A Dance Performance with a Humanoid Robot using a Real-time Gesture Responsive Framework

We present a lightweight, real-time gesture responsive framework designed to investigate dynamics of human-robot interaction in live dance performances, merging aspects from choreography and robotics. In particular, it is tailored for artists to integrate human-sized humanoid robots into their dances seamlessly, offering an intuitive solution without the complexities of mastering robot control systems. Unlike existing interaction methods relying on wearable sensors or predetermined music cues, our framework, integrated in a task-space controller, enables the robot to dynamically respond to the dancer's movements, generating unpredictable yet artistically meaningful gestures without the need for body-mounted sensors. This design choice emphasizes the artistic intention behind the improvisation, illustrating that it is not merely about creating movements spontaneously but rather a way of expression in intuition guided by perception. We assess our framework on the HRP-4 humanoid robot and the result has been successfully demonstrated in a public human-robot dance performance at the Arts Center of Enghien-les-Bains's 23/24 season launch event, aiming to contribute to the exploration of co-creation between human and human-sized humanoid robot in improvised dance performances.