Long-horizon DLO routing on DLO Routing Simulation (Implicit)
80Success Rate (SR)VLM w/ Failure Reasoning
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| VLM w/ Failure ReasoningHigh Level=VLM w/ Failure Reasoning, Low Level=RL Policy2025.10 | 80 | 2.7 | 13.9 | |
| Symbolic PlannerHigh Level=Symbolic Planner, Low Level=RL Policy2025.10 | 68 | 2.3 | 10 | |
| Fixed OrderHigh Level=Fixed Order, Low Level=RL Policy2025.10 | 53 | 2.1 | 9.7 | |
| VLM w/o Failure ReasoningHigh Level=VLM w/o Failure Reasoning, Low Level=RL Policy2025.10 | 47 | 2 | 16.2 | |
| VLM w/ Failure ReasoningHigh Level=VLM w/ Failure Reasoning, Low Level=Heuristic Policy2025.10 | 13 | 1.3 | 8 |