Robotouille is a new interactive benchmark showing LLM agents succeed on 47% of synchronous plans but only 11% when tasks require overlapping timed actions.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.RO 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Robotouille: An Asynchronous Planning Benchmark for LLM Agents
Robotouille is a new interactive benchmark showing LLM agents succeed on 47% of synchronous plans but only 11% when tasks require overlapping timed actions.