iMMEC Seminar with Isabella Thiel and Ronald Heisser
bis
Building 3, 370, Middlesex
Speakers: Isabella Thiel and Ronald Heisser Affiliations: Isabella Thiel, PhD Candidate, MIT Department of Mechanical Engineering Ronald Heisser, Postdoctoral Associate, MIT Department of Mechanical Engineering Hosts: Andrew Akerson, Professor, MIT Department of Mechanical Engineering Mohammad Charara, Postdoctoral Fellow, MIT Department of Mechanical Engineering Lebo Molefe, Postdoctoral Fellow, MIT Department of Mechanical Engineering SPEAKER: ISABELLA THIEL Talk title: What Can Dynamics Tell Us About Extremes We Rarely Observe? Abstract: Extreme events in chaotic dynamical systems are difficult to model reliably, particularly when high-fidelity data are scarce and the events of interest occur on timescales far exceeding available training records. Yet the dynamics themselves contain information about transient amplification that is not readily captured by conventional data-driven approaches. We developed a method to access finite-time instability geometry to improve the statistical emulation of rare extremes. By evolving a small stochastic ensemble around synchronized coarse-resolution dynamics, we obtain a computationally inexpensive probe of local perturbation growth. We establish a theoretical connection between ensemble covariance and the deformation kernels of the variational dynamics, providing a Jacobian-free characterization of finite-time instability. This information is then incorporated into an existing and otherwise unchanged neural corrector without modifying the underlying physical solver. Using quasi-geostrophic flow, we demonstrate improved estimates of extreme-event frequency and spatial extent with substantially less high-resolution training data, extrapolating trajectories 1,000 times longer than the training window. Biography: Isabella Thiel is third-year Ph.D. candidate in the Department of Mechanical Engineering at MIT advised by Themistoklis Sapsis. Before MIT, she received a M.Sc. in Applied Mathematics from ETH Zürich in 2024, and a B.Sc. in Mathematics from TUM 2021. Her research lies at the intersection of dynamical systems, stochastic modeling, and rare and extreme events. She develops analytical, computational, and data-driven methods for high-dimensional nonlinear systems with persistent or intermittent instabilities, with a particular interest in inexpensive predictive tools for anticipating rare transitions and extreme behavior. SPEAKER: RONALD HEISSER Talk title: Biohybrid Toggle Switches: Leveraging Bistability in Muscle-Actuated Machines Abstract: Biohybrid mechanisms actuated by muscle tissue can revolutionize robotics by harnessing the adaptive, energy-efficient force generation of living muscle cells. Muscle’s hierarchical, modular structure enables scalable actuation from the μm- to m-scale, making it compelling for millimeter-scale devices, where conventional robotic actuators fail. Electrically stimulated skeletal muscle efficiently transitions between passive resting and active contracted states. Current biohybrid robots couple muscles to simple spring-like mechanisms, converting rhythmic stimulation and actuation cycles into walking, swimming, and pumping. In these systems, maintaining actuated states requires sustained power, accelerating fatigue and cell damage. We extend biohybrid functionality by coupling engineered muscles to bistable mechanisms, enabling switching between stable deformation states, consuming power only during switching. These “biohybrid toggles” reduce muscle power consumption and stimulation exposure while opening opportunities for musclepowered mechanical computing. Using elastomer micromolding, we produced a tunable bistable flexure mechanism with muscle-compatible stiffness and displacement ranges. We couple muscles to the flexure and demonstrate, to our knowledge, the first biohybrid bistable mechanism. Biography: Ronald Heisser’s research focuses on developing design principles and manufacturing techniques to realize soft, millimeter-scale electromecha