The robot that can see! New AI that can inspect an unfamiliar object and pick it up

The robot that can see! New AI that can inspect an unfamiliar object and pick it up could lead to machines that can clean your house for you

  • The robots make distinctions between objects and accomplish specific tasks
  • They will one day  ‘see’ well enough to clean people’s homes, experts say
  • Robot looks at objects as collections of points that serve as ‘visual roadmaps’ 
  • This gives it an understanding of what it needs to grasp and where 
  • e-mail

View
comments

Scientists have created a robot that can ‘see’ unfamiliar objects using breakthrough computer vision. 

It can make basic distinctions between objects and accomplish specific tasks without having ever seen them before.

The AI will one day be sophisticated enough to create house-cleaning robots that can be put to work in homes and offices, researchers from MIT say. 

People could give their robot an image of a tidy house and ask it to clean while they are at work.

It could also be given an image of dishes and be asked to put the plates away, scientists say.

Scroll down for video


Scientists have created a robot that can ‘see’ unfamiliar objects using break-through computer vision. The AI will one day be sophisticated enough to create house-cleaning robots that can be put to work in homes and offices, researchers from MIT say

Humans have long been masters of dexterity, a skill that is largely be credited to the fact we can see.

Robots, meanwhile, are still catching up. 

For decades robots in controlled environments like assembly lines have been able to pick up the same object over and over again but now they are one step closer to actually seeing. 

Researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) have created a new system which they have dubbed ‘Dense Object Nets’ (DON).

  • Artificial intelligence poses a greater challenge to the… Volvo’s transforming robo-taxi: Firm unveils fully… New York scientists reveal new DNA analysis technique… Pluto SHOULD be a planet: Astronomers claim controversial…

Share this article

The robot looks at objects as collections of points that serve as ‘visual roadmaps’. 

This gives it an understanding of what it needs to grasp and where.

After training, if a person specifies a point on a object, the robot can take a photo of that object and identify and match points.

For example, someone might use DON to get a robot to grab onto a specific spot on an object such as the tongue of a shoe.


The new robots will be able to make basic distinctions between objects and accomplish specific tasks without having ever seen them before


People could give their robot an image of a tidy house and ask it to clean while they are at work. It could also be given an image of dishes and be asked to put the plates away, scientists say

From that it can look at a shoe it has never seen before and successfully grab its tongue.

This approach lets robots better understand and manipulate items, scientists say. 

Most importantly, it allows them to even pick up a specific object among a clutter of similar objects.

This will be a valuable skill for the kinds of machines that companies like Amazon and Walmart use in their warehouses.

‘Many approaches to manipulation can’t identify specific parts of an object across the many orientations that object may encounter,’ said PhD student Lucas Manuelli, who wrote a new paper about the system with lead author and fellow PhD student Pete Florence.

‘For example, existing algorithms would be unable to grasp a mug by its handle, especially if the mug could be in multiple orientations, like upright, or on its side.’ 

‘In factories robots often need complex part feeders to work reliably,’ said Mr Florence.

https://youtube.com/watch?v=OplLXzxxmdA%3Ffeature%3Doembed

Most importantly, it allows them to even pick up a specific object among a clutter of similar objects. This will be a valuable skill for the kinds of machines that companies like Amazon and Walmart use in their warehouses

‘But a system like this that can understand objects’ orientations could just take a picture and be able to grasp and adjust the object accordingly.’ 

In one set of tests on a soft caterpillar toy, a Kuka robotic arm powered by DON could grasp the toy’s right ear from a range of different configurations.

This showed that, among other things, the system has the ability to distinguish left from right on symmetrical objects.

When testing on a bin of different baseball hats, DON could pick out a specific target hat despite all of the hats having very similar designs — and having never seen pictures of the hats in training data before.

In the future, the team hopes to improve the system so it can perform specific tasks with a deeper understanding of the corresponding objects, like learning how to grasp an object and move it with the ultimate goal of say, cleaning a desk.

The team will present their paper on the system next month at the Conference on Robot Learning in Zürich, Switzerland.

COULD AUTONOMOUS ROBOTS PAINT OUR HOMES?

US engineers have developed an autonomous robot that could one day paint people’s walls. 

Scientists behind the robot, named Maverick, claim that it can paint walls evenly and with striking accuracy. 

MIST, or mobile intelligent spraying technologies, is the team behind the robot and is made up of engineering students from the University of Waterloo.  

Maverick is equipped with an array of sensors that make it a fully-functioning autonomous robot, MIST says.

The robot uses mapping technology and an elevator-like shaft to spray paint up and down the walls.  

It’s fitted with a platform, arm and spray system that allows for even coating. 

MIST believes that the home painter industry is ripe for disruption, noting that there are over 5,000 painting-related injuries each year, with $1.5 billion spent on painters annually in North America. 

‘The traditional manual painting process is slow, costly, inefficient and hazardous,’ MIST notes on its website. 

 

Source: Read Full Article