Submitted : 2026 Mar.Accepted by CVPR 2026 https://arxiv.org/abs/2603.12967 Language-Grounded Decoupled Action Representation for Robotic ManipulationThe heterogeneity between high-level vision-language understanding and low-level action control remains a fundamental challenge in robotic manipulation. Although recent methods have advanced task-specific action alignment, they often struggle to ge..
Submitted : 2026 FebICRA 2026 Accepted (로보틱스 학회-2025년 기준 45% Acceptance Rate, 우수 컨퍼런스) https://arxiv.org/abs/2602.02142 FD-VLA: Force-Distilled Vision-Language-Action Model for Contact-Rich ManipulationForce sensing is a crucial modality for Vision-Language-Action (VLA) frameworks, as it enables fine-grained perception and dexterous manipulation in contact-rich tasks. We present Force-Distilled ..
Published : 2024 , Oct. 31 https://www.pi.website/blog/pi0 Our First Generalist PolicyOur first generalist policy, π0, a prototype model that combines large-scale multi-task and multi-robot data collection with a new network architecture to enable the most capable and dexterous generalist robot policy to date.www.pi.website 벤치마크 평가에서 비교군으로 꾸준히 나오는만큼 볼 필요가 있다. 마찬가지로 수식이 깨져서 $ab= ^ _ 이런 형식으로 보이면..
Published : 2026/03/03https://www.pi.website/research/memory VLAs with Long and Short-Term MemoryMulti-Scale Embodied Memory (MEM) gives our models both long-term and short-term memory, enabling complex tasks longer than ten minutes.www.pi.websitehttps://arxiv.org/abs/2603.03596 MEM: Multi-Scale Embodied Memory for Vision Language Action ModelsConventionally, memory in end-to-end robotic learnin..
NeurIPS 2025 poster Acceptedhttps://sites.google.com/view/forcevla2025 force-vlaAbstract Vision-Language-Action (VLA) models have advanced general-purpose robotic manipulation by leveraging pretrained visual and linguistic representations. However, they struggle with contact-rich tasks that require fine-grained control involving forcesites.google.com 수식적으로 기법 원리에 대해 설명을 추가할 예정. 현재는 모두 글로 이루어져 있음..
IEEE Access Acceptedhttps://vla-survey.github.io/ Vision-Language-Action Models for Robotics: A Review Towards Real-World ApplicationsA comprehensive review of Vision-Language-Action models for robotics, covering architectures, learning paradigms, and real-world applications.vla-survey.github.io논문 제목 : Vision-Language-Action Models for Robotics: A Review Towards Real-World Applications 본 리뷰 논문은 ..
[데이콘] 2025 전력사용량 예측 AI 경진대회건물의 전력사용량 예측 AI 모델 개발먼저, LSTM, CatBoost는 성능이 크게 다르지는 않았다. 정말 XGboost가 좋은 것 같다. 또한 건물 유형별과 건물 100개 따로 학습 후 사용한 것에 대해서는 건물 유형별로 모델 돌린 것이 더 좋게 나왔다. 여러 전략이 중요한 것 같다.먼저, 동일한 문제의 이전 대회에서 불쾌 지수와 같은 피쳐를 추가했고, 우리도 추가하게 됐다.train_ft['THI'] = 9/5*train_ft['기온(C)'] - 0.55*(1-train_ft['습도(%)']/100)*(9/5*train_ft['습도(%)']-26)+32test_ft['THI'] = 9/5*test_ft['기온(C)'] - 0.55*(1-test_ft[..
