PROJECT TITLE :

Trial and Error: Using Previous Experiences as Simulation Models in Humanoid Motor Learning

ABSTRACT:

Since biological systems have the power to efficiently reuse previous experiences to vary their behavioral ways to avoid enemies or find food, the number of required samples from real environments to boost behavioral policy is greatly reduced. Even for real robotic systems, it is fascinating to use solely a restricted variety of samples from real environments because of the restricted durability of real systems to cut back the desired time to boost management performance. In this article, we tend to used previous experiences as environmental native models therefore that the movement policy of a humanoid robot can be efficiently improved with a limited range of samples from its real atmosphere. We tend to applied our proposed learning methodology to a real humanoid robot and successfully achieve two challenging management tasks. We applied our proposed learning approach to accumulate a policy for a cart-pole swing-up task in a real-virtual hybrid task environment, where the robot waves a PlayStation (PS) Move motion controller to move a cart-pole in a virtual simulator. Furthermore, we have a tendency to applied our proposed methodology to a challenging basketball-shooting task in a very real atmosphere.