MoRL: Reinforced Reasoning for Unified Motion Understanding and Generation

2026-02-16

AI Summary

This work addresses the limitations in human motion understanding and generation stemming from insufficient reasoning capabilities and inadequate planning during inference. The authors propose MoRL, a unified multimodal motion model that integrates supervised fine-tuning with reinforcement learning guided by verifiable rewards, and introduces Chain-of-Motion (CoM)—a novel test-time stepwise reasoning mechanism—to enhance the logical coherence and physical plausibility of generated motions. To support this approach, they design task-oriented composite rewards that enforce semantic-physical consistency and construct two large-scale chain-of-thought datasets, MoUnd-CoT-140K and MoGen-CoT-140K. Evaluated on the HumanML3D and KIT-ML benchmarks, MoRL significantly outperforms existing methods, achieving superior performance in both motion understanding coherence and perceptual realism of synthesized actions.

Technology Category

• Machine Learning: Multimodal Learning
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Computer Vision: Motion & Tracking

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

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.

Problem

• motion understanding
• motion generation
• reasoning capability
• test-time planning
• multimodal motion

Innovation

• reinforcement learning
• Chain-of-Motion
• multimodal motion modeling
• verifiable rewards
• reasoning coherence