Intend to Move: A Multimodal Dataset for
Intention-Aware Human Motion Understanding

Ryo Umagami¹  Liu Yue¹  Xuangeng Chu¹  Ryuto Fukushima¹  Tetsuya Narita¹
Yusuke Mukuta^1,4  Tomoyuki Takahata²  Jianfei Yang³  Tatsuya Harada^1,4
1The University of Tokyo, ²Tokyo Denki University, ³Nanyang Technological University, ⁴RIKEN AIP

Intend to Move (I2M) is a new multimodal dataset for embodied AI, designed for intention-aware human motion understanding in real-world environments.
I2M includes diverse modalities: 3D human motion, multi-view RGB-D video, 3D point cloud & mesh, and intention text.

✅️ Data and code will be released soon!