Robots that have the ability to “learn” to do specific tasks are nothing new, but truly autonomous models are still a thing of the future. The Tokyo Institute of Technology is working on a robot that’s supposed to be able to learn, adapt to new situations and act in a human-like way someday.
When compared to similar robots out there, “HIRO” is pretty impressive. As you can see in the video embedded below, the robot uses a algorithm called SOINN (Self-Organizing Incremental Neural Network) [JP] to do its magic (hardware-wise, HIRO [JP] is rather bland).
When confronted with a new problem, HIRO tries to “remember” past actions and accesses stored information in order to solve it (in the simulated environment shown in the clip below, HIRO is preparing a glass of cold water by itself). It also can be “taught” to do certain tasks, and it’s able to learn and react by analyzing the world around him (video and audio data).
We’ve seen that before, but what’s interesting about this robot is that it can connect to other sources to obtain more information to solve problems. When in trouble, its makers say HIRO will be able to access the web by itself to get information or even connect to other robots that have experience in overcoming a specific obstacle.
Here’s a video showing the robot in action (shot by Diginfonews in Tokyo, in English):
Seattle, WA
San Diego, CA
Menlo Park, CA
San Francisco, CA
Berlin, Germany
Boston, MA
