HAZARD Challenge: Embodied Decision Making in Dynamically Changing Environments (ICLR 2024)

Qinhong Zhou*, Sunli Chen*, Yisong Wang, Haozhe Xu, Weihua Du, Hongxin Zhang, Yilun Du, Joshua B. Tenenbaum, Chuang Gan

News Mar 20: HAZARD challenge will be one of the challenges in CVPR Embodied AI Workshop 2025. The leaderboard submission is open now, welcome 2025 submissions! (will end in 1 June)

Detailed documentations

Get started

• Overview
• Installation
• Create your own agent and submit
• Common utils

Agent documents

• Observations

• Action space

• Default agents

  • RL agent
  • MCTS agent
  • Random agent
  • Greedy agent
  • Rule-based agent
  • LLM-based agent
    • Version 1.0
    • Version 2.0

• Customized agent

Baseline results

• Full log files can be found in this link
• W/o perception (use ground truth segmentations)