Institution: Massachusetts Institute of Technology, University of Washington, Cornell University, University of Pennsylvania Proposal Number: 0735953
EFRI-ARESCI: Controlling the Autonomously Reconfiguring Factory
This project, with investigators from the Massachusetts Institute of Technology, Cornell University, the University of Pennsylvania, and the University of Washington, seeks to establish new fundamental theories for understanding autonomously reconfigurable systems under conditions of uncertainty. Natural systems possess the remarkable ability to create deterministic structures and processes out of a huge variety of raw materials. They have extreme robustness with respect to the source of raw materials and high adaptability with respect to their behaviors, due in part to their stochastic nature. These properties are also desirable for engineered systems such as automated factories, cooperative robotic systems, and networked computational systems. However, currently the design and assembly of these systems relies on deterministic processes and supply chains, which makes them fragile with respect to fluctuations in supply and limited in their ability for structural reconfiguration and functional adaptation. The goal of this project is to explore, and physically demonstrate, a novel paradigm for robust construction and adaptive reconfiguration of physical systems from elementary components, under uncertainty and variability of material resources. The investigators envision a manufacturing process where the source and target are defined indirectly, and the path between them is subject to stochastic fluctuations requiring strategic decisions. The project addresses (1) the theoretical foundations of reconfiguring systems by examining distributed algorithms, control theory, and statistical physics approaches to modeling system behavior; (2) methods for analysis and synthesis by analyzing the information flow in these systems and the development of a synthetic design methodology; and (3) experimental validation by using the investigator's existing and new platforms to demonstrate construction and swarming tasks.
The goal of the proposed system is to be built on-the-fly and instantiated at a disaster site to provide support by creating physical structures and facilitating information flow for first responders. The system can also be instantiated in the context of construction and fabrication, bringing manufacturing processes to new levels of customization and robustness and automation. This study can lead to a better understanding of biological systems, which are self-organizing at many different levels. Finally, the proposed approaches to engineering and analyzing stochastic adaptive reconfiguring machines may generate hypotheses for neuroscientists, psychologists and biologists regarding the organizational and algorithmic nature of adaptation and robustness in complex systems.
