Robot-learning researcher working on latent world models and task-semantic VLA calibration for manipulation
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-13) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
Not yet scored: the input contains no publication-count or productivity metrics beyond the two cataloged 2026 preprints. This axis would measure total output volume and per-year production once a fuller bibliographic record is linked.
Impact—
Unscored for now: no citation counts, h-index, or influence signals are available in the snapshot. This axis would measure how widely the two preprints are cited and taken up by the robot-learning community.
Focus0/100
The focus score is 0, driven by an even field distribution (0.5 / 0.5) across the two cataloged fields: robotics / embodied machine learning and robot learning / world models. In practice the split is less scattered than the score suggests, since both papers sit on adjacent sides of the same robot-learning problem — one on decision-aligned latent world models, the other on VLA calibration for manipulation.
Collaboration55/100
Collaboration scores 55, based on 15 distinct coauthors across the cataloged record. You appears as a middle author on both papers and as corresponding author on none, consistent with contributing inside a larger team. The frequent-coauthor list is led by Shuaijun Liu and Chengyu Wu, who each appear twice, alongside Shuyang Hao, Chenglong Zhang, Ningxin Su, Jingyao Cai, Xingwei Chen, Qifu Wen, and Xi Lin.
Activity—
Not scored: the snapshot provides no publication cadence or yearly volume data beyond two preprints both dated 2026, so recent activity cannot be measured yet. This axis would track output recency and sustained production over time.
Reproducibility55/100
Reproducibility averages 55 across the two papers Evolve has scored (reproPapers = 2), each rated medium readiness. Evolve's review of D-JEPA flags no released code plus cohort-selection and tiny-driving-set concerns; BAS-VLA is backed by solid simulations, though its preservation evidence is narrow. Reproducibility therefore rests mainly on reported sim and robot results rather than shared artifacts.
Research summary
Feiyang You is a robot-learning researcher whose Evolve-visible record centers on two 2026 arXiv preprints at the interface of world models and vision-language-action (VLA) manipulation. The first, D-JEPA, proposes a decision-aligned latent world model in which a bounded set transformer re-ranks candidate latent futures to close a decision-local gap in model-based planning; Evolve's review notes consistent simulation and real-robot gains while flagging cohort selection, a tiny driving set, and the absence of released code. The second, BAS-VLA, adds a task-semantic action calibration layer on top of frozen VLA models, staying stable under appearance shift but breaking under semantic change, where stale success falls to 0%. Both papers list You as a middle author rather than corresponding author. The bibliometric snapshot splits the cataloged output evenly (0.5/0.5) between robotics/embodied machine learning and robot learning/world models, yields a moderate collaboration score (55) from 15 distinct coauthors, and a reproducibility mean of 55 across the two scored papers, both rated medium readiness. Output, impact, and activity axes are unscored because citation and volume data are not in scope, and no institutional affiliation, country, or ORCID is attached; the profile is therefore name-resolved from public bibliographic data and should be read as an early-stage, fast-moving record.
Collaboration pattern
The record shows team-embedded work: 15 distinct coauthors, with You in the middle author position on both papers and no corresponding authorship. Shuaijun Liu and Chengyu Wu recur most often (twice each in the frequent-coauthor list), followed by Shuyang Hao, Chenglong Zhang, Ningxin Su, Jingyao Cai, Xingwei Chen, Qifu Wen, and Xi Lin. The cluster is consistent with a core group collaborating across both 2026 papers.
Research keywords
Highlights
D-JEPA: re-ranking latent futures for decision-aligned world models
BAS-VLA: task-semantic action calibration on frozen VLA models
Counterfactual evaluation separating appearance from semantic shift
Timeline
- 2026Posts the D-JEPA preprint on arXiv, introducing a decision-aligned latent world model for model-based planning.
- 2026Posts the BAS-VLA preprint on arXiv, adding task-semantic action calibration for vision-language-action models.
- 2026Both preprints enter the Evolve review catalog with medium reproducibility readiness, forming the two-paper base behind the reproducibility mean of 55.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Dimension scores are computed from public bibliometric data only; the record carries no institutional affiliation, country, or ORCID, so identity is name-resolved and namesakes cannot be fully excluded.
- Output, impact, and activity axes are unscored because citation counts, publication volume, and yearly cadence are not part of the provided snapshot.
- The profile draws only on the two 2026 arXiv preprints in the Evolve catalog; earlier or parallel work outside the catalog is not reflected.
- The frequent-coauthor list contains duplicate entries and includes the researcher's own name, so recurring-collaborator counts should be read as approximate.