horizontal line
Living Machines Home navigation buttonseparator lineOverview navigation barseparator linePeople navigation buttonseparator lineProjects navigation buttonseparator linePublications navigation bar

Ideas for Living Machines

People: Una-May O'Reilly

Here is an annotated list about various ideas I've been considering under topics that I'm especially interested in:

Evolution and self adaptation in evolutionary systems

  • Let Evolution Change: If evolution were different, what would be the outcome?  How is an outcome shaped by its process of creation? What is the biological evidence of self-adaptation in evolution?
  • Reinforcement Learning and GP:  Can we use reinforcement learning to determine better parameters for the evolutionary process?
  • Tags and Mimicry:  Essentially tags are identification information that an agent in a complex adaptive system arries and outwardly displays to its environment and other agents.  What is their role in mimicry?

Development and differentiation in living systems

  • Evolving Solid Constructions:  How about designing a new growth process (with an efficient, expressive (possibly grammatical) definition like L-systems) that works in solid space (e.g. a solids growth process in SolidWorks)?

Self-organization, self-assembly

  • Self Assembly-Joints and Compositions:  The endless loop that grabs me in self-replication is the one that goes like this: Whatever it is that's building it, has to be built.  Or, to build it, you have to use it.  This leads me to think about the tasks of assembly and decomposition.   I want to think about how things can be designed so that they're easy to construct.  Maybe it's useful to ask the following paired questions:    
    1. What's simple to attach together?, and,
    2. What's useful as a robot 'attacher' component?
  • Computational Chemistry:  How can biochemical properties such as physical and chemical affinities of molecules serve as metaphors that aid us in directing an evolutionary programming system?

What properties might be necessary though not sufficient for a system to be living