skip to main content
Caltech

Mechanical and Civil Engineering Seminar

Thursday, April 7, 2022
11:00am to 12:00pm
Add to Cal
Gates-Thomas 135
Structural Computing: Leveraging Mechanics for Information Processing
Phil R. Buskohl, Research Mechanical Engineer, Functional Materials Division, U.S. Air Force Research Laboratory,

Mechanical and Civil Engineering Seminar Series

Title: Structural Computing: Leveraging Mechanics for Information Processing

Abstract: The collection and processing of information are essential for the successful operation of natural and synthetic systems. Adaptive materials that respond mechanically to their environment present an opportunity to consolidate this collection and processing pipeline by correlating the deformation states of the mechanical structure with abstract computing operations. Nonlinearity and multistability are important features to leverage in mechanical structures to perform this collection-processing consolidation, both for digital (mechanologic) and analog abstractions. To explore these concepts, we first harness the bistable behavior of the waterbomb origami unit constructed from a humidity-responsive polymer to abstract local bit states based on popped-up or popped-down configurations. The waterbomb units are structured such that their mechanical state and environmental inputs reproduce the truth table of simple logic gate building blocks. Networks of waterbomb origami units were further analyzed to estimate how neighboring stable state assignments modify the energetics of the individual waterbomb units, which could serve as a method to control the order of stable state switching in the array. We then shift toward analog computing concepts by investigating the computing capacity of 2D nonlinear spring networks using a reservoir computing approach. Reservoir computing is a class of recurrent neural networks that trains only a readout layer of the network dynamics in contrast to tuning all the internal parameters of the network. We introduce a mechanical analog for the rectified linear unit (ReLU) activation function from neural network community and benchmark the memory capacity, nonlinearity and output tasks of mechanical ReLU networks sampled from a distribution of spring properties. Preliminary results indicate that the stiffness ratio of the ReLU spring (ie. ratio of the bilinear slopes) is a key driver of the nonlinearity score of the network, even more so than the incidence of activating the spring nonlinearity. Collectively, the results highlight the potential to harness multistability and nonlinear dynamics as a source of physical computation to augment the collection and processing of information in adaptive mechanical systems.

Bio: Philip R. Buskohl is a Research Mechanical Engineer in the Functional Materials Division at the U.S. Air Force Research Laboratory. The Division delivers materials and processing solutions to revolutionize AF capabilities in Survivability, Directed Energy, Reconnaissance, Integrated Energy and Human Performance. Phil has authored over 36 peer-reviewed papers ranging from the chemical-mechanical feedback of self-oscillating gels, design of reconfigurable origami structures and mechanical computing concepts. His research interests include nonlinear elasticity, optimization methodology for material design, and mechanically adaptive materials.

NOTE: At this time, in-person Mechanical and Civil Engineering Lectures are open to all Caltech students/staff/faculty/visitors with a valid Caltech ID.

For more information, please contact Stacie Takase by phone at (626) 395-3389 or by email at [email protected] or visit https://www.mce.caltech.edu/seminars.