Feldenkrais Robot

This small monkey like robot, unlike Asimo, gets Feldenkrais lessons which teach it to move gracefully, and without fear, to become highly flexible and fast, it can jump, roll and hang on to things, ride a bike and even drive a car.

It learns to do so by gentle trial and error, through these lessons of "Awareness Through Movement"

Adaptive learning is a known AI concept with endless implementations. Positive feedback is part of the adaptive learning process.

Of course the motors will not be one or two but rather many small motors all controlled via an array of sensors, affecting each other and notifying the control center. Perhaps a "Neural Network" device, a MAS (Multi-Agent System) or just a (currently) conventional computer running "parallel" control software.

The robot will be programmed either to receive a set of commands (there is only a restricted amount of information given in these sessions), which would be developed by Feldenkrais practitioners and "translated" into instructions, i.e. via some Laban Symbol input system, or better yet, will simply use audio to text, and translate the instructions into it's own "trial and error" sessions, resulting with a robot that can jump, run, roll and tiptoe.

If you ever had a Feldenkrais "lesson", (or gave one) you know that this method works by either giving general instructions or by gently moving the practicing person's body, while receiving constant feedback, alerted to what they should expect, called to awareness of what they feel, and cautioned to be careful not to harm themselves, by observing and attending to the body's warnings, received as pain, pressure or other bodily sensations.

The robot will emulate these through the sensor array, and respond with similar actions through its array of motors or artificial muscles (see link).