Search

4 results for “RoboTwin 2.0”

1. Research · Robotics & embodied AI

   Robot-relative 3D features improve pretrained manipulation policies across diverse tasks

   Spatial Grafting: Grounding 3D Features for Flow-Matching Robot Policies

   A lightweight module grounded reconstruction features in metric robot coordinates and improved four vision-language and world-action policy hosts without redesigning their perceptual backbones.

   BEHAVIOR-1K: 0.63 Q-score (as reported)

   29 Sept 2026University of Toronto, Huawei Noah’s Ark Lab, 2012 Labs

2. Research · Robotics & embodied AI

   Robots complete more two-handed tasks when AI repeatedly refines actions

   Vision-Language Models Are Joint Denoisers in Robotic Action Generation

   Gong and colleagues report a 20.8-percentage-point improvement across four real-world tasks by involving a vision-language model throughout action generation.

   RoboDojo: 18.29 score (as reported)

   3 Nov 2026Peking University, The Hong Kong University of Science and Technology (Guangzhou), Microsoft Research Asia

3. Research · Robotics & embodied AI

   Autoregressive video pretraining unlocks the value of long context for real-time robot control

   Long-WAM: Scaling the Context of World-Action Models

   Long-WAM scales visual history up to 19.2 seconds for action prediction, achieving high success on dynamic tasks.

   DOMINO: 45.1 manipulation score (as reported)

   9 Oct 2026NVIDIA, MIT, HKU

4. Research · Robotics & embodied AI

   Predictive visual models improve action generation for generalist robot manipulation

   InternW0-$Δ$: A World Action Model Bridging Predictive Dynamics and Actions with 20K+ Hours of Open Data

   InternW0-Delta combines video, semantic, geometric and action models, trained on more than 20,000 hours of heterogeneous demonstrations.

   AC-One: 95% success rate (as reported)

   28 Sept 2026Shanghai AI Laboratory