MoRL: Reinforced Reasoning for Unified Motion Understanding and Generation
This addresses the need for improved logical reasoning and perceptual realism in motion tasks for vision and robotics applications, representing a novel integration of methods rather than a foundational breakthrough.
The paper tackles the problem of limited reasoning capability and test-time planning in human motion understanding and generation by proposing MoRL, a unified multimodal motion model trained with supervised fine-tuning and reinforcement learning with verifiable rewards, achieving significant gains over state-of-the-art baselines on HumanML3D and KIT-ML datasets.
Human motion understanding and generation are crucial for vision and robotics but remain limited in reasoning capability and test-time planning. We propose MoRL, a unified multimodal motion model trained with supervised fine-tuning and reinforcement learning with verifiable rewards. Our task-specific reward design combines semantic alignment and reasoning coherence for understanding with physical plausibility and text-motion consistency for generation, improving both logical reasoning and perceptual realism. To further enhance inference, we introduce Chain-of-Motion (CoM), a test-time reasoning method that enables step-by-step planning and reflection. We also construct two large-scale CoT datasets, MoUnd-CoT-140K and MoGen-CoT-140K, to align motion sequences with reasoning traces and action descriptions. Experiments on HumanML3D and KIT-ML show that MoRL achieves significant gains over state-of-the-art baselines. Code: https://github.com/AIGeeksGroup/MoRL. Website: https://aigeeksgroup.github.io/MoRL.