Over the past 5 to 7 years, non-overlapping (NOV) clock phase generators have conventionally relied on downstream, dedicated auxiliary logic—such as cross-coupled SR latch delay blocks. While effective at nominal voltages, these additional stages create severe power overhead and fail under ultra-low-voltage (ULV) constraints due to reduced transistor gain and output swing. Recent advances attempted to integrate NOV phase generation directly within ring oscillators to improve energy efficiency. However, topologies relying on bootstrap architectures or heavy transistor stacking (e.g., laddered inverters) remain restricted to minimum operating voltages ranging from 0.5V to 1.0V. Present implementations lack the ability to directly produce non-overlapping clock phases at sub-100 mV supply levels without incurring high area and power costs from auxiliary phase-generation circuits.

Two closely related sonic devices measure airflow characteristics by detecting differential times of flight between pairs of sensors. Developed by UC Berkeley researchers, this technology encompasses a room anemometer for measuring three-dimensional air velocity vectors and air speed, alongside a duct flow anemometer designed to measure volumetric air flow. The architecture leverages arrays of microelectromechanical systems sensor chips, which provide lower production costs, higher measurement accuracy, faster response times, and significantly lower power requirements compared to conventional airflow sensing technologies. The devices can incorporate integrated radios for wireless data transmission and are highly optimized for efficiency, allowing them to operate for multiple years on a single small battery. Furthermore, onboard signal processing algorithms enable precise flow and temperature analysis across both homogeneous and inhomogeneous flow fields, while an integrated compass and accelerometers continuously correct the room anemometer for physical orientation changes.

Chemoselective conjugation is achieved through redox reactivity by reacting a nitrogen-transfer oxidant with a thioether substrate in an aqueous environment to form a stable conjugation product. Developed by UC Berkeley researchers, this platform utilizes Redox-Activated Chemical Tagging strategies for methionine-based protein functionalization. Specifically, novel urea-oxaziridine compounds serve as oxidant-mediated reagents for direct biomolecule functionalization, converting target methionine residues into their corresponding sulfimide conjugation products. This biocompatible reaction occurs efficiently under mild, aqueous conditions, offering a powerful tool for modifying complex proteins without disrupting their native structures or biological activities.

A four-dimensional-printed pneumatically actuated soft joint provides a high actuation range, controlled folding degree, high customizability, and streamlined assembly driven by air pressure. Developed by UC Berkeley researchers, this technology utilizes spherical bellow structures whose peaks and valleys are printed with flexible materials, which are then adhered to both sides of the joint to facilitate folding. The unique joint design, ease of assembly, and wide material palette for this pneumatically actuated material can be applied to a diverse array of robotic actuation mechanisms. For example, a versatile robotic gripper can be constructed to safely manipulate a wide variety of delicate, soft, or hard objects across multiple fields including medical, aerospace, and defense applications.

Computer-implemented methods identify putative nested open reading frames within prokaryotic deoxyribonucleic acid. Developed by UC Berkeley researchers, this computational platform accurately detects overlapping or entirely contained protein-coding sequences that traditional gene-finding algorithms frequently overlook. The method maps out alternative and nested open reading frames, providing a more comprehensive understanding of microbial genomes, hidden viral elements, and compact bacterial expression systems.

A system for training robots leverages monocular red-green-blue videos to streamline behavioral learning in humanoid systems. Developed by UC Berkeley researchers, this platform utilizes one or more processors configured to recover three-dimensional human motion geometry and scene geometry from standard video recordings over time. The system then retargets this recovered motion directly to a humanoid robot and trains a reinforcement learning policy. This process produces a unified policy that allows the robot to autonomously execute complex behaviors in various real-world contexts, bypassing the need for expensive motion-capture setups or tedious manual programming.

Researchers at the University of California, Davis have developed methods to treat medulloblastoma by restoring calcium signaling that promotes neuronal differentiation and inhibits tumor growth.

Researchers at the University of California, Davis have developed a closed-loop laundry system that integrates human-powered exercise with textile washing and water recycling, designed for resource-constrained environments including space habitats.