ai2thor

[arXiv 2023] Embodied Task Planning with Large Language Models

🐍 A Python Package for Seamless Data Distribution in AI Workflows

SIMS-V: simulated instruction-tuning data generation for spatial video understanding

Learning Embeddings that Capture Spatial Semantics for Indoor Navigation, NeurIPS ORLR 2020

🤖 Lightweight and customizable reinforcement learning environment based on AI2-THOR

Visible AI-driven robotics demos powered by VLM policies, MCP skills, and AI coding agents. Reviewable HTML reports — frames, maps, tool traces, scores — instead of buried terminal logs.

Zero-shot object navigation in RoboTHOR combining VLM-guided frontier exploration, Tree-of-Thoughts planning, and real-time RGB-D semantic mapping; requires no object-specific training.

MultiAgent Cooperative instruction following Env.

AI Robotics Demos 2026 - VLM Policies, MCP Skills & HTML Reports

AI-powered predictive navigation assistant for visually impaired students that uses DINOv3 semantic perception, JEPA world modeling, cognitive mapping, YOLOE open-vocabulary detection, EasyOCR landmark reading, and AttnLRP explainability to enable trustworthy, map free campus navigation with real time voice guidance.

Vision-Language-Action system for humanoid manipulation. Achieves 77% accuracy on real-world cooking videos.

Safe hierarchical embodied-AI framework with multi-layer safety filtering and formal recovery — IEEE TENSYMP 2026

Object Detection server on Interactive AI2THOR environment using Rust and yolov5s

Autonomous safety transfer for embodied agents in Reactor-generated worlds — Team Oneiros

ALFRED 具身任务下一步宏动作模型微调项目：数据清洗、Qwen2.5-7B LoRA SFT 与严格/物理容错评估

Have you ever dreamed of a Smart Pet which can be trained in modern world based on AI-powered approaches?

AI2-THOR to Blender Dataset Converter

AI2-THOR와 CoELA 구조를 활용한 LLM 기반 다중 로봇 협업 연구

Verified spatial counterfactual dataset generation