Researchers are developing advanced frameworks for mobile manipulation in robotics, aiming to improve data efficiency and robustness. VOMMI utilizes portable demonstrations and offline trajectory reconstruction to enhance learning for embodied intelligence, showing reduced velocity errors compared to robot-collected data. MobileVISTA generates pose-perturbed training data from canonical demonstrations, improving robustness for robots like humanoids without needing generative models. MobiAgent employs a dual-loop system for lifelong learning, decoupling high-level reasoning from low-level control and achieving significant success rate improvements on benchmarks like BEHAVIOR-1K and RoboCasa. MM-ABC, a generalist foundation model, integrates perception, coordination, and imagination for mobile manipulation, achieving high success rates across multiple benchmarks and real-world tasks. AI
IMPACT These advancements in mobile manipulation frameworks could accelerate the development of more capable and robust robots for complex real-world tasks.
RANK_REASON Multiple research papers introducing new frameworks and models for mobile manipulation.
- alphaXiv
- Astribot S1
- BEHAVIOR-1K
- CatalyzeX Code Finder for Papers
- CORE Recommender
- DagsHub
- Gotit.pub
- Hugging Face
- Influence Flower
- MobiAgent
- RoboCasa
- ScienceCast
- Vision-Language Action Models
- Vision-Language models
- EBench
- LIBERO
- LIBERO-Plus
- ManiSkill-HAB
- MM-ABC
- MM-APT
- RoboCasa365
- Galaxea R1 Pro
- MobileVISTA
- Openpicus
- R2-VO
- Suzannah Wistreich
- VOMMI
AI-generated summary · Google Gemini · from 4 sources. How we write summaries →