Currently the development of a robot It can be like that of a newborn human being because to the extent that he is working like this, only that he does it faster thanks to systems that have gradually perfect themselves, do you want to know about this surprising new advance?
This humanoid robot is like a little baby learns by reinforcement while walking
Until a few years ago one of the main difficulties for developers of robotic structures It had been to provide them with natural movements similar to human, so algorithms based on mathematical models planning each movement were created.
However, the locomotion of these humanoid robots continued to be and seem artificial, but recently a Chinese company has devised a solution that is giving a lot to talk about. It is a new algorithm where a learning theory is used that has traditionally been applied in people.
Reinforcement learning through a Algorithm is being implemented with an automaton called Adamso that it acquires locomotive skills similar to human. In other words, as if it were a baby when he starts crawling, falling and getting up to take the first steps.
The manufacturer of ADAM, PNDBOTTICS, has reported that the objective is the improvement of flexibility for an effective adaptation to a dynamic and changing environment, a continuous improvement process that began in June 2023 and in which the lower extremities and feet have been reinforced also making it more durable.
If each step means the advance towards perfection, it would not surprise that in the end he plays better than Messi
The robotic development company based in China is making an automaton that will be able to walk as if it were a personusing an algorithm based on reinforcement or RL learning, according to acronym in English, waiting for this humanoid to be able to:
-- Stay standing
- Walking fluidly
- Move stability
- Irregular surfaces
The distinction in this case is that the applied algorithm leads to an adjustment in this robotized machine of human movements at the Real moment, so that it can achieve the necessary changes in the environment where it interacts with a high level of autonomy, a transcendental achievement in the judgment of the publication in the publication in Interesting Engineering.
Achievement is having overcome a technical challenge that had always been present in the field of robotics Humanoid and this implies the successful closure of the existing gap between the induced simulation at the digital level and the application of these robots in real conditions, so the next step can be to give greater freedom to the robots.
Overcoming this technical challenge means free and free humanoids in the future
The realization of complex movements have been led by neuronal networks in this humanoid machine that weighs approximately 60 kilos and measures 1.6 m, getting a lot of freedom in arms, legs and waist and even a couple of 360 nm in its limbs, which forms an almost perfect drive with a high degree of control.
The system of this humanoid robot is integrated into an Intel i7 processor and a Real -time network in which data from human mobility were used adapting them to the structure of this robotic machine. Also, in the optimization of locomotion, public data and precise capture of movements were used.
One of the peculiarities of this automaton is not only that walking as a human but that every time it will be doing better because the RL system allows you to improve the efficiency in energy administration and provides a security environment, helping to reach the following advantages:
- Reduction of mechanical wear
- Realization of soft movements
- Coordinated motor articulation
In conclusion, through new control algorithms based on learning and imitation learning that have allowed developers to have better access to their training, Adam has been built, a surprising robot humanoid that learns as a newborn but a thousand times faster.
