Fondazione Bruno Kessler researcher: LLM coding agents for task and motion planning, RL from VLM teachers.
2
papers here
1
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
2
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-17) 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 underlying output data is insufficient in this snapshot. This axis would measure total publication volume over time — preprints, peer-reviewed papers, and venue spread — beyond the 11 recent-3y works already recorded.
Impact—
Not yet scored: the available citation signals (h-index 2, i10-index 1) are too early-stage to rank impact reliably. This axis would measure citation influence, uptake by other research groups, and scholarly reach once enough citing literature accumulates.
Focus100/100
Scores 100 on focus, reflecting tightly concentrated topics: OpenAlex topic counts are led by AI-based Problem Solving and Planning (7) and Reinforcement Learning in Robotics (6), with Software Engineering Research (4) and Robot Manipulation and Learning (3) adjacent; the single classified field entry is robotics/AI task and motion planning, and both catalog papers sit squarely in learning-based planning.
Collaboration48/100
Scores 48 on collaboration: the recent three-year window records 11 works with 10 distinct co-authors, a compact working-size network. Evolve's catalog adds ten frequent co-authors — Giovanni Bonetta, Davide Zago, Rossella Cancelliere and Bernardo Magnini on one line, and Tom Silver, Bowen Li, Josh Roy, Yichao Liang, Qianwei Wang and Yixuan Huang on another. Roles alternate between leading (first-author TAMP preprint) and contributing (middle-author SAGE paper).
Activity72/100
Scores 72 on activity. OpenAlex counts by year show 5 works in 2024, a dip to 1 in 2025, and a rebound to 5 in 2026 — 11 works across the recent three-year window. Both 2026 catalog papers are arXiv preprints, one first-authored, indicating a publication cadence that is accelerating again after the quiet 2025.
Reproducibility100/100
Scores 100 on reproducibility, computed from a mean of 100 across 1 reproducibility-assessed paper — a positive but narrow signal. The first-authored coding-agents TAMP preprint carries Evolve's "high" repro-readiness rating, consistent with its large, well-controlled 28-environment study, though the review flags training-data contamination risk; the SAGE paper is rated medium, with 3 seeds noted.
Research summary
Matteo Merler is a researcher at Fondazione Bruno Kessler in Italy whose work, as captured in this OpenAlex-resolved profile, centers on learning-based robot planning: task and motion planning (TAMP), reinforcement learning in robotics, and large language models used as program-synthesis engines for manipulation problems. The bibliometric snapshot is early-stage — h-index 2, i10-index 1 — but recent activity is steady: 11 works across the last three years (5 in 2024, 1 in 2025, 5 in 2026) with 10 distinct co-authors. OpenAlex topic counts are led by AI-based Problem Solving and Planning (7) and Reinforcement Learning in Robotics (6), followed by Software Engineering Research (4), Robot Manipulation and Learning (3), and Domain Adaptation and Few-Shot Learning (2). Two 2026 arXiv preprints anchor the Evolve catalog: as first author, "Coding Agents for Generalized Task and Motion Planning Problems," a large, well-controlled 28-environment study in which coding agents beat hand-engineered planners, rated high for reproducibility with a flagged training-data contamination risk; and as a middle author, "Selective Agent Guidance via Entropy" (SAGE), which distills entropy-gated vision-language-model guidance into PPO policies via behavioral cloning, showing sparse-reward gains at low teacher cost. The profile is name-resolved from public bibliographic data; no ORCID is linked.
Collaboration pattern
Collaboration scores 48: 11 recent-3y works with 10 distinct co-authors, a compact working-size network. The frequent-coauthor list pairs a long-running line (Giovanni Bonetta, Davide Zago, Rossella Cancelliere, Bernardo Magnini) with a robotics-learning circle (Tom Silver, Bowen Li, Josh Roy, Yichao Liang, Qianwei Wang, Yixuan Huang). Roles alternate between leading — first author on the 2026 coding-agents TAMP preprint — and contributing, as a middle author on SAGE.
Research keywords
Highlights
First-authored coding agents beat hand-engineered planners on generalized TAMP
Co-authored SAGE distills entropy-gated VLM guidance into RL policies
Strong open-reproducibility signal for the planning line of work
Timeline
- 2024OpenAlex records 5 works in this year, the start of the recent three-year publication window.
- 2025Recorded output dips to 1 work, the quietest year in the recent three-year window.
- 2026Output rebounds to 5 works; two arXiv preprints enter the Evolve catalog — the first-authored coding-agents study on generalized TAMP and the co-authored SAGE paper on entropy-gated VLM guidance.
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 come from OpenAlex only; the h-index 2 and i10-index 1 reflect an early-stage, name-resolved profile that may not capture every publication.
- Output and impact dimension scores are missing because the underlying data is insufficient; no volume or citation ranking is estimated.
- The reproducibility score of 100 is computed from only 1 assessed paper (reproPapers: 1), a narrow basis.
- No ORCID is linked and the profile is disambiguated from public bibliographic data, so identity should be confirmed before citing.