Robotic Task Planning in Dynamic Environments on OmniGibson
74Success RateLookPlanGraph
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LookPlanGraphBackbone=Llama3.2, Ground-truth Augmentation=true2025.12 | 74 | 86 | |
| LookPlanGraphBackbone=GPT-4o, Ground-truth Augmentation=true2025.12 | 71 | 84 | |
| ReActBackbone=GPT-4o, Ground-truth Augmentation=true2025.12 | 66 | 78 | |
| ReActBackbone=Llama3.2, Ground-truth Augmentation=true2025.12 | 63 | 75 | |
| LookPlanGraphBackbone=GPT-4o, Ground-truth Augmentation=false2025.12 | 42 | 69 | |
| ReActBackbone=GPT-4o, Ground-truth Augmentation=false2025.12 | 41 | 64 | |
| SayPlan LiteBackbone=GPT-4o, Ground-truth Augmentation=false2025.12 | 40 | 67 | |
| LLM+PBackbone=GPT-4o, Ground-truth Augmentation=false2025.12 | 37 | 62 | |
| LookPlanGraphBackbone=Llama3.2, Ground-truth Augmentation=false2025.12 | 35 | 60 | |
| LLM-as-PBackbone=GPT-4o, Ground-truth Augmentation=false2025.12 | 33 | 59 | |
| ReActBackbone=Llama3.2, Ground-truth Augmentation=false2025.12 | 33 | 52 | |
| SayPlan LiteBackbone=Llama3.2, Ground-truth Augmentation=false2025.12 | 27 | 55 | |
| LLM+PBackbone=Llama3.2, Ground-truth Augmentation=false2025.12 | 21 | 40 | |
| LLM-as-PBackbone=Llama3.2, Ground-truth Augmentation=false2025.12 | 16 | 36 | |
| SayPlanBackbone=GPT-4o, Ground-truth Augmentation=false2025.12 | 12 | 31 | |
| SayPlanBackbone=Llama3.2, Ground-truth Augmentation=false2025.12 | 10 | 26 |