SynIL: Leveraging Synergy for Offline Imitation Learning from Imperfect Demonstration Datasets

What happened

arXiv:2609.38225v1 Announce Type: new Abstract: Imitation learning enables robots to acquire (one company buying another) complex skills directly from massive demonstration datasets, but its performance degrades severely when datasets are contaminated with suboptimal or noisy demonstrations. Comprehensive evaluations on D4RL locomotion benchmarks and multi-human Robomimic manipulation datasets demonstrate that synergy-derived rewards correlate strongly with ground-truth rewards.

While prior quality-assessment methods attempt to filter or reweight data, they typically rely on manual pre-selection of expert reference data or task-specific heuristics, limiting scalability. Grounded in neuroscientific evidence that motor synergy, a low-dimensional coordinated structure in movement, correlates directly with motor proficiency, SynIL algorithmically quantifies synergy manifestation to generate dense, transition-level reward signals via self-supervised reward regression.