these simple systems tends to find simpler solutions (here simpler loops) over time. It is still an open research question how to “encourage” evolution (set up systems) to develop more sophistication over time.

Another of my Alife favorites is the moving oil droplet in water developed by Martin Hanczyc’s lab, which is a beautiful example of Alife wetware.  These dirt simple droplet systems demonstrate metabolic driven motility (chemical energy transformed into mechanical energy), path following as well as agency modification of the environment.

Several people pointed to the non-trivial evolutionary processes so vividly narrated by zoologist Tom Ray in his Tierra system in 1991. Unfortunately, I was not able to find good movie clips from these Tierra simulations, so I’ve instead picked a simulation clip from the related Avida system developed Chris Adami’s group. In this clip self-replicating programs compete for resources (space and computing-cycles) located in a two dimensional plane.  Different colors indicate different mutants and you can clearly see several evolutionary successions in this short clip where color (new mutant) after color (yet another mutant) takes over this 2D world.

The flying and flocking small robots developed by Dario Floriano’s lab were also emphasized: http://lis.epfl.ch/videos (click on the video “Deployment of large aerial swarms”) as well as the single flyers developed by the Delph University of Technology.

The flocking behavior generated by Floriano’s autonomous flying robots is based on the very simple algorithms initially developed in simulation by Craig Reynold back in 1986. He implemented the following algorithm for each object: (i) move in the direction with highest object density, (ii) don’t get to close to each other, (iii) and avoid obstacles. That’s it, and then you can generate the kind of flocking behavior you see everywhere in nature with birds, fish, sheep, etc.

I clearly remember when Craig showed us this original video of his bird-like “Boids” at the first Alife workshop in September 1987 at Los Alamos. It was awesome.

Karl Sims’ evolving virtual creatures were also mentioned as a hallmark for the power of design through evolution. Karl has evolved an array of impressive virtual creatures where their body design is selected based on the performance. This was developed in the mid 90’s and it’s still very impressive.

I have to emphasize that these above examples in no way constitute a comprehensive list of the best Alife science. It’s just some delightful examples cherry picked from the field’s first 25 years. There exist a massive body of beautiful work, which for a variety of reasons cannot be captured in a 30 seconds flashy video.  However, that does not make the less flashy work less important. Sometimes it’s actually the other way around.

I’ll soon post a few of my other impressions from ECAL 2011. It was an exciting meeting.