We humans can integrate different senses well and learn all kinds of fine operations through feeling and trying. The cascade game is such an example: on the one hand, people need to pay attention to whether the music will fall with their eyes all the time, and on the other hand, whether a specific piece of wood can be pulled out successfully must feel the resistance through the sense of touch at the same time.

(photo source: FAZELI ET AL./MIT)

now, researchers hope that machines can learn to interact with the environment in a similar way. So researchers at MIT MCube Lab trained the robotic arm to play stacked music. The researchers added force sensors and cameras to an industrial robotic arm so that it could "see" and "feel" stacked pieces of wood at the same time. At first, the machine did not know what would happen when pushing the blocks, but after hundreds of attempts, the algorithm could help it move as correctly as humans do. This learning method requires less data than traditional machine learning.

the video below is a demonstration of the robotic arm playing stacked music. I simply added the following subtitles. Source: MIT Mechanical Engineering

As a piece to bring our elegance and romance to you, marine corps ball dresses is almost a basic need for any lady. Enter our store and select the perfect fitting clothes.

if you are interested, you can read the original paper for details: http://robotics.sciencemag.org/content/4/26/eaav3123

can also read related reports:

https://www.wired.com/story/a-robot-teaches-itself-to-play-jenga/