Skip to content
RobotWorld
← Tags

#Emergent Capabilities (2)

WaterfallPagination
π0.7: a Steerable Generalist Robotic Foundation Model with Emergent Capabilities

π0.7: a Steerable Generalist Robotic Foundation Model with Emergent Capabilities

π0.7 is a steerable generalist robotic foundation model conditioned on diverse multimodal context—language commands, subgoal images, and task metadata—enabling strong out-of-the-box performance in unseen environments, zero-shot cross-embodiment transfer, and emergent capabilities matching RL-finetuned specialists.

Physical Intelligence, Bo Ai, Ali AminApr 16, 2026
Robot foundation modelsVLAvision-language-actionApr 16, 2026
Emergent Compositional Skills in Mixture-of-Experts VLAs

Emergent Compositional Skills in Mixture-of-Experts VLAs

We consider the problem of learning compositional robot policies end-to-end from expert demonstrations, without any pre-specified notion of task decomposition or hierarchy. We ask whether a VLA trained with a simplified Mixture-of-Experts (MoE) action head can emergently learn to decompose tasks into reusable, interpretable primitives. We find that learned experts are heavily reused across tasks and consistently correspond to qualitatively distinct low-level behaviors, suggesting that the router implicitly learns to perform high-level sequencing while experts serve as compositional primitives. Our MoE matches the task performance of a monolithic baseline while demonstrating meaningful expert specialization, a step toward modular, interpretable robot policies that emerge from data alone.

Shlok Shah, Rhiaan Jhaveri, Tharun Kumar Tiruppali KalidossJul 22, 2026
VLAHybrid ExpertCompositional GeneralizationJul 22, 2026