Jesse Zhang

University of Washington

収録論文 18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

• Flex-π: 計算柔軟性を備えたマルチストリーム世界行動モデル
  VLA2026/8/11

  凍結したビデオ生成VAEが3D点群も符号化できることを発見し、RGBに加えて3D幾何学と物体意味論を共同で学習する世界行動モデルを提案。推論時にストリームを選択可能で、実世界の両腕操作タスクで高い性能を達成。

• Flex-$\pi$: A Multi-Stream World-Action Model with Compute Flexibility
  2026/8/1
• TMRL: Diffusion Timestep-Modulated Pretraining Enables Exploration for Efficient Policy Finetuning
  2026/5/1
• OGPO: Sample Efficient Full-Finetuning of Generative Control Policies
  2026/5/1
• Robometer: Scaling General-Purpose Robotic Reward Models via Trajectory Comparisons
  2026/3/1
• PEEK: Guiding and Minimal Image Representations for Zero-Shot Generalization of Robot Manipulation Policies
  2025/9/1
• HAND Me the Data: Fast Robot Adaptation via Hand Path Retrieval
  2025/5/1
• ReWiND: Language-Guided Rewards Teach Robot Policies without New Demonstrations
  2025/5/1
• HAMSTER: Hierarchical Action Models For Open-World Robot Manipulation
  2025/2/1
• EXTRACT: Efficient Policy Learning by Extracting Transferable Robot Skills from Offline Data
  2024/6/1
• RL-VLM-F: Reinforcement Learning from Vision Language Foundation Model Feedback
  2024/2/1
• LiFT: Unsupervised Reinforcement Learning with Foundation Models as Teachers
  2023/12/1
• TAIL: Task-specific Adapters for Imitation Learning with Large Pretrained Models
  2023/10/1
• Bootstrap Your Own Skills: Learning to Solve New Tasks with Large Language Model Guidance
  2023/10/1
• RoboCLIP: One Demonstration is Enough to Learn Robot Policies
  2023/10/1
• SPRINT: Scalable Policy Pre-Training via Language Instruction Relabeling
  2023/6/1
• COG: Connecting New Skills to Past Experience with Offline Reinforcement Learning
  2020/10/1
• REPLAB: A Reproducible Low-Cost Arm Benchmark Platform for Robotic Learning
  2019/5/1