2025 - 2026 Roberts Innovation Fund Awards

Advanced technologies in aerospace, energy, and electronics increasingly demand materials that maintain performance under extreme conditions. Hu’s technology leverages a scalable ultrahigh-temperature stable plasma (USP) process to introduce a new paradigm for materials synthesis for designing metastable, multielement, and compositionally complex particles across metals, ceramics, and semiconductors. 

Petroleum refining is one of the most energy- and carbon-intensive industries, consuming over 1,100 terawatt-hours annually. The Zhong lab’s technology aims to reduce carbon emissions and energy consumption with their design of intrinsically porous polymer membranes for petroleum fractionation. Operating under mild, pressure-driven conditions, their membrane platform enables precise and rapid molecular separations of crude oil engineered through polymeric structures, with over 90% energy reduction compared to conventional distillation system. 

ExpoSeq addresses a critical gap in precision medicine by providing comprehensive, personalised data on what individuals are exposed to in their daily environments and translating those findings into accessible, personalised health reporting. The Fresh Air wristband passively captures over 1,000 airborne contaminants, including organic chemicals, microplastics, and respiratory pathogens. ExpoSeq integrates these wearable exposomic measures with biological markers to generate a personalised exposure profile, equipping individuals, clinicians, and researchers with the environmental data needed to understand and ultimately reduce environment-related health risks. 

The Miller-Jensen Lab and Kueh Lab's collaborative technology represents a potential breakthrough in the treatment of chronic degenerative diseases including Alzheimer’s, through enabling advances in engineering macrophage-based therapies.  The team is developing a novel genetic screening approach to identify targets to overcome obstacles that currently limit macrophage-based therapies, and will use these targets to optimize macrophage functionality for future disease-specific applications. This project holds the promise of unlocking the potential of macrophages as a next-generation cell therapy. 

Unlike most current AI scientists operating as data analysis assistants, Geomancer proposes an AI scientist that infers a simulatable model of the underlying system from which it generates hypotheses, designs experiments, and uses feedback to update the model. Geomancer builds on the Krishnaswamy Lab's vast prior work in deep dynamics inference to use multimodal data to automatically learn virtual neural generative models of the data, allowing for live feedback and dynamics inference for pharma applications and research institutions. 

Geometric Amplification will dramatically improve amplifiers intended to avoid the loss or corruption of data over long distances to support Internet infrastructure. With their invention of a new type of amplifier, the Harris Lab’s Geometric Amplification will advance telecommunication hardware with greater power-efficiency, phase insensitivity, and greater tunability. 

GlycoDx is developing the first clinically deployable platform that harnesses the diagnostic power of distinct cell-surface sugar molecules with an AI-empowered spatial Glyco-Code for early disease detection and therapeutic stratification. Unlike genomic or proteomic markers, glyco-codes capture very early disease transitions and therapy-relevant immune and tumor states. This uniquely multiplexed glyco-profiling and AI-driven analytics approach represents a fundamentally new dimension in molecular diagnostics. 

For the 52 million people living with epilepsy—of which 35% of patients are drug-resistant—NeuroLeaf seeks to leverage cutting-edge flexible bioelectronic technology to detect and treat epilepsy in real-time. Compared to current rigid, battery-powered solutions, the Cea lab solution works to deliver drugs locally via ionic communication and a multimodal interface with goals to further optimize the device’s circuit design and system integration. 

SilhouetteDB is a fault-tolerant, distributed, and oblivious coordination layer for privacy-preserving algorithms.  Built for the next generation of secure distributed systems, this project represents the future of oblivious coordination.  This solution will fill a current need in the market for enterprises in finance, healthcare, government, and social platforms which rely on graph analytics, but can’t safely use or share them. 

Six-axis force-torque (FT) sensors are commonly used by robots to measure contact forces, but traditional sensors are expensive, fragile, and prone to environmental interference. VisualFT overcomes these limitations by leveraging camera-based sensing and a novel mechanical amplification structure. This approach eliminates the need for complex electronics and high-precision manufacturing, resulting in a robust and affordable sensor.