Tech & AI

Researchers have demonstrated that drones can detect nearby obstacles by listening to their own engine noise—much like biological echolocation. The technique could serve as a backup safety system alongside cameras and radar, reducing drone crash risks and enabling operations in environments where traditional sensors fail.EN

Researchers modeled a hybrid system where autonomous cranes handle loading while operators remotely drive timber-hauling vehicles, achieving cost savings roughly double previous approaches. The finding suggests forestry companies could deploy fewer machines with remote human oversight, reducing capital spending while maintaining output near current levels.EN

Researchers have demonstrated that specialized silicon carbide sensors can accurately monitor ammonia levels in power plant exhaust in real time. The technology enables automatic fine-tuning of pollution control systems, potentially reducing both nitrogen oxide and ammonia emissions simultaneously—a critical advance as regulators worldwide tighten air quality standards.EN

Researchers found that transit agencies can significantly reduce the number of buses needed by planning timetables, vehicle scheduling, and charging station placement simultaneously rather than sequentially. The discovery matters because it offers cities a concrete way to lower capital costs for electrifying bus fleets—a major hurdle in the shift away from diesel.EN

Researchers have developed timetabling rules that make train networks more resilient to disruptions while maintaining efficiency—a finding validated on Sweden's busiest rail corridor. The breakthrough, which incorporates practical insights from experienced planners, offers transit operators a policy tool to improve punctuality without major infrastructure investment.EN

Researchers have identified boron-related defects as a major culprit behind poor performance in silicon carbide semiconductors used in power electronics. The finding could help manufacturers design better chips for electric vehicles and renewable energy systems, where these high-temperature semiconductors are increasingly critical.EN

Researchers have developed a way to detect single muscle fiber contractions using ultrasound alone—eliminating the need for needle electrodes. The breakthrough could enable faster, less painful muscle assessment in clinical settings and sports medicine, opening new markets for portable diagnostic devices.EN

Researchers have developed an early warning system for when artificial intelligence models used in pathology labs start making mistakes due to data differences between hospitals. The finding matters because AI systems trained in one setting often fail unexpectedly when deployed elsewhere—a critical problem as hospitals adopt these tools at scale.EN

Researchers found a way to predict biological brain age from MRI scans using simplified 2D images instead of full 3D volumes, cutting computational costs dramatically. The method maintains reasonable accuracy and could accelerate processing of massive biobanks like the U.K.'s 29,000-subject dataset, unlocking faster breakthroughs in neurodegenerative disease research.EN

Researchers have cracked a long-standing problem in industrial robotics: accurately modeling how robots actually move so control systems can run them faster and more efficiently. The breakthrough matters because factories worldwide rely on robots for manufacturing, and better models could reduce energy consumption and improve safety without expensive hardware upgrades.EN

Researchers have developed a mathematical approach that allows multiple robotic arms to coordinate perfectly when carrying shared loads—solving the problem in the fastest possible time. For manufacturers relying on collaborative robots, this means faster production cycles and the ability to handle heavier, more complex tasks without adding more equipment.EN

Researchers have developed MEMMAL, a framework that combines machine learning with traditional mechanistic modeling to create more accurate predictions of how drugs affect cells. The hybrid approach could accelerate drug development and enable faster personalized medicine decisions by marrying AI's pattern-finding power with biology's explanatory rigor.EN

A Swedish study found that second-graders using artificial intelligence to practice basic arithmetic improved faster than peers using conventional teaching methods. The finding could reshape how schools deploy AI in classrooms and appeals to edtech companies seeking evidence that their tools outperform traditional instruction.EN

A new metatext examines the relationship between artistic creation and audience need, using an electronic music piece as a case study. For creators and cultural enterprises, the work raises practical questions about matching what you make with what markets actually want.EN

Scientists have developed a mathematical framework that lets AI systems understand complex sentence meanings 10x more efficiently. The breakthrough solves a decades-old computer science problem and could accelerate deployment of next-generation language models in enterprise applications, from document analysis to customer service automation.EN

Researchers have solved a fundamental problem in autonomous vehicle navigation: how robots can reliably map and locate themselves in environments where physical objects shift position. The breakthrough matters for mining operations, warehouses, and any setting requiring long-term robot autonomy—where static mapping assumptions fail in the real world.EN

Researchers have mapped how snow, ice, and cold temperatures derail high-speed rail schedules, identifying which track segments are most vulnerable to delays. The findings could help transit operators and infrastructure planners prepare better for climate volatility and improve customer reliability promises.EN

Photon-counting CT technology dramatically improves image clarity around metal implants like joint replacements and surgical plates, outperforming standard hospital scanners. The advance matters for hospitals weighing equipment upgrades and insurers assessing reimbursement strategies, as clearer imaging could reduce repeat scans and improve diagnostic accuracy in orthopedic and trauma cases.EN

Battery-powered trucks emit far less carbon over their lifetime than those running on renewable fuels, according to new research comparing energy sources for heavy transport. The finding reshapes how companies and governments should invest in decarbonizing freight—the sector responsible for 5% of global emissions—with major implications for fleet strategy and energy policy.EN

A team has fixed a long-standing gap in automated theorem proving by enabling specialized math solvers to handle floating-point arithmetic—the backbone of scientific computing. The upgrade significantly boosts success rates in formal verification, a critical tool for ensuring software reliability in safety-critical applications like aviation and autonomous systems.EN

Researchers have developed a way to strengthen plywood adhesives by adding micro-fibrillated cellulose to lignin-based formulations, reducing reliance on petroleum-derived chemicals. The finding could help plywood manufacturers lower costs and environmental impact while maintaining structural quality.EN

Researchers have developed a new prediction system that forecasts human movement patterns far longer and more accurately than existing methods, critical for autonomous vehicles and service robots operating around people. The breakthrough could significantly reduce collision risks and improve safety in shared human-robot spaces.EN

Researchers have developed a formal system for encoding burden-of-persuasion rules into AI reasoning engines, enabling machines to model legal arguments with multiple levels of proof requirements. The approach could automate legal decision-making in contract disputes, regulatory compliance, and courtroom support tools—reducing inconsistency in how different parties' claims are weighted.EN

Researchers reveal that data centers don't just process information—they fundamentally alter how time operates in the digital economy. The finding challenges conventional thinking about infrastructure and suggests policymakers must account for temporal dimensions when regulating the data economy.EN

Researchers have developed a neural network that lets robots understand their surroundings by analyzing video in real time, mimicking human vision. The breakthrough could reduce the need for human operators and enable autonomous systems to work safely alongside people in dynamic environments—a critical capability for factories, warehouses, and delivery operations.EN