Scientific ML and robot learning researcher: LLM-driven algorithm design and embodied benchmark building
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-15) 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—
No publication-count data is available yet, so this axis is unscored. It would measure total scholarly output — number of papers, venues, and authorship roles — and cannot be estimated from the current snapshot.
Impact—
No citation counts or h-index data are available, so impact is unscored. This axis would measure citation-based influence — citations, h-index, and uptake of the two reviewed papers — and should not be inferred.
Focus0/100
The focus score is 0, reflecting maximal spread rather than concentration: the raw field distribution splits exactly evenly between scientific machine learning (0.5) and robotics, robot learning / embodied AI (0.5). With one cataloged paper on each side, current evidence shows a two-field portfolio rather than a single dominant specialization; additional publications would be needed to tell whether this even split is a durable working pattern.
Collaboration60/100
The collaboration score is 60, computed from a raw count of 19 distinct coauthors, 12 of whom recur across projects — including Mingyang Yu, Xu Yang, Jun Zhang, Jing Xu, Keqian Li, Ziyu Chen, Henghui Bao, Haoran Chang, Alan Yu, Ran Choi, Kai McClennen, and Gio Huh. On both cataloged papers Wang sits in a middle author position and is not corresponding author, consistent with a contributor role inside larger multi-person teams.
Activity—
No publication-timing data is available, so activity is unscored. It would measure research cadence — recent output rate and consistency across years — beyond the two 2026 papers visible in the current snapshot.
Reproducibility55/100
Reproducibility scores 55, the mean readiness rating across exactly 2 scored papers. Both rate medium: PINNForge reports solid statistics while winning 24/25 PDE tasks but leaves LLM invocation costs uncounted; the Neverwhere suite ships a 60+ scene 3D Gaussian Splatting benchmark whose sim-to-real proof rests on a single policy with n=20 trials and weak statistics. Together these drive the moderate 55 mean.
Research summary
Xiaolong Wang works at the intersection of scientific machine learning and robotics (robot learning / embodied AI), with the field distribution splitting exactly evenly at 0.5 and 0.5. The Evolve catalog holds two 2026 arXiv papers, on both of which Wang is a middle-listed author rather than corresponding author. PINNForge applies large language models with execution feedback to evolve physics-informed neural network designs for PDE solving; Evolve's review credits it with winning 24/25 PDE tasks and solid statistics, while tempering novelty relative to Lang-PINN and noting uncounted LLM costs. The Neverwhere Visual Parkour Benchmark Suite contributes a 60+ scene 3D Gaussian Splatting benchmark with closed-loop evaluation for quadruped visual parkour, though its sim-to-real evidence rests on one policy and n=20 trials with weak statistics. Collaboration metrics show 19 distinct coauthors and 12 recurring partners, a collaboration score of 60; reproducibility averages 55 (medium readiness) across the two scored papers. Output, impact, and activity scores are missing from the snapshot, so no publication-count or citation claims are made. No institution, country, or ORCID is attached, and because the name is common, this profile is name-resolved from public bibliographic data.
Collaboration pattern
The collaboration score is 60, derived from 19 distinct coauthors with 12 recurring partners — including Mingyang Yu, Xu Yang, Jun Zhang, Jing Xu, Keqian Li, Ziyu Chen, Henghui Bao, Haoran Chang, Alan Yu, Ran Choi, Kai McClennen, and Gio Huh. On both reviewed papers Wang holds a middle author position and is not the corresponding author, suggesting a contributor role inside larger teams that span the scientific ML and robotics sides.
Research keywords
Highlights
LLM-driven evolutionary design of physics-informed neural networks (PINNForge)
A 60+ scene 3DGS benchmark for quadruped visual parkour (Neverwhere)
An evenly split two-field research portfolio
Across the Evolve catalog, Wang's field distribution splits exactly 0.5 scientific machine learning and 0.5 robotics (robot learning / embodied AI), with one middle-author 2026 paper in each field and 19 distinct coauthors bridging both communities.
Timeline
- 2026PINNForge appears on arXiv: large language models with execution feedback drive evolutionary design of physics-informed neural networks for PDE solving; Evolve's review credits 24/25 winning tasks with solid statistics.
- 2026The Neverwhere Visual Parkour Benchmark Suite appears on arXiv, contributing a 60+ scene 3D Gaussian Splatting benchmark with closed-loop evaluation for quadruped visual parkour.
- 2026Both cataloged papers land in the same year, one on each side of the 0.5/0.5 field split, and each receives a medium reproducibility readiness rating (mean 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.
- "Xiaolong Wang" is a common name; this profile is name-resolved from public bibliographic data and identity could be ambiguous without an ORCID.
- Bibliometrics come from a public bibliometric snapshot only; output, impact, and activity scores are missing because publication counts and citation data were not available.
- No institution, country, or ORCID is attached, so affiliation and career context are absent from this profile.
- The evidence base is small: 2 reviewed papers, both 2026 arXiv preprints with middle authorship; conclusions may not generalize.