Preferred Networks

History

2018: Use of AI technologies in training of promrobot of Toyota and Fanuc

On May 17, 2018 it became known that the artificial intelligence of Preferred Networks began to be applied in production of industrial robots of Toyota and Fanuc companies. 

According to Bloomberg, investors estimated Preferred Networks at $2 billion thanks to investments of clients. So, Toyota Motor invested over $110 million in algorithm elaboration which will help the company to compete with Google in the field of unmanned vehicles. 

The main difference of Preferred Networks from hundreds of other startups in the company consider its communications with industrial enterprises of Japan and application of algorithms of deep learning on production. Use of deep learning has huge potential in the field of production, Yutaka Matsuo, the specialist in the field of computers of the University of Tokyo and the president of the Japanese association of deep learning considers.

The AI startup worth $2 billion teaches robots to think

At the Consumer Electronics Show exhibition in Las Vegas demonstration using toy cars showed some opportunities of this technology. Then several miniature Toyota Prius cars set in the field with obstacles were provided. At first machines could hardly move and all the time collided. But after two hours of tests and errors, they moved as if they had professional drivers inside.

Any programmer did not write for them the instruction. Robots developed own rules from experience — as though people studied on errors of your friend, Bloomberg notes.

After Preferred Networks showed how its technology in the future can be used to make industrial robots similar to the qualified masters. Programming of the robot collector of Fanuc for capture of objects from lump can take several days the engineer person. Experiments showed that machines can cope with this task for one night.^[1]

Notes