NTU Releases ACE-Data-0: Large-Scale Synchronized Multimodal Robotics Dataset

liuziwei7 · x · 2026-08-05

The MMLab team at Nanyang Technological University (NTU) has released ACE-Data-0, a large-scale synchronized multimodal dataset. The core highlight is that it aligns multiple modalities—such as ego and multi-view exo videos, human and hand motions, object motion, audio, and tactile sensing—onto a unified timeline and a shared coordinated system.

The dataset aims to establish benchmark tasks for perception, state recovery, and interaction understanding. The team notes its significant application value in imitation learning, world models, Vision-Language-Action (VLA) models, and general robotic intelligence.

Related event: NTU Open-Sources Multimodal Embodied AI Dataset(2 posts)→

Original post →

More from Embodied

Embodied channel →