New Scientist


Robot that's the width of a hair masters Pac-Man and cuts cheese

New Scientist

You'd need your glasses on to play this version of Pac-Man. Tiny metal robots can plot their own route around a maze modelled on the iconic video game. Similar devices could one day be used to travel around the body, delivering drugs or performing surgery. Sarthak Misra from the University of Twente, the Netherlands, and colleagues created four different types of micro-gripper robots, with the smallest being just 100 micrometres long. The biggest was still less than one millimetre.


Food delivery robots are teaching themselves how to cross roads

New Scientist

That'll be the robot with my pizza. Such a scenario probably seems a bit far-fetched but, in the US and UK, delivery firms like JustEat and DoorDash are already experimenting using small robots to deliver groceries and meals. Currently these systems need human chaperones to monitor the robot's progress, jumping in if it gets into trouble. But now Kiwi, a company based at the University of California, Berkeley, is using machine learning to teach its delivery robots how to cross the road safely, without any human intervention. It could be an important step in making these robots more autonomous, something that is vital if they are ever going to be delivering our dinners at scale.


Robot's terrible jokes are a new test of machine intelligence

New Scientist

Pretend for a minute you're the captain of a ship that's being attacked by enemy cannons. Now – say something funny. Making up jokes on the spot is a real test of wits. Yet two comedians have developed an improv show in which many of the ad-libbed gags are delivered by a toy robot. In the last couple of years this unlikely comedy trio – known as HumanMachine – has performed 30 times to nearly 3000 people at comedy festivals in the UK and Canada.


Robofish floats about tracking antibiotics in the Great Lakes

New Scientist

A ROBOTIC fish may be an unlikely ally in the fight against antibiotic resistance. Swimming through streams and lakes, it will monitor the levels of antibiotics in the water, among other pollutants. The prototype will soon be sent below the surface of the lakes near Michigan, which are under threat from industrial pollution and contaminants from farming. "The water is getting increasingly contaminated with multiple pollutants," says Alicia Douglas at the Water Rising Institute. Antibiotics are among them, she says, and are becoming a growing risk because we know so little about how they spread.


Feather-light artificial muscles lift 1000 times own weight

New Scientist

Foldable artificial muscles can lift 1000 times their weight, be made in just 10 minutes and cost less than a dollar. Soft robots are getting better, but greater flexibility has a trade-off as softer materials are often weaker and less resilient than inflexible ones, limiting their use. But now researchers have created a 2.6-gram "muscle" that can lift a 3-kilogram object – a weight to strength ratio equivalent to a newborn lifting a Land Rover. The origami-inspired robot looks a bit like a bag with many chambers, each filled with air or water. When a vacuum is applied, the fluid is sucked out, contracting the muscle and driving the motion.


Who will really benefit from the coming smart-city revolution?

New Scientist

Have you ever felt like you are destined to become a supplicant whose chief purpose is to be sensed electronically, generate data and have it processed by intelligent machines for somebody else's benefit? If not, you are probably lucky enough to have been spared the hype of the smart-city lobby. That hype is hitting fever pitch. An investment group, backed by Microsoft founder Bill Gates, just pledged $80 million to kick-start Belmont, an 80,000-home smart city near Phoenix, Arizona, which will reportedly be replete with driverless vehicles, high-speed internet, jobs in advanced manufacturing and autonomous delivery services. In Canada, Google owner Alphabet has partnered with the city of Toronto to consult on and develop a section of Lake Ontario waterfront into a high-tech smart district.


Soft robot that squeezes your heart keeps damaged organs pumping

New Scientist

IT'S a pump that could bring your heart back to life. A lack of donated hearts often means people with heart failure die waiting for a replacement. But now a robotic device has been designed to help out with pumping duties to keep diseased hearts beating for longer. Nikolay Vasilyev at Boston Children's Hospital, one of the creators of the device, hopes it may even allow a full heart recovery, rendering a transplant unnecessary. The device consists of an implanted semi-circular brace that hugs the diseased chamber, surrounding it with an inflatable sleeve.


?utm_campaign=RSS%7CNSNS&utm_source=NSNS&utm_medium=RSS&utm_content=news&campaign_id=RSS%7CNSNS-news

New Scientist

Boston Dynamics has released footage of a humanoid robot that can do box jumps and back flips. This is a significant improvement on robot capabilities – despite what years of sci fi movies have suggested, it's actually really hard to make a bipedal robot walk, climb stairs and just generally keep its balance. That's why some of the entrants in recent humanoid robot Grand Challenges have looked less human and more like a cross between an orang-utan and a wolf spider. The new video suggests that for Boston Dynamics' Atlas robot, some of these problems have been solved. Last year, Atlas demonstrated that it can already go for a nice hike in the woods, and now it looks like it's ready for the gym too.


What's the best way to scare an elephant? Use an AI scarecrow

New Scientist

When you think of agricultural pests, elephants are probably near the bottom of the list. But they do an enormous amount of damage to nut and banana plantations precisely because they are too big, tough and smart to scare off once they start eating. Now, Australian researchers have developed an AI scarecrow that can do the job. It has been so successful that they are looking to adapt it to other smart pests – the long term goal is a scarecrow that understands the type of pest approaching and can tailor its scaring strategy. Scarecrow technology has a long history of ignominious failure, and not just for elephants – animals quickly learn to tune out a deterrent if it becomes apparent that there is no threat.


Why we should build AI that sometimes disobeys our commands

New Scientist

The future of human-AI interactions is set to get fraught. With the push to incorporate ethics into artificial intelligence systems, one basic idea must be recognised: we need to make machines that can say "no" to us. Not just in the sense of not responding to an unrecognised command, but also as the ability to recognise, in context, that an otherwise proper and usable directive from a human must be refused. That won't be easy to achieve and may be hard for some to swallow.