Robot learning researcher: decision-time world-model re-ranking and task-semantic calibration for VLA policies
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 — no publication-count data is available for this profile. This axis would measure total output, venue spread, and authorship roles over time; the visible record so far is limited to two 2026 arXiv preprints.
Impact—
Not yet scored — no citation counts or influence metrics are available. This axis would measure citations, uptake, and downstream reuse; for a brand-new 2026 preprint record, citation evidence has simply not accumulated yet.
Focus0/100
This axis scores 0 because the field distribution splits evenly — 0.5 for "robotics / embodied machine learning" and 0.5 for "robot learning / world models". The two strands sit on one continuum, though: world-model planning and calibration of vision-language-action policies are adjacent parts of robot learning, so in a two-paper record the 0 reflects label granularity more than genuinely scattered research.
Collaboration55/100
Scores 55 from 15 distinct co-authors across the record. A recurring core — Shuaijun Liu, Feiyang You, and Chengyu Wu, each appearing twice in the frequency list — works alongside Shuyang Hao, Ningxin Su, Jingyao Cai, Xingwei Chen, Qifu Wen, and Xi Lin. Both cataloged papers place this researcher in a middle author position without corresponding-author responsibility, consistent with contributor roles in team-led robotics projects.
Activity—
Not yet scored — no per-year publication series is available. This axis would measure publishing cadence and continuity over time; the only dated signals in hand are the two 2026 arXiv preprints, too sparse to infer a pace.
Reproducibility55/100
Scores 55, the mean reproducibility readiness across 2 assessed papers, both rated medium. Evolve's review notes that D-JEPA ships no code despite consistent simulation and real-robot gains, while BAS-VLA provides solid simulation evidence but narrow preservation testing. Protocol-level rigor is present; artifact release and broader evidence remain the gap before a higher readiness rating.
Research summary
Chenglong Zhang works across robot learning and embodied machine learning; the Evolve catalog splits the visible record evenly between "robotics / embodied machine learning" and "robot learning / world models", each carrying 0.5 of the field distribution. Two 2026 arXiv preprints define that record. D-JEPA proposes a decision-aligned latent world model that re-ranks candidate futures with a bounded set transformer, closing a decision-local gap; Evolve reports consistent simulation and real-robot gains while flagging cohort selection, a tiny driving set, and no code release. BAS-VLA adds a task-semantic action calibration layer on frozen vision-language-action models: stable under appearance shift but breaking under semantic change, where stale success falls to 0%, with solid simulations yet narrow preserving evidence. Methodologically, both papers inject lightweight decision-time modules rather than retraining the underlying policy. Collaboration data shows 15 distinct co-authors with a recurring core — Shuaijun Liu, Feiyang You, Chengyu Wu — and both cataloged papers list this researcher as a middle author, not corresponding. Reproducibility readiness averages 55 across two papers rated medium. Citations, output volume, and yearly activity data are absent, so this remains an early-stage profile.
Collaboration pattern
The record lists 15 distinct co-authors. A stable core recurs — Shuaijun Liu, Feiyang You, and Chengyu Wu each appear twice in the frequency list — joined by Shuyang Hao, Ningxin Su, Jingyao Cai, Xingwei Chen, Qifu Wen, and Xi Lin. In both cataloged 2026 papers this researcher sits in a middle author position without corresponding-author responsibility, consistent with a specialist contributor inside a team-led robotics group rather than solo or PI-led output.
Research keywords
Highlights
D-JEPA: decision-aligned re-ranking for latent world models
BAS-VLA: task-semantic action calibration for frozen VLA models
A decision-time pattern: adapt frozen models instead of retraining
Both 2026 preprints inject lightweight modules at decision time — re-ranking candidate futures for a latent world model, calibrating actions for a frozen VLA policy — rather than retraining the underlying model; both carry medium reproducibility readiness.
Timeline
- 2026D-JEPA preprint appears on arXiv: a decision-aligned latent world model whose candidate futures are re-ranked by a bounded set transformer.
- 2026BAS-VLA preprint appears on arXiv: task-semantic action calibration for frozen vision-language-action models, robust to appearance shift but not to semantic change.
- 2026Both cataloged papers land in the same year, spanning world-model planning and VLA policy calibration within robot learning.
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.
- Bibliometrics derive from OpenAlex-sourced data; no institution, country, or ORCID is attached to this record.
- Impact, output, and activity dimensions are unscored because citation, publication-count, and per-year data were unavailable.
- Coverage is limited to the two 2026 arXiv preprints reviewed by Evolve; other works by this author fall outside this snapshot.
- "Chenglong Zhang" is a common name; this profile is name-resolved from public bibliographic data and may merge or split identities.