Event
Authors Propose PIFR Benchmark for Navigation Interruption Recovery
1 reports · 1 sources
Zerun Liu and other authors propose and release the PIFR benchmark, designed to evaluate the ability of systems in long-horizon embodied navigation to complete an inserted task and then resume the original task after a high-priority instruction interrupts the current task. The benchmark includes three interruption settings—single, repeated, and nested—and defines three stage metrics: completion of the inserted task, resumption of the original task, and completion of the entire sequence. The authors also propose a lightweight candidate-ranking baseline, which they say is used to improve target re-localization after interruption. What has been released so far is the design of the benchmark and baseline; the report does not provide evaluation results.
Report timeline
- PIFR: A Benchmark for Preemptive Instruction Following and Task Resumption in Long-Horizon Embodied Navigation
The authors propose the PIFR benchmark to evaluate preemption and resumption capabilities in long-horizon embodied navigation when a higher-priority instruction interrupts the current task sequence. PIFR includes single, repeated, and nested interruption settings, with stage-specific metrics for inserted-task completion, resumption of the original task, and completion of the full sequence. The authors also propose a lightweight candidate-ranking baseline to improve target re-grounding after interruption.
arXiv cs.RO ·