MIRROR: Motion-Internalized Reasoning with Reciprocal On-Policy Reinforcement for Unified Motion Generation and Understanding

Submitted to AAAI, 2027

Authors: Wenhao Shen, Hao Wang, Nanjie Yao, Shanshan Zhao, Qing-Guo Chen, Weihua Luo, Guosheng Lin

Venue/status: Submitted to AAAI 2027.