ResearchBenchmarksGoal-conditioned navigation on MiniGrid 16x16 (RGB image input) (test)Follow0.44Clean RewardTDRT-PPO0.37760.39380.410.4262Jun 6, 2024Evaluation ResultsMethodMethodLinksClean RewardAttack Reward: Rew Max (SA-RL)Attack Reward: Rew Max (PA-AD)Attack Reward: BIA-ILfD (ours)Attack Reward: AvgTDRT-PPOTarget Reward=0.98 ± 0Target Reward=0.98 ± 02024.060.440.470.630.390.5PPOTarget Reward=0.98 ± 0...Target Reward=0.98 ± 0, Defense Status=No defense2024.060.380.490.940.480.64