Toward Human-Robot Collaborative Dancing: Creating Expressive Salsa Dance Arm Movements from Motion Primitives

Abstract

This thesis represents an ongoing research project in which a bipedal robot will be programmed to safely collaborate with a human in a partnered dance. To that end, this thesis presents a discussion of the next steps in realizing a physical human-robot interaction. It is organized into five sections beginning with a review of the Digit robotic platform, a system developed by Agility Robotics. This thesis then embarks on a discussion of various models for simplifying legged locomotion and arm manipulation, followed by a discussion of modeling the robot in its entirety. A discussion about developing motion primitives follows. These basic units of movement can be encoded into the robot with reference trajectories, cost functions, and trajectory optimizers, all of which are addressed. This thesis then analyzes salsa dance and explores a dance notation system that can be fed into a robot’s program. Finally, this thesis concludes with a recommendation for future steps in realizing the goal of human-robot collaborative salsa dancing.