“

Miroslav Trinko Geek, specialty programmer, but journalist by profession. Rider, tennis player and fan of Formula-1. I write about technologies, smartphones and electric vehicles.
The uniqueness of this development is to use a bioin -aged controller Tumblernetworking on a deep training with reinforcement (Deep Reinforcement Learning). It allows the work to simulate the human course and adapt to different conditions – even on sand, loose stones or mats.
Tumblernet combines information about the mass center and the pressure center into a single system that mimics human balancing. This can make the robot freely move from four -legged to two -to -two mode, using front or rear paws, as well as performing complex movements, including reversals and motion in a circle.
Another feature of it is stability. It can withstand strokes, pushing and even falling – and does not require an additional balance of balance. In case of loss of balance, robots can rise on its own, even if it caught on the objects.
Researchers also tested work on the beach – an environment where it is difficult to keep balance. Despite the difficult conditions, it successfully moved on two paws, which indicates a high adaptability and reliability of development.
Unlike traditional four-legged robots, this model is better adapted to the environment where people live. It can overcome the stairs, avoid obstacles and leave “free paws” to perform tasks. This makes it especially useful for people care, rescue and interaction with a person in a real environment.
”, – WRITE: mezha.media

Miroslav Trinko Geek, specialty programmer, but journalist by profession. Rider, tennis player and fan of Formula-1. I write about technologies, smartphones and electric vehicles.
The uniqueness of this development is to use a bioin -aged controller Tumblernetworking on a deep training with reinforcement (Deep Reinforcement Learning). It allows the work to simulate the human course and adapt to different conditions – even on sand, loose stones or mats.
Tumblernet combines information about the mass center and the pressure center into a single system that mimics human balancing. This can make the robot freely move from four -legged to two -to -two mode, using front or rear paws, as well as performing complex movements, including reversals and motion in a circle.
Another feature of it is stability. It can withstand strokes, pushing and even falling – and does not require an additional balance of balance. In case of loss of balance, robots can rise on its own, even if it caught on the objects.
Researchers also tested work on the beach – an environment where it is difficult to keep balance. Despite the difficult conditions, it successfully moved on two paws, which indicates a high adaptability and reliability of development.
Unlike traditional four-legged robots, this model is better adapted to the environment where people live. It can overcome the stairs, avoid obstacles and leave “free paws” to perform tasks. This makes it especially useful for people care, rescue and interaction with a person in a real environment.