Recursive, hyperspherical behavioral learning for robotic control

Robots, undoubtedly, are governed by a set of behavioral policies. However, embedding these policies becomes problematic and complex due to the nondeterministic properties of the task and environment. Learning from demonstration, or LFD, alleviates this vexatious conundrum and expedites the mapping process, for the robot implicitly learns the desired objective. This paper presents a novel method for facilitating behavior learning in robots. The algorithm employed, called Recursive, Hyperspherical Behavioral Learning, or RHBL, actively translates the teacher´s reactions to various stimuli into a behavioral tree, which defines the robot´s current domain knowledge. Once the tree is formulated, the anthropomorphic robot demonstrates proficiency in the observed, complex task, for the tree elicits responses from various stimuli, defining the robot´s autonomous behavior. Details of the algorithm and the results of its application are presented in the paper.