Tongtong Cao

Department of Foundation Model, 2012 Labs

収録論文 17本 ・ フィジカルAI/ロボット学習

※arXiv著者名で収集。同姓同名の別人の論文が含まれる場合があります。

論文

• Faster-WAM: 世界行動モデルに深い行動モジュールは必要か？
  VLA2026/8/3

  ビデオ世界モデルと行動予測を組み合わせた世界行動モデル（WAM）において、行動モジュールの深さをビデオバックボーンから切り離し、軽量な単層ヘッドをドッキングする設計（DoT）を提案。推論速度を3.2倍高速化しつつ、性能と汎化を維持した。

• Faster-WAM: Do World Action Models Need Deep Action Modules?
  2026/8/1
• RoboHarness: Memory-Driven Orchestration of Heterogeneous Robot Policies for Long-Horizon Planning
  2026/7/1
• KAM-WM: Kinematic Affordance Maps from Latent World Models for Robot Manipulation
  2026/7/1
• Decoupling Semantics and Geometric Grounding: Spatial Visual Prompts for Language-Conditioned Imitation Learning
  2026/6/1
• Anticipation-VLA: Solving Long-Horizon Embodied Tasks via Anticipation-based Subgoal Generation
  2026/5/3
• Anticipation-VLA: Solving Long-Horizon Embodied Tasks via Anticipation-based Subgoal Generation
  2026/5/1
• RePO-VLA: Recovery-Driven Policy Optimization for Vision-Language-Action Models
  2026/5/1
• Do World Action Models Generalize Better than VLAs? A Robustness Study
  2026/3/1
• Efficient Camera Pose Augmentation for View Generalization in Robotic Policy Learning
  2026/3/1
• H-WM: Robotic Task and Motion Planning Guided by Hierarchical World Model
  2026/2/1
• CAPE: Context-Aware Diffusion Policy Via Proximal Mode Expansion for Collision Avoidance
  2025/11/1
• UnPose: Uncertainty-Guided Diffusion Priors for Zero-Shot Pose Estimation
  2025/8/1
• HIPPo: Harnessing Image-to-3D Priors for Model-free Zero-shot 6D Pose Estimation
  2025/2/1
• Mem2Ego: Empowering Vision-Language Models with Global-to-Ego Memory for Long-Horizon Embodied Navigation
  2025/2/1
• SpatialCoT: Advancing Spatial Reasoning through Coordinate Alignment and Chain-of-Thought for Embodied Task Planning
  2025/1/1
• ET-Plan-Bench: Embodied Task-level Planning Benchmark Towards Spatial-Temporal Cognition with Foundation Models
  2024/10/1