New AI Lets Robots Learn From Sloppy Human Videos — No Expert Needed

Imitation learning lets robots acquire complex skills directly from massive demonstration datasets, but its performance degrades severely when those datasets are contaminated with suboptimal or noisy demonstrations. Prior quality-assessment methods typically rely on manual pre-selection of expert reference data or task-specific heuristics, which limits scalability. Enter SynIL (Synergy-based Imitation Learning), a framework for automated, label-free demonstration quality assessment in offline reinforcement learning. It's 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. Evaluations on D4RL locomotion benchmarks and multi-human Robomimic manipulation datasets show synergy-derived rewards correlate strongly with ground-truth rewards. SynIL also substantially outperforms Behavior Cloning and achieves performance comparable to — and in sparse-reward human teleoperation scenarios, superior to — offline reinforcement learning trained on true environment rewards.