Xiao Ma

ByteDance Seed

収録論文 30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

• BridgeVLA++: データ効率的で汎化性が高く、メモリ拡張された3D操作のための視覚-言語-行動フレームワーク
  VLA/操作2026/8/5

  事前学習済み視覚言語モデルを活用した3Dロボット操作フレームワークBridgeVLAを拡張し、空間的・時間的メモリを統合することで、データ効率と汎化性を保ちつつ、記憶依存の操作タスクで最先端の性能を達成した。

• SiMDex: 類似する自己中心視点動画のマイニングによるクロス身体性器用操作
  データマイニング/器用操作/VLA2026/8/4

  ロボットの器用操作のためのVLAモデル訓練に有効な人間の自己中心視点動画を、類似性に基づくデータマイニングで選別するフレームワークを提案。約3200万サンプルから1.49M件を選び、成功率を47.7%から61.1%に向上させた。

• BridgeVLA++: A Data-Efficient, Generalizable, and Memory-Augmented Vision-Language-Action Framework for 3D Manipulation
  2026/8/1
• SiMDex: Mining Similar Egocentric Videos for Cross-Embodiment Dexterous Manipulation
  2026/8/1
• TTT-VLA: Test-Time Latent Prompt Optimization for Vision-Language-Action Models
  2026/6/1
• World Value Models for Robotic Manipulation
  2026/6/1
• Hand-in-the-Loop: Improving VLA Policies for Dexterous Manipulation via Seamless Hand-Arm Intervention
  2026/5/1
• GR-Dexter Technical Report
  2025/12/1
• GR-RL: Going Dexterous and Precise for Long-Horizon Robotic Manipulation
  2025/12/1
• WMPO: World Model-based Policy Optimization for Vision-Language-Action Models
  2025/11/1
• ByteWrist: A Parallel Robotic Wrist Enabling Flexible and Anthropomorphic Motion for Confined Spaces
  2025/9/1
• GR-3 Technical Report
  2025/7/1
• Human-assisted Robotic Policy Refinement via Action Preference Optimization
  2025/6/1
• Chain-of-Action: Trajectory Autoregressive Modeling for Robotic Manipulation
  2025/6/1
• BridgeVLA: Input-Output Alignment for Efficient 3D Manipulation Learning with Vision-Language Models
  2025/6/1
• Environment-Driven Online LiDAR-Camera Extrinsic Calibration
  2025/2/1
• What Matters in Building Vision-Language-Action Models for Generalist Robots
  2024/12/1
• Effective Tuning Strategies for Generalist Robot Manipulation Policies
  2024/10/1
• Green Screen Augmentation Enables Scene Generalisation in Robotic Manipulation
  2024/7/1
• BiGym: A Demo-Driven Mobile Bi-Manual Manipulation Benchmark
  2024/7/1
• Redundancy-aware Action Spaces for Robot Learning
  2024/6/1
• Hierarchical Diffusion Policy for Kinematics-Aware Multi-Task Robotic Manipulation
  2024/3/1
• InsActor: Instruction-driven Physics-based Characters
  2023/12/1
• DaXBench: Benchmarking Deformable Object Manipulation with Differentiable Physics
  2022/10/1
• Imitation Learning via Differentiable Physics
  2022/6/1
• Hierarchical Reinforcement Learning under Mixed Observability
  2022/4/1
• Ab Initio Particle-based Object Manipulation
  2021/7/1
• Learning Latent Graph Dynamics for Visual Manipulation of Deformable Objects
  2021/4/1
• Spatio-Temporal Graph Transformer Networks for Pedestrian Trajectory Prediction
  2020/5/1
• Differentiable Algorithm Networks for Composable Robot Learning
  2019/5/1