Reinforcement Learning Results
Qualitative rollout comparisons across models and environments
Random Policy Baseline
Center Point
Inside Room
Trapped Corner
Local Attention ResNet-18
Center Point
Inside Room
Trapped Corner
Sliding Window Transformer (Random Embedding)
Center Point
Inside Room
Trapped Corner
Sliding Window Transformer (No Curiosity)
Center Point
Inside Room
Trapped Corner
Sliding Window Transformer (PCA Head)
Center Point
Inside Room
Trapped Corner
New Environment (PCA Head)
Center Point
Bedroom
Toilet
