A new, AI-powered training method created by UC Berkeley researchers helps robots learn complex tasks with startling speed and accuracy. The protocol combines human demonstrations and feedback with the robot’s own real-world attempts to help it master tasks such as assembling a computer motherboard, building a shelf or even “Jenga whipping” with a 100% success rate within a matter of hours.

Researchers at Stanford Engineering have developed a new drug delivery platform that would allow antibodies and other protein therapeutics to be administered via injection, instead of through a time-consuming IV infusion.

World-renowned plant biologist Lucia Strader, previously at Duke University, will join the faculty at the Salk Institute in October 2025. Her research will help Salk’s Harnessing Plants Initiative design more resilient crop varieties that can thrive in changing environments.

The U.S. National Science Foundation has renewed funding for BioPACIFIC MIP — Biomaterials, Polymers and Advanced Constructs from Integrated Chemistry Materials Innovation Platform — a collaboration between UC Santa Barbara and UCLA that provides a unique scientific ecosystem for the United States. This $19.8 million renewal empowers BioPACIFIC MIP to continue advancing the frontier of biobased materials, uniting synthetic biology, chemistry, automation and artificial intelligence to reshape how materials are discovered, designed and deployed.

Since the era of film photography, capturing and documenting high-speed dynamic processes has been a longstanding pursuit in science and engineering. With the rise of solid-state imaging technologies such as charge-coupled devices (CCD) and complementary metal-oxide-semiconductor (CMOS) sensors, high-speed imaging has garnered increasing attention and become a cornerstone in fields such as aerospace, industrial inspection, and national defense.

Conventional imaging sensors primarily record two-dimensional (2D) image sequences that lack depth information. In recent years, the rapid progress in optoelectronic technologies has made 3D imaging and sensing one of the most dynamic and critical frontiers in optical metrology and information optics. Among these, Fringe Projection Profilometry (FPP) has emerged as a leading non-contact 3D surface measurement technique due to its high accuracy, flexible encoding, and wide applicability. Nevertheless, conventional structured-light 3D imaging techniques such as FPP always rely on a one-to-one synchronization between pattern projection and image acquisition, fundamentally limiting the system’s temporal resolution to the native frame rate of the imaging sensor. Current methods to increase speed often depend on high-refresh-rate hardware, which significantly increases system complexity and cost. This hardware bottleneck has become a major obstacle in advancing high-speed and ultra-high-speed 3D imaging.

Worcester Polytechnic Institute researcher Catherine Whittington has been awarded a prestigious CAREER Award from the National Science Foundation to develop three distinct laboratory models for the study of fibrosis in pancreas, skin, and uterine fibroids.