#Embodied AI

Habitat-Sim is a physics-enabled 3D simulation platform optimized for fast environmental rendering using 3D scans and CAD models. It supports datasets like HM3D and MatterPort3D, and facilitates robot simulation through URDF. It complements Habitat-Lab in developing embodied AI tasks. Available via Conda, it offers various installation options, along with comprehensive documentation and community forums.

Explore an advanced open-source platform for Embodied AI research by Allen Institute for AI. Supports environments like iTHOR and Habitat, offering task and algorithm versatility with PyTorch integration, visualizations, and multi-agent features. Provides tutorials, pre-trained models, and adaptable action spaces for diverse research applications.

Discover a comprehensive platform with over 200 environments and more than 2600 realistic objects tailored for embodied AI research. This framework supports multiple agents and a variety of AI tasks across diverse image settings, bridging simulation and real-world scenarios in a neutral and effective manner.

Explore up-to-date research on Embodied AI and agents powered by Large Language Models. Maintained by experienced researchers, this resource features pioneering studies, self-evolving agents, and innovative social agent frameworks. Key updates include the Mobile-Agent-v2 for improved mobile navigation and notable papers like 'MetaGPT,' acclaimed at ICLR 2024. The compilation offers insights into practical applications and research surveys showcasing the union of LLMs with Reinforcement Learning and World Models, keeping you informed on the latest in AI developments.