The Golem Project Automatic Design and Manufacture of Robotic Lifeforms Hod Lipson and Jordan B. Pollack

Next steps

The machines and tasks we described in this work are fairly simple from the perspective of what human teams of engineers can produce, and what biological evolution has produced. The hard question remains: will this process progress beyond toys to more complex machines? The graphs below show how fitness of creatures progresses over the generations. The horizontal axis counts generations, and the vertical axis measures fitness (locomotion distance, in this case). Each dot is an individual being born. It is evident that the progress rate becomes slower as the fitness becomes higher. This rate slowdown is enhanced by the fact that evaluation time also increases as machines become more complex. Also, looking at individuals that are generated after many generations, we see highly coupled mechnisms that are difficult to improve on.

There are several approaches to the question of sustained evolution. We are exploring these alternatives:

1. Very large populations and lots of CPU power, might still make progress. In particular, individuals with decoupled functionality might find it easier to adapt, thereby promoting modularity without explicit intervention.
2. It is possible that creatures reach some maximum level of complexity because the environment is too dull - in our case only an infinite plane. Perhaps with inclusion of more sophisticated environment, more building blocks like sensors, and interaction between robots, might promote more complexity.
3. The design space is exponential - as more components are added, the permutations of parameters becomes so large that the space of possibilities becomes intractable. Hoever, if we can find a way for the evolutionary process to discover and reuse modules, then the complexity of the building blocks increases exponentialy too, and the design space might be tractable after all.