MoConVQ: Unified Physics-Based Motion Control via Scalable Discrete Representations

and National Key Lab of General AI, China (a) Motion Tracking (b) Interactive Control (c) Text2Motion (d) LLM Integration Fig. 1. We present a method for learning discrete motion representation from a large-scale unstructured motion dataset for physics-based characters. The framework allows various applications, including those shown in this figure, to be accomplished in a unified fashion.

In this work, we present MoConVQ, a novel unified framework for physicsbased motion control leveraging scalable discrete representations. Building upon vector quantized variational autoencoders (VQ-VAE) and model-based reinforcement learning, our approach effectively learns motion embeddings from a large, unstructured dataset spanning tens of hours of motion examples. The resultant motion representation not only captures diverse motion skills but also offers a robust and intuitive interface for various applications.

We demonstrate the versatility of MoConVQ through several applications: universal tracking control from various motion sources, interactive character control with latent motion representations using supervised learning, physics-based motion generation from natural language descriptions using the GPT framework, and, most interestingly, seamless integration with large language models (LLMs) with in-context learning to tackle complex and abstract tasks.