Automated-planning researcher evaluating how reliably frontier LLMs translate natural-language goals into PDDL
1
papers here
1
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-09-21) 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—
Output metrics are not yet available: the data includes no total publication count or venue list. This axis would measure the volume and recency of publications beyond the single 2026 first-author paper currently in the Evolve catalog.
Impact—
Impact cannot be scored: no citation counts, h-index or venue influence data were provided. This axis would measure how often the researcher's work is cited and adopted, which cannot be estimated from a single recently posted 2026 preprint.
Focus100/100
The focus score is 100, the maximum, because the field distribution in the snapshot contains exactly one field: artificial intelligence (automated planning). The reviewed 2026 paper sits squarely inside planning, applying PDDL to LLM evaluation — an undiversified, tightly targeted research concentration.
Collaboration22/100
The collaboration score is 22, based on 2 distinct co-authors recorded in the snapshot. Lukas Chrpa and G. Michael Youngblood appear as frequent co-authors, and the 2026 catalog paper is first-authored. The small collaborator pool suggests a compact, specialist working pattern rather than a broad international network.
Activity—
Research activity is unscored because recent-year publication frequency, submission cadence and career-stage indicators are absent from the data. This axis would measure ongoing research pace; the catalog currently shows one 2026 output and nothing about surrounding years.
Reproducibility20/100
The reproducibility score is 20, computed from 1 reviewed paper with a repro-readiness of 'low'. Evolve's review notes the headline result — six frontier LLMs above 92% on NL-to-PDDL translation, Gemini 2.5 Flash best at 96% — rests on single-run, n=90 results without statistical support, limiting independent verification.
Research summary
Tomas Balyo is an early-stage profile in Evolve's catalog, positioned in artificial intelligence with a single captured field: automated planning. The snapshot's field distribution contains exactly that one entry, producing a maximum focus score of 100, while the collaboration score of 22 reflects just 2 distinct co-authors — Lukas Chrpa and G. Michael Youngblood — and the reproducibility score of 20 comes from one reviewed paper rated low on repro-readiness. That paper, a 2026 first-author arXiv preprint titled 'Which LLM is Best for Translating Natural Language Goals to PDDL', benchmarks six frontier large language models on converting natural-language goal statements into PDDL. It reports that all six exceed 92% accuracy, with Gemini 2.5 Flash best at 96%. Evolve's verdict judges the claims plausible but statistically under-supported: results come from a single run over n=90 instances without significance testing. The profile therefore describes a focused planning researcher working at the LLM-planning interface, using prompt engineering and benchmark methodology to probe how reliably generative models produce formal planning specifications. Output, impact and activity axes remain unscored because no publication counts, citations or multi-year productivity data are available. The evidence base is thin — one reviewed 2026 preprint and a narrow bibliometric snapshot — so conclusions should be treated as provisional until more of the researcher's record is captured.
Collaboration pattern
Tomas Balyo's recorded collaboration footprint is compact: 2 distinct co-authors appear in the bibliometric snapshot, yielding a collaboration score of 22. Lukas Chrpa and G. Michael Youngblood are the recurring partners, consistent with a small specialist team in automated planning. The single reviewed 2026 paper is first-authored (non-corresponding), suggesting hands-on technical contribution within a tight group rather than leadership of a large distributed consortium.
Research keywords
Highlights
Six frontier LLMs all exceed 92% at natural-language-to-PDDL goal translation
Controlled head-to-head benchmark for a core planning translation task
Verdict: plausible results, but single-run n=90 without statistical support
Timeline
- 2026First-author arXiv preprint published: 'Which LLM is Best for Translating Natural Language Goals to PDDL'.
- 2026Benchmark of six frontier LLMs on NL-to-PDDL goal translation reported, with Gemini 2.5 Flash best at 96%.
- 2026Paper enters the Evolve catalog with a reproducibility review scoring its repro-readiness low (repro mean 20).
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.
- Dimension scores and co-author counts come from a narrow bibliometric snapshot and may not reflect the researcher's full publication record.
- No institution, country or ORCID was provided; the profile is name-resolved from public bibliographic data and identity is not independently verified.
- The evidence base is one reviewed 2026 arXiv preprint; the reproducibility score of 20 rests on a single paper.
- Output, impact and activity axes are unscored because publication counts, citations and multi-year activity data were not available.