arXiv:2611.10061 · v1

UniLatent: Cross-Embodiment Transfer via Latent Observation Alignment

首次提交 2026-11-06 15:35

版本更新 2026-11-06 15:35

雷达观测 2026-11-10 10:12

摘要

Robot datasets are costly to collect and are usually tied to a specific embodiment, making it difficult to reuse demonstrations across robots with different geometry, appearance, and kinematics. We introduce UniLatent, a cross-embodiment transfer framework that reduces the embodiment-induced observation gap in a shared latent space. UniLatent renders paired source-target robot observations in simulation and aligns a source visual encoder to a frozen target encoder. A policy trained on encoded source demonstrations can then run directly on target observations, with no pixel-level image translation. Across three simulation tasks and two target robot arms, UniLatent reaches up to 93% zero-shot success, exceeding the strongest pixel-translation baseline by over 30 points on average. In real-world xArm-to-Franka transfer, fine-tuning with only 10 target demonstrations per task yields 27/30 successful rollouts. Code, videos and data are available at https://selina2023.github.io/unilatent/.

作者备注：Accepted to CoRL 2026