Modeling Physarum Polycephalum Growth With Neural Networks on Bioceramic Porcelain Surfaces
Here we describe a system for the exploration of the interface between chaos and control through the manipulation of physarum polycephalum (slime mold) on ceramic topologies.
Slime mold has gained interest across an array of fields due to its ability to solve shortest walk scenarios on varied topologies. At the beginning of its life-cycle, the species rapidly coats a surface in order to identify food sources. Once these nutrients nodes have been located, the species retract excess plasmodium, leaving thickened channels that connect each node.
These channels follow the shortest-walk path between food sources on the topology with striking accuracy. By manipulating the topology of ceramic tiles, the shortest walk between food sources, and therefore the channels formed by the slime mold, can be manipulated.
At exaggerated topologies with extremely high peaks, the shortest walk to a food source is more easily calculated given the severity of the gradient on steep edges. In these scenarios, the slime mold extends its channels in a predictable manner. At more subtle topologies, however, a subset of paths are accessible to the slime mold.
Factors beyond shortest-walk calculations influence its decision-making process, leading to the deliberate introduction of chaos into the system. This system allows for the novel exploration of the interface between chaos and control in the design process.
Team Diana Yan | Nicole Bakker | Nathan Peters | Nic Hogan
Coursework SCI-6317 Material Systems: Digital Design and Fabrication (2018)
Instructors: Co-taught by Prof. Martin Bechthold and Jose Luis García del Castillo y López, Harvard Graduate School of Design