具身智能科技前瞻探索（第3期）：多任务操作、第一人称世界模型、低光照与模糊感知

微信扫码，打开到小程序

2 01 02 03 3 / 3 / CONTENTS 01 02 03 04 05 06 07 4 / 4 5 MoE-ACT 1.1   6 MoE-ACT 1.2 数据来源：《MoE-ACT: Scaling Multi-Task Bimanual Manipulation with Sparse Language-Conditioned Mixture-of-Experts Transformers》  7 / 7 8 EgoSim 2.1   9 EgoSim 2.2 数据来源：《EgoSim: Egocentric World Simulator for Embodied Interaction Generation》  10 / 10 11 E-VLA VLA 3.1   12 E-VLA VLA 3.2 数据来源：《E-VLA: Event-augmented Vision-Language-Action Model for Dark and Blurred Scenes》  13 / 13 14 CRAFT 4.1   15 CRAFT 4.2 数据来源：《CRAFT: Video Diffusion for Bimanual Robot Data Generati》  16 / 16 17 Heracles 5.1   18 Heracles 5.2 数据来源：《Heracles: Bridging Precise Tracking and Generative Synthesis for General Humanoid Control》  19 / 19 20 ThermoAct VLA 6.1   21 ThermoAct VLA 6.2 数据来源：《ThermoAct: Thermal-Aware Vision-Language-Action Models for Robotic Perception and Decision-Making》  22 / 22 23    24 THANKS FOR LISTENING