D-JEPA proposes decision-aligned latent world model, hits 87.89% on PushT
NEBULIS-Lab · hf · 2026-09-29
NEBULIS-Lab introduces D-JEPA, a decision-aligned latent world model.
Problem: latent world models predict action consequences, but accurate prediction doesn't guarantee latent distance reflects which candidate will execute successfully — the paper identifies a "decision-local prediction gap" where candidates predicted closer to the goal can yield worse realized outcomes.
Method: D-JEPA learns decision-relevant relations among candidate futures from executed outcomes, using a bounded permutation-equivariant operator that jointly reasons over goal-relative predictive features and ordinal evidence, refining pretrained predictive geometry where action choices matter most. It realizes this structure in JEPA-compatible representations for native latent-distance planning.
Results: gains across latent control, manipulation, pretrained action models, physical robots and autonomous driving — 87.89% success on PushT, +15.04 avg on RoboTwin, +17 points on physical robot tasks.
More from Embodied
- ProHand Gen 2 Power robotic hand ships with 2M-cycle rating and 66% more abduction — imankitgoyal · 2026-09-29
- Sentdex Ports Robot Locomotion Training to mjlab, Walk Still Needs Work — Sentdex · 2026-09-29
- OpenAI Researcher Sheryl Hsu Departs, Turning to Post-AGI Robotics Work — SherylHsu02 · 2026-09-29
- YC-backed Aviern flies autonomous planes to inspect oil & gas pipelines, alerts in ~30s — ycombinator · 2026-09-29
- Robotics startup Roboteur hiring gripper/hand specialist, more home demos next — neurosp1ke · 2026-09-29
- Volvo's autonomous haulers have moved 3 million tonnes of earth — cisc · 2026-09-29