Department of Computer Science Drexel University
It has been noted that two types of systems exist in the world: 1) systems designed and built by humans to perform a specific job, and 2) biological systems that self-organize to also perform complex tasks without a master plan or overseer directing the actions. The self-organization found in biological systems is a powerful untapped paradigm that, if formulated and exploited, could lead to a superior approach to designing human-made structures. Self-organization algorithms could enable collections of software agents, micro-devices and engineered cells to form into user-defined shapes using only minimal, local information. Developing these algorithms should lead to advances in geometric computer-aided design, tissue engineering, swarm robotics, sensor networks and the custom-design of living cell aggregates. This research project combines and extends concepts from morphogenesis (the cell-level spatial process that produces the structure of an organism), cell aggregation simulation and evolutionary computing in order to produce a new, organic approach to shape generation. Called Morphogenetic Primitives (MPs), this approach captures the fundamental algorithmic constructs of geometric self-organization and harnesses their capabilities within a software framework. MPs are being applied to a number of innovative and important problem-domain areas. The approach creates a new class of self-organizing modeling primitives that may be programmed to ?grow? shapes that meet a specific need while adapting to constraints in the environment. The conceptual framework of MPs is rooted in developmental biology and is based on the multiple biological/physical mechanisms of morphogenesis. Local cell-level interactions aggregate to create high-level shapes and structures. Given the extreme challenges of determining the correct local actions that produce a desired final structure, evolutionary computing techniques are utilized in order to computationally produce the instructions and equations needed for a specific morphogenetic primitive application.
