Robots are already playing a big role in manufacturing environments, labs, and other places of business. In the future, they are going to become even more autonomous. Take this Agile Justin for instance. It is an advanced humanoid robot programmed to classify materials by touching them, using deep learning.
The robot relies on a flexible tactile skin on its finger tip and deep learning network architecture to complete the task. The skin gathers data and sends them to the network, which happens to be trained from only 540 samples. According to the researchers, it has a classification accuracy of up to 97.3%. Pretty impressive, don’t you think?