People involved: Jonathan Grizou
Keywords: shape engineering, automating science
A minimal mold could be defined as the minimal set of constraints that can bias a physical system into expressing desirable properties. In the case of shape engineering, we can harvest existing growth mechanisms of known organisms and aim at finding the minimal set of environmental constraints that can lead the system to adopt a desired physical shape. Such constraints can include physical constraints (walls, surface properties), chemical inhibitors and exhibitors, and conditions (temperature, humidity, light).
This project aims at developing a proof of concept of the minimal mold approach. Because the set of variables is likely to be large and the system of study not fully understood, the search of the minimal mold would be driven by a trial-and-error process using automation and machine learning. Two important challenges will have to be solved:
- The first challenge will be to identify a physical system well suited for the study (e.g. bacteria cellulose, chemical gardens, …). It will need to be easily observable (e.g. 2D via a webcam), have a relatively quick growth (e.g 1 day), and be automatable (easy to prepare, observe and clean).
- The second challenge will be to define the set of environmental variables that are both susceptible to bias the system’s growth and well suited for optimization via machine learning algorithms.
This project also aims to explore if we can devise machines that can help us generate specific engineering goals without the prerequisite to understand the inner working of a system. Hence, the desired outcome would be a compelling visual demonstration of the evolution of the shape-mold pairs created on a system whose inner workings are yet mostly unknown.