Engineering advances and applied technology.
Aggressive driving was associated with more than 2.5 times the model-predicted battery capacity fade of eco-driving in a study using real-world electric-vehicle data. The result comes from a physics-based causal-inference framework, not a direct long-term measurement of battery loss.
Neural networks running on low-power microcontrollers decoded surface muscle signals in real time, a requirement for wearable devices and prosthetic control. On a large-scale benchmark, the models matched or exceeded earlier accuracy with up to 5,000 times lower computational cost in resource-constrained settings.
A single ionic wearable sensor simultaneously captured the heart’s electrical signals and arterial pulse waves in a wristband. The design could support beat-to-beat blood-pressure monitoring without timing errors between separate sensors, although the reported testing was limited to daily resting conditions.
A manganese dioxide cathode containing isolated, high-spin cobalt atoms helped an aqueous zinc–manganese battery deliver 658 mAh per gram and run for more than 15,000 cycles. The researchers say the cobalt enabled reversible three-electron storage and helped convert deposited manganese compounds back into active material without added acid or a redox mediator.
A two-molecule layered coating helped organic solar cells maintain higher efficiency when scaled into a mini-module. The module reached 17.0% efficiency, compared with 12.2% for a single-molecule control.
Chitosan scaffolds carrying two bone-related peptides improved bone-cell adhesion, growth and mineralization in laboratory tests. In vivo, scaffolds with both peptides supported greater cell colonization and reduced the thickness of fibrous tissue around the implant material.
Two different scandium treatments improved the cycling life of a layered sodium-battery cathode. After 100 cycles in sodium half-cells, capacity retention rose from 18.6% without the treatments to 67.8% with scandium inside the material and 75.4% with a scandium oxide coating.
An engineered thermophilic fungus converted raw corncobs into malic acid without a separate biomass-pretreatment step. In a 5-liter bioreactor, it produced 157.6 grams per liter of malic acid, which reached 99% purity after downstream processing.
Researchers demonstrated a programmable photomask that changed pixel patterns during exposures in a standard semiconductor photolithography stepper. The device produced photoresist features as small as about 1.7 micrometres, providing a basis for masks that could be reprogrammed without replacing the whole reticle.
An AI system using roadside cameras mapped traffic-related emissions by road segment and hour in Atlanta’s urban core. The researchers found that conventional inventories underestimated localized pollution burdens, partly because they missed the impact of inter-city freight traffic.
Scientists describe a global method for GPS-like navigation that avoids relying on a central processing hub. Using experiments with hundreds of widely distributed IGS stations, the approach produced satellite correction products similar to centralized baselines, while converging faster and cutting communication compared with other decentralized designs.
China began phasing out electric-vehicle (EV) purchase subsidies in 2018–2019, and the study finds buyers’ decisions shifted toward charging access. In cities with weaker charging networks, adding charging facilities raised EV market share in simulations, while the best subsidy type depended on local infrastructure.
Corrosion-causing chloride and dissolved iron can build up in amine-based gas-sweetening systems, raising maintenance costs. A biodegradable composite made from zein nanoparticles and humic substances removed both impurities from a methyldiethanolamine solution and reduced carbon-steel corrosion in experiments.
A new field study measured drivers’ pupil size as vehicles moved through four road tunnels with different shapes and lighting. Visibility emerged as the strongest driver of pupil changes, and the researchers proposed a “visibility-driven compensation” model to match the non-linear link between visibility and pupil response.
A study in Scientific Reports describes an AI framework that generates different residential landscape layouts while explicitly weighing trade-offs among comfort, appearance, accessibility, and ecological value. Tested on three parcels in Changsha, China, the approach produced more diverse, realistic-looking morphologies and found optimized schemes that, on average, scored 6.2% above professional human benchmarks on a combined index.
