Juno pretrains a lightweight action-conditioned JEPA on the same embodiment-matched trajectories used for policy learning.
One action-conditioned JEPA improves vision-language-action policies across all lifecycle stages
Juno unifies representation, prediction, and adaptation in a single model, raising SimplerEnv success to 68.5% and retaining 70-75% real-robot success under shifts that cripple the base policy.
Academic
Yuchen Zhu · Chenyi Xu · Yulin Zhang · Gang Xu · Wentao Zhu
University of Science and Technology of China · Eastern Institute of Technology, Ningbo · Hangzhou Dianzi University · ShanghaiTech University
Research Digest··2 min read
The authors present Juno, a framework that repurposes a single action-conditioned joint-embedding predictive architecture (JEPA) as a control-aligned backbone, a predictive teacher, and an adaptable dynamics model across the vision-language-action (VLA) lifecycle.
Why this paper
From University of Science and Technology of China and 3 others
In one line
Juno uses a single action-conditioned JEPA to improve VLA success rates across pretraining, policy learning, and deployment.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ✓Reports numbers on named benchmarks
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§