Interactive Expressive Motion Generation Using Dynamic Movement Primitives

TL;DR

This work addresses the challenge of producing realistic, expressive nonverbal behavior in social robots. It proposes a unified framework that encodes a subset of the 12 Principles of Animation as parameterized Dynamic Movement Primitives, enabling learnable, explainable, and online-adaptable expression through principled modulations. The approach demonstrates diversity of expression using a single base model across simulation and three robots, and validates perceptual efficacy via a user study showing above-chance emotion recognition and modulation-aligned intensity. The method offers a data-efficient, composable pathway for advancing human-robot interaction by delivering controllable, goal-directed expressive motion.

Abstract

Our goal is to enable social robots to interact autonomously with humans in a realistic, engaging, and expressive manner. The 12 Principles of Animation are a well-established framework animators use to create movements that make characters appear convincing, dynamic, and emotionally expressive. This paper proposes a novel approach that leverages Dynamic Movement Primitives (DMPs) to implement key animation principles, providing a learnable, explainable, modulable, online adaptable and composable model for automatic expressive motion generation. DMPs, originally developed for general imitation learning in robotics and grounded in a spring-damper system design, offer mathematical properties that make them particularly suitable for this task. Specifically, they enable modulation of the intensities of individual principles and facilitate the decomposition of complex, expressive motion sequences into learnable and parametrizable primitives. We present the mathematical formulation of the parameterized animation principles and demonstrate the effectiveness of our framework through experiments and application on three robotic platforms with different kinematic configurations, in simulation, on actual robots and in a user study. Our results show that the approach allows for creating diverse and nuanced expressions using a single base model.

Figures (6)

• Figure 1: Our framework takes demonstrated trajectories and modulation parameters as input and generates expressive robot motion. Shown are two joint angles over time with black (dashed) demonstrations and two example modulations in red and blue. Some animation principles define primary and secondary motion, which are shown as opaque for the primary and translucent for the secondary joints.
• Figure 2: Phase adaptions for progression modulation with $\mathcal{P}_{Slow}$ and $\mathcal{P}_{Time}$. The linear phase, used during learning, is shown as a dashed line.
• Figure 3: Modulation with individual principles. The demonstrated trajectories are shown as a dashed line.
• Figure 4: Selected modulations in higher dimensions. The demonstrated trajectory is shown as a dashed line.
• Figure 5: Two of the robots used in our experiments: Pepper (left), Daryl (right)