Talk Abstracts

Mitigating industrial carbon emissions requires technologies that operate efficiently under realistic flow, thermal, and chemical constraints, where separation and reaction processes are often tightly coupled. Many existing approaches rely on powder-based materials and packed-bed reactors, which impose fundamental limitations on pressure drop, heat management, and scalability. In this talk, I will present recent work on monolithic material platforms designed to address these challenges by decoupling structural transport properties from surface chemistry.

The talk will focus on how mesoscale architecture in washcoated monolithic materials governs gas transport, thermal behavior, and reactive performance under integrated capture–reaction conditions. By systematically varying operating parameters such as flow rate, temperature, and feed composition, this work identifies structure–function relationships that emerge in structured reactor environments. These results highlight both the opportunities and constraints of monolithic architectures and illustrate how reactor-relevant material design can enable more efficient and scalable approaches to carbon management across engineering applications.

Robotic technology is evolving at a remarkable pace, and with this rapid evolution comes an urgent need for safe robotic autonomy. This talk presents the mathematical foundations for achieving collision avoidance on robotic systems, enabling robots to generate safe actions directly from real-world perception (e.g. camera, LiDAR). By bridging elliptic partial differential equations and modern control theory, we establish formal mathematical guarantees of safety while adapting changes in the environment. We validate these theoretical approaches on a variety of robotic platforms---including humanoids, quadrupeds, drones and manipulators---demonstrating the generality of the underlying mathematical principles, and their capacity to enable safe and reliable operation in real-world settings.

Ubiquitous amongst any living organism is the need to conserve energy via processes such as respiration. Respiration requires the reduction of an electron acceptor, such as oxygen, to generate an ion gradient driving the synthesis of adenosine triphosphate (ATP) via the action of the ATP synthase. In the absence of oxygen, the preferred terminal electron acceptor, non-fermenting and denitrifying bacteria such as Pseudomonas aeruginosa can respire using nitrate instead of oxygen, reducing it to dinitrogen. In this study, we focused on the utilization of a specific N-oxide intermediate: nitric oxide (NO). NO is of particular interest due to its production as a toxin as part of the human immune response to infections. 

This talk will provide an example of a common pathogen’s ability to use a substrate initially produced as an antimicrobial agent for energy conservation, highlighting the importance of considering the full repertoire of energy-conservation strategies used by pathogens to succeed in eradicating them. 

As AI-generated synthetic media becomes increasingly indistinguishable from authentic content, understanding how humans perceive and evaluate artificial stimuli is both a scientific and societal imperative. This talk presents findings from neuroimaging and behavioral experiments investigating whether the human brain contains reliable signals that distinguish real from AI-generated faces.

Using fMRI and machine learning-based neural decoding, we show that distinct patterns of brain activity emerge when participants view authentic versus synthetic faces — even when behavioral judgments approach chance. Multivariate classification of neural responses identifies regions involved in authenticity perception, revealing that the brain processes subtle cues related to artificiality that may not reach conscious awareness. Behavioral results further characterize the conditions under which human detection performance degrades, particularly as generative models improve in photorealism.

This study examines the effects of microbiome on activation of host regeneration processes. As an injury model, we utilized the Drosophila limb, which does not normally regenerate. We find that supplementing Lactobacillus brevis ATCC 367 promotes activation of regeneration processes in the amputated limb, as characterized by change in wound healing, enhanced tissue survival, and eventually partial regrowth of the limb. To investigate how L. brevis influences host regeneration, we combined genome-scale metabolic modeling with experimental measurements, and found that L. brevis secretes the amino acid ornithine. Flies fed with L. brevis indeed show higher extract levels of ornithine and altered levels of enzymes that metabolize ornithine. Directly administering ornithine to the flies promotes regeneration processes in the limb. In summary, this study presents evidence that supplying a bacteria that secretes ornithine can promote regeneration processes in a host organ that does not normally regenerate. Further, L. brevis supplementation can be performed prior to injury, opening up the possibility of prophylactically improving response to injury.