Early-stage researcher at Shanghai Jiao Tong University: LLM-enhanced EV-grid optimization and safe robot learning
2
papers here
2
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-16) 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 catalog holds too few indexed outputs to compute a reliable publication-volume measure. This axis would count arXiv and journal papers, venues, and authorship positions over time once a fuller bibliometric record is linked.
Impact—
Not yet scored: citation counts, h-index, and downstream adoption are unavailable in the current snapshot, so research influence cannot be measured. This axis would track citations, citing works, and uptake of the two 2026 preprints as they accrue.
Focus0/100
Scores 0, the lowest possible concentration: the reviewed output splits exactly 50/50 between reinforcement learning for smart energy systems and robot learning for safe embodied-AI control. One line builds LLM-enhanced multi-agent RL for EV charging and grid optimization; the other enforces hard constraints on flow-matching vision-language-action policies. The shared thread is learning-based decision-making under constraints, but the two communities and evaluation cultures remain distinct.
Collaboration48/100
Scores 48, a mid-range collaboration profile built on 10 distinct coauthors in the current snapshot. The named network spans Lindong Xie, Chongyu Wang, Gaojunjie Li, Siqi Bu, Edward Chung, Jianming Ma, Rongjun Jin, Xiaxi Si, Yiheng Li, and Yue Gao, mixing energy-systems and robotics partners. Both cataloged papers are team efforts — one first-authored, one middle-authored, neither corresponding — suggesting contributions within larger groups rather than solo or corresponding-author-led work.
Activity—
Not yet scored: the snapshot lacks publication-frequency and per-year output data needed to gauge research tempo. This axis would measure how actively new preprints and papers appear over time; only two 2026 arXiv entries are available, too thin to score.
Reproducibility38/100
Scores 38, the mean reproducibility-readiness across the 2 reviewed papers, and the gap between them is wide. The EV-charging MARL paper rates low: its headline metric is self-defined, which hinders external comparison. ActSafeGuard rates medium, with strong results but Fast-WAM table inconsistencies and no error bars. Code release, standardized benchmarks, and variance reporting would be the clearest levers to raise this score.
Research summary
Yang Zhang is a researcher at Shanghai Jiao Tong University working across two learning-based control domains: reinforcement learning for smart energy systems and robot learning for safe embodied AI. The Evolve catalog holds two 2026 arXiv papers. The first, first-authored, claims the first unified multi-agent reinforcement learning framework spanning electric vehicles, charging stations, and the grid in public charging systems, adding LLM-based feature selection and reward weighting on top of MADDPG; Evolve credits broad simulation gains but notes the headline metric is self-defined. The second, ActSafeGuard, is a middle-author contribution adding a differentiable, training-aligned ray-scaling layer that enforces hard constraints on flow-matching vision-language-action policies; results are strong, though Fast-WAM table inconsistencies and missing error bars temper confidence. The current bibliometric snapshot is small: 10 distinct coauthors (collaboration score 48), an exact 50/50 split between the two fields (focus score 0), and a mean reproducibility-readiness of 38 across the two reviewed papers (low and medium). Neither paper lists Zhang as corresponding author, and output, impact, and activity cannot be scored for lack of citation and volume data. The result is an early-stage, preprint-driven profile: technically ambitious in bridging LLM-augmented multi-agent optimization and safe policy enforcement, but awaiting peer-reviewed venues, standardized benchmarks, and fuller bibliometric coverage before its influence can be judged.
Collaboration pattern
Yang Zhang works in teams rather than solo: both cataloged papers are multi-author efforts, one first-authored and one middle-authored, neither corresponding. The snapshot lists 10 distinct coauthors — including Lindong Xie, Chongyu Wang, Gaojunjie Li, Siqi Bu, Edward Chung, Jianming Ma, Rongjun Jin, Xiaxi Si, Yiheng Li, and Yue Gao — spanning the energy-systems and robotics lines, consistent with the mid-range collaboration score of 48.
Research keywords
Highlights
First unified LLM-enhanced MARL for EV–station–grid optimization
ActSafeGuard: hard constraints for flow-matching VLA policies
Timeline
- 2026Posts a first-authored arXiv paper on LLM-enhanced multi-agent reinforcement learning for unified EV–charging station–grid optimization in public charging systems.
- 2026Co-authors ActSafeGuard, a differentiable, training-aligned constraint-enforcement layer for flow-matching vision-language-action policies.
- 2026Both papers enter the Evolve catalog with reproducibility-readiness ratings of low (EV-charging MARL) and medium (ActSafeGuard).
- 2026The snapshot records 10 distinct coauthors across the two papers and an exact 50/50 split between smart-energy RL and safe 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.
- “Yang Zhang” is a common name; this profile is name-resolved from public bibliographic data and the two reviewed papers, so attribution to other same-named researchers is possible.
- Radar scores and the coauthor count come from a small bibliometric snapshot covering only the two reviewed 2026 arXiv papers; output, impact, and activity dimensions lack data.
- Both papers are arXiv preprints; reproducibility ratings reflect Evolve review notes (self-defined metrics, table inconsistencies, no error bars), not peer-reviewed replication.
- No citation counts, h-index, ORCID, or country information was provided, and earlier or unpublished work is not represented in the 2026 timeline.