Tianyu Xin

University of Electronic Science and Technology of China · China

Unclaimed · AI-structured profileOpenAlex

Early-career robotics researcher at UESTC working on vision-language-action models for precise robot manipulation

papers here

reviewed by Evolve

first-author

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—

This axis measures publication volume and productivity over time, but the underlying scoring data is insufficient for this profile, so no output score is reported and none should be estimated from the reviewed papers alone.

Impact—

This axis measures citation-based influence, such as citations accrued to recent works, but the data needed to compute it is not yet available for this profile; no impact score is reported here.

Focus0/100

Scores 0/100, the lowest measured axis, reflecting low topical concentration. The field distribution splits evenly — 0.5 robotics/robot learning and 0.5 robot learning (vision-language-action models) — and OpenAlex topic counts scatter across five clusters: advanced image fusion techniques (6), plus reinforcement learning in robotics, image enhancement, image processing, and domain adaptation with few-shot learning (3 each).

Collaboration66/100

Scores 66/100. The OpenAlex record lists 26 distinct coauthors, and 12 recur frequently, including Xingjie Fan, Lijie Wang, Zheng Lu, Yiming Wang, and Heyang Yu. Both cataloged 2026 arXiv papers place this researcher in middle-author positions without corresponding authorship, consistent with contributing inside larger teams on vision-language-action robotics. That breadth of stable collaborators supports the moderately strong collaboration reading.

Activity70/100

Scores 70/100. Per-year counts show 4 works in 2025 rising to 6 in 2026, with 10 works inside the recent three-year window — an accelerating tempo. Both cataloged papers appeared on arXiv in 2026, so the recent output is concentrated in the current year rather than spread thinly, consistent with an active early-stage publishing rhythm.

Reproducibility55/100

Scores 55/100, averaged over the 2 reproducibility-assessed papers in the catalog, both graded medium readiness. Reviews credit solid experimental design — 'Acting in Meters' is described as well-ablated with consistent simulation and real-robot gains — but flag single-seed statistics, unreported variance on small LIBERO-Long margins, and no code release, which cap confidence in replication.

Research summary

Tianyu Xin is a robotics researcher at the University of Electronic Science and Technology of China whose recent work centers on vision-language-action (VLA) models for robot manipulation. The available bibliometric record (OpenAlex) is early-stage: 10 works in the recent three-year window, with per-year counts of 4 in 2025 and 6 in 2026, an h-index of 1, and 26 distinct coauthors. Two 2026 arXiv papers appear in the Evolve catalog, both with middle-author positions: 'Acting in Meters,' which introduces metric-interaction adapters (ICT + MAIF) that consistently improve VLA/WAM manipulation success in simulation and on real hardware, and 'LayerRoute,' which adds action-conditioned mixture-of-layers routing with gains up to +7.2pp on LIBERO-Long. Topic profiling splits evenly between robotics/robot learning and robot learning (vision-language-action models), alongside clusters in advanced image fusion (6), reinforcement learning in robotics, image enhancement, image processing, and domain adaptation/few-shot learning (3 each). Collaboration is the strongest measured dimension (66/100), activity scores 70, and reproducibility 55; output and impact lack sufficient data to score. Both cataloged papers carry medium reproducibility readiness.

Collaboration pattern

Collaboration is the strongest measured dimension (66/100). The OpenAlex record lists 26 distinct coauthors, with 12 recurring collaborators including Xingjie Fan, Lijie Wang, Zheng Lu, Yiming Wang, and Heyang Yu. Both cataloged 2026 papers carry middle-author positions without corresponding authorship, consistent with contributing inside larger teams on vision-language-action robotics rather than solo-led efforts.

Research keywords

vision-language-action modelsrobot manipulationworld-action modelsflow matchingmetric scalepoint cloud geometrylayer routingLIBERO benchmarkreinforcement learning in roboticsimage fusionimage enhancementdomain adaptation and few-shot learning

Highlights

Metric-interaction adapters lift precise manipulation
Introduces ICT and MAIF adapters that learn metric-scale interactions for robotic manipulation; Evolve's review finds consistent success gains for VLA and WAM policies in simulation and on real hardware, with strong ablations — though single-seed statistics and no code release temper certainty.
Action-conditioned mixture-of-layers routing for VLA policies
Adds action-conditioned VLM-layer routing plus an action-state reread mechanism to vision-language-action policies, with multi-benchmark improvements reaching +7.2pp on LIBERO-Long; the margins are real but small and reported without variance statistics.

Timeline

2025Earliest year in the available bibliometric window: OpenAlex per-year counts record 4 publications.
2026Per-year publication count rises to 6, the highest of the covered years and part of a 10-work recent three-year record.
2026'Acting in Meters: Learning Metric Interactions for Precise Robotic Manipulation' appears on arXiv, introducing ICT + MAIF metric-interaction adapters for VLA/WAM manipulation.
2026'LayerRoute: Action-Conditioned Mixture-of-Layers Routing for Vision-Language-Action Policies' appears on arXiv, reporting gains up to +7.2pp on LIBERO-Long.

Papers in the Evolve catalog

Frequent co-authors

Xingjie Fan×2 Lijie Wang×1 Zheng Lu×1 Yiming Wang×1 Heyang Yu×1 Kenghou Hoi×1 Bowen Hu×1 Di Cui×1 Haoran Liao×1 Wanqi Zhong×1 Yizhao Xu×1 Ziliang Wang×1

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.

Bibliometric figures (26 coauthors, h-index 1, per-year counts) come from an OpenAlex-derived snapshot; no ORCID was available, so identity is name-resolved from public bibliographic data.
Output and impact dimensions are unscored because the underlying data is insufficient; citation standing should not be inferred from this profile.
Only the two 2026 arXiv papers in the Evolve catalog have been individually reviewed; other works in the recent record are not assessed here.
OpenAlex topic labels (e.g., image fusion, image enhancement) may partly reflect keyword-level assignment rather than current research focus.