Click For Photo: https://i.dailymail.co.uk/1s/2019/01/17/21/8667222-0-image-a-40_1547760055616.jpgClick For Video: https://videos.dailymail.co.uk/video/mol/2019/01/12/4826379301719330467/1024x576_MP4_4826379301719330467.mp4
It gets knocked down, but it gets up again.
A team of Swiss researchers have trained a robot dog to be able to fight off repeated kicks from a human.
Matter - Times - Robot - 'ANYmal - Back
No matter how many times the robot, called the 'ANYmal,' gets kicked, it's able to roll over onto its back and stand back up again.
A team of Swiss researchers have trained a robot dog to be able to fight off repeated kicks from a human. No matter how many times the 'ANYmal' gets kicked, it stands back up again
Researchers - Findings - Paper - Wednesday - Science
The researchers' findings were laid out in a new paper published Wednesday in Science Robotics.
Scientists from Switzerland's ETH Zurich University trained the robot using artificial intelligence, which enables it to move and respond faster than ever before.
Speed - Record - Percent
So fast, they say, that it was able to beat the previous speed record by 25 percent.
According to researchers, this is not an easy thing to do.
Robots - Challenges - Robotics - Study - Notes
'Legged robots pose one of the greatest challenges in robotics,' the study notes.
'Dynamic and agile maneuvers of animals cannot be imitated by existing methods that are crafted by humans.'
Researchers - Simulations - Dynamics - Robots
Typically, researchers rely on simulations, which are 'cheap and safe,' but cannot 'accurately capture the dynamics of complex robots,' they explained.
Instead, scientists developed a neural network system from which the robot would be able to quickly learn and adapt.
Policies - Simulation - Machine - Locomotion - Skills
'Using policies trained in simulation, the quadrupedal machine achieves locomotion skills that go beyond what had been achieved with prior methods: ANYmal is capable of precisely and energy-efficiently following high-level body velocity commands, running faster than before, and recovering from falling even in complex configurations,' the study continues.
It also uses a relatively new method called reinforcement learning, where the robot basically learns through trial and error, completing a series...
Wake Up To Breaking News!