AI2-THOR

AI2-THOR is a benchmark that contains a simulated indoor environment used to evaluate the performance of agents in Object-Goal Navigation tasks, where agents must autonomously explore and navigate toward target objects based on semantic labels.

Papers using AI2-THOR (5)

Dualthor: A Dual-arm Humanoid Simulation Platform For Contingency-aware Planning2025AION: Aerial Indoor Object-goal Navigation Using Dual-policy Reinforcement Learning2026GRIP: A Unified Framework For Grid-based Relay And Co-occurrence-aware Planning In Dynamic Environments2025DualTHOR: A Dual-Arm Humanoid Simulation Platform for Contingency-Aware Planning2025AION: Aerial Indoor Object-Goal Navigation Using Dual-Policy Reinforcement Learning2026