A team at Beihang University, working with scientists from Beijing Institute of Technology, built an AI chip shaped by how brains work. Instead of relying on standard computing methods, this device handles movement detection like living organisms do. Four times quicker than human sight can catch motion, the system sharpens response speed in machines. Because it mirrors brain pathways, lag drops when interpreting live visuals.
While most processors struggle under fast-changing scenes, this design keeps pace through biological mimicry. Faster reactions emerge not from power, but structure – closer to neurons talking in real time. So robots see shifts sooner; self-driving vehicles adjust earlier. Where old models delay, this one moves ahead without pause. Efficiency grows because timing aligns with nature’s own rhythm. Machines gain smoother awareness, almost lifelike, yet rooted in silicon.
Right now machines like driverless vehicles, flying drones, and robotic helpers rely on visual tech to make sense of their surroundings. Instead of guessing movement, older artificial sight methods grab complete images frame by frame. Each one gets broken down completely, pixel after pixel, searching for shifts in light. Because so much math happens per shot, delays build up fast. Half a second might pass before any response kicks in – sometimes more. When moving fast, a brief lag might add up to multiple meters of travel before reaction kicks in. By focusing on key shifts in visuals rather than analyzing every detail of a scene, the brain-like AI chip cuts down response time – motion sensing becomes swifter, uses less energy.
Scientists say their new computer chip takes ideas from part of the brain known as the LGN. While eyes take in scenes, only certain parts get full attention – like sudden motion or flashes. Instead of treating every pixel the same, the mind zooms in on shifts, such as something darting across view. Because of this filter, people can dodge a fast-approaching object without thinking twice. Much like that natural shortcut, the device highlights dynamic spots and skips flat stretches. Its design runs on noticing what changes, just as nerves do when spotting threats nearby. Where motion shows up, that is what gets focused on instead of pushing entire video streams to the central chip. By zeroing in there first, less power drains away while things move faster through the system.
What stands out about this breakthrough lies in how it mimics brain structure. Instead of traditional circuits, the device runs on components similar to nerve cells and connections between them. Built with transistor elements acting like synapses, the chip handles data while storing it at once. Because of this setup, small variations in light across moments become noticeable. When even a patch of pixels grows dimmer or brighter, that spot gets flagged as motion by the system. Faster processing happens when attention shifts just to moving parts of a scene. About four times quicker, say the team, than older AI methods when studying videos.
Out on the road, they put the mind-like computer chip to the test. During a self-driving trial, the car responded in 0.035 seconds instead of 0.23. What seems like a tiny shift makes a big difference. Traveling at 80 kph, every fraction of a second trims how far it rolls before stopping. Spotting hazards quicker leads to smarter moves around obstacles. Safety gets a quiet boost when split-second timing tightens up. Surprisingly, movement detection got much sharper – results jumped beyond twice what earlier models managed. That kind of leap suggests a new kind of chip, modeled on brains, might just reshape how driverless vehicles sense their world.
Flying machines gained an edge thanks to the new AI chip. When zipping through cluttered spaces, spotting barriers swiftly matters a lot. Thanks to circuits modeled on brains, these devices reacted faster to nearby objects. Nearly three-fourths of lag vanished during image analysis tasks. Decisions about where to steer now happen at sharp speed. When just certain areas get processed, the machine uses less electricity. Running on less juice leads to extended battery runtime – key for drones and mobile robots that work without a plug.
Something strange happened when robot arms got a new kind of computer chip – suddenly they moved smarter. Not just faster, but better at catching things zipping past without warning. One test saw them grab flying targets seven times more often than before. Even tricky items, say a bouncing ping-pong ball, didn’t escape their grip anymore. Machines began reacting almost like living hands, adjusting mid-movement without freezing up. Factories might start using these in places where every millisecond counts. Games involving robots could change too, maybe even how people team up with machines on assembly lines.
A new paper in Nature Communications comes from a team across multiple countries, with scientists from Tsinghua University working alongside peers at the University of Hong Kong, Cambridge, Northeastern, and King Abdullah University of Science and Technology. Smarter artificial intelligence might not come just from better code; these experts think hardware itself needs a rethink. Merging ideas from brain science into chip development could open different paths forward. While most focus lands on software tweaks, upgrading physical systems may bring speedier gains, using less energy along the way.
What makes the brain-inspired AI chip stand out? It helps robots understand people better. Because they read gestures and faces faster, their reactions feel less mechanical. For helpers in homes, clinics, or care settings, timing matters a lot. When someone moves, response has to be immediate safety depends on it. With quicker motion sensing and fewer lags, these chips let robots perceive actions as they happen.
Though things get tricky when lots of items shift rapidly in cluttered scenes, work already done hints at real promise. Moving forward, instead of staying locked in lab trials, efforts will shift toward building chip designs that grow easily for use in self-driving cars and factory machines. Success here might weave this tech into the backbone of future artificial intelligence systems.
One step at a time, AI keeps moving forward – speed now matters more than before. Road safety gets better when machines see faster, just like warehouses run smoother with quick-eyed robots. Drones find their way easier, factories hum along without delay. A new chip, built like a mind, shows life’s design holds tech answers. Human sight reacts in about 150 milliseconds; this chip hits nearly the same mark. Machines begin to sense the world, not just scan it.
Some think building better physical parts could shape where artificial intelligence goes next. Rather than just boosting raw speed, new chips focused on handling key information might run leaner. This model mimics thinking patterns, linking tiny circuits, nerve-like design, while using light-guided calculations to push performance without raising risks.
Down the road, efforts shift toward scaling up manufacturing while trying out the chip in more diverse settings. Out there in actual use inside self-driving vehicles and flying drones will show what it truly can do. Should it keep delivering solid results beyond controlled labs, a fresh benchmark might emerge for how machines see. Speed meets low power use plus sharper recognition, forming something rare: hardware shaped like brains but built for tomorrow’s smart machines.
All on its own, the mind-shaped computer chip pushes artificial intelligence hardware into new territory. Instead of copying standard models, it copies how eyes see movement, letting machines sense change quicker – four times speedier responses in bots, flying cameras, and driverless vehicles. Safety climbs up, power use drops down, everything runs smoother. While tech reshapes work everywhere, findings such as this show what happens when nerve science meets tiny circuit building – machines think sharper, move faster, hold steady without fail.
