피지컬 AI Lab https://validator.w3.org/feed/docs/rss2.html Nav2 RateController: Why Running Actions Still Receive Ticks Nav2 Velocity Smoother: Choosing Command or Odometry Feedback openpi Remote Inference: Where Observation Transforms Belong TACTO: What Tactile Rendering Can and Cannot Validate AnySkin: Checking Policy Reuse After a Tactile Skin Replacement VoxPoser: Turning Language into 3D Planner Objectives SayCan: Why a Useful Robot Skill May Still Be Infeasible RT-H: Choosing Language Corrections or Teleoperated Corrections Where2Act: Why an Actionable Pixel Is Not a Complete Action PerAct vs Act3D: Two Ways to Represent 3D Robot Actions VRB: Learning Robot Affordances from Human Video RoboPoint: Turning a 2D Action Point into a Robot Target Code as Policies: What Generated Robot Code Needs to Run Touch100k: Choosing Tactile Language Labels for Your Task Sparsh: When to Freeze a Tactile Encoder BEHAVIOR 2026: Separating Policy Tuning from Evaluation ManiSkill 3: Comparing Visual Simulator Throughput Fairly SIMPLER: Diagnosing Sim-to-Real Policy Ranking Mismatches FoAR force feedback: when contact should change a robot policy NVIDIA アイザック GR00T N1.7：アーキテクチャ、データ、ファインチューニング、ロボット検証 Octo on a new robot: bind the checkpoint inputs and actions LIBERO suites: choose the right transfer test NVIDIA GR00T N2 與 N1.7 差在哪裡？版本與使用狀態 NVIDIA Isaac GR00T N1.7：架構、資料與機器人驗證 피지컬 AI Lab UMI demonstration transfer: what handheld robot data must preserve 사람 영상에서 로봇 정책까지, 오류가 시작된 단계를 찾는 법 How to find where a video-to-data robot pipeline fails ReKep 제약을 만족했는데 로봇 작업은 왜 실패할까 ReKep keypoint constraints: when the plan fits but the robot still fails GELLO 리더 장치가 시연 데이터 수집에 맞는 경우 When GELLO is worth building for robot demonstrations Why action encoding changes a robot world model’s forecast NVIDIA Isaac GR00T N1.7：架构、数据与机器人验证 NVIDIA Isaac GR00T N1.7: Architecture, Data, Fine-Tuning and Robot Validation ロボット推論遅延：カメラ露出からアクチュエータコマンドまで Isaac Lab-Arena policy regression testing: build a versioned evaluation gate DROID calibration repair: what corrected extrinsics can actually fix GIFT and the action-sufficiency gap in robot policy features TEMPO와 VLA의 시간 맥락: 최신 영상에도 놓치는 이동 정보 TEMPO and VLA temporal context: why a fresh frame can still miss motion ロボット機能安全：PL、 SIL および安全機能検証 ロボットグラスプ計画：候補、接触品質、実行可能な選択 TacO tactile sensors: choose by task, not resolution ROS 2 リアルタイム制御：締め切り、ジッター、ros2_control Gemini Robotics-ER 1.6: Spatial Reasoning, API Outputs and Robot Integration 같은 동작인데 예측이 달라진다면: 로봇 월드 모델의 행동 인코딩 점검 GR00T N2와 N1.7 차이: DreamZero 구조와 2026 공개 상태 DJI AP100 降落伞与 FTS：在 Matrice 400 BVLOS 安全论证中的作用 글 목록