Autonomous animation of humanoid robots
Date of Issue2016
School of Mechanical and Aerospace Engineering
Robotics Research Centre
Carnegie Mellon University
A humanoid robot can actuate its multiple degrees of freedom to form whole body motions that convey meanings in different input signals. This thesis investigates how to autonomously animate a real humanoid robot given an input signal, such as gesturing to speech or dancing to the emotion expressed in music. This thesis addresses five core challenges: Representation of motions, Mappings of motions to meanings where meanings are represented as labels, Selection of relevant motions that considers the similarity between labels and audience preferences, Synchronization of motions to the input signal to form motion sequences, and Stability of the motion sequences (RMS^3). This thesis introduces a complete algorithm that solves the challenges of RMS^3, and selects a motion sequence to animate using a weighted criteria of audience preferences and stability. The approach and algorithms in this thesis are general to autonomously animate humanoid robots, and this thesis uses the NAO humanoid robot to autonomously animate speech and music.