ResearchBenchmarksEmbodied Planning on WorldLines 21 action-dense samples downstream planning probeFollow68.4Plan JudgeObsMem32.31241.68151.0560.419Jun 17, 2026Evaluation ResultsMethodMethodLinksPlan JudgeState ConsistencyPrecondition ValidityMemory UseAction OrderObsMemN=21N=212026.0668.472.170.20.6967.6A-memN=21N=212026.0654.256.658.10.52455.3Mem0N=21N=212026.0652.655.156.30.51253.5GraphMemN=21N=212026.0648.149.350.70.46249.6MemoryOSN=21N=212026.0633.736.137.60.31935.2