@inproceedings{cai-etal-2026-atc,
title = "{ATC}-Bench: A Simulation-Grounded Benchmark for Evaluating {LLM}-Generated Air Traffic Control Instructions",
author = "Cai, Jinjun and
Li, Jiaxing and
Wang, Xingyue and
Zheng, Weihuang and
Kong, Youyong and
Ding, Hui",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.1552/",
doi = "10.18653/2026.findings-emnlp.1552",
pages = "33032--33055",
ISBN = "979-8-89176-462-0",
abstract = "Air traffic control (ATC) requires structured, safety-critical decision making whose effects unfold through aircraft trajectories, so text-level correctness alone is insufficient for evaluating large language models (LLMs) in this domain. We introduce ATC-Bench, a simulation-grounded benchmark for ATC instruction generation with Chinese-facing scenario descriptions and controller-style instruction fields. Given a structured Chinese traffic situation, a model generates a JSON control plan that is translated into BlueSky commands and evaluated through closed-loop simulation against separation, no-fly-zone (NFZ) avoidance, and route-recovery constraints. The benchmark spans four task families that form a controlled complexity gradient: single-aircraft NFZ avoidance, pairwise conflict resolution, joint conflict-airspace constraints, and multi-aircraft conflict graphs. Experiments with 11 representative LLM endpoints show that while most models produce format-valid instructions, task success varies widely across models and task families: the highest-scoring endpoint under our fixed protocol achieves 63.5{\%} macro-average task success rate (TSR), while the two lowest endpoints fall below 25{\%}, with the largest gaps appearing under joint constraints and multi-aircraft coordination. These results expose a persistent execution gap between static validity and simulation-passing behavior, motivating simulation-in-the-loop evaluation for LLM-based decision making in safety-critical aviation settings."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="cai-etal-2026-atc">
<titleInfo>
<title>ATC-Bench: A Simulation-Grounded Benchmark for Evaluating LLM-Generated Air Traffic Control Instructions</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jinjun</namePart>
<namePart type="family">Cai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiaxing</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xingyue</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Weihuang</namePart>
<namePart type="family">Zheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Youyong</namePart>
<namePart type="family">Kong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hui</namePart>
<namePart type="family">Ding</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Air traffic control (ATC) requires structured, safety-critical decision making whose effects unfold through aircraft trajectories, so text-level correctness alone is insufficient for evaluating large language models (LLMs) in this domain. We introduce ATC-Bench, a simulation-grounded benchmark for ATC instruction generation with Chinese-facing scenario descriptions and controller-style instruction fields. Given a structured Chinese traffic situation, a model generates a JSON control plan that is translated into BlueSky commands and evaluated through closed-loop simulation against separation, no-fly-zone (NFZ) avoidance, and route-recovery constraints. The benchmark spans four task families that form a controlled complexity gradient: single-aircraft NFZ avoidance, pairwise conflict resolution, joint conflict-airspace constraints, and multi-aircraft conflict graphs. Experiments with 11 representative LLM endpoints show that while most models produce format-valid instructions, task success varies widely across models and task families: the highest-scoring endpoint under our fixed protocol achieves 63.5% macro-average task success rate (TSR), while the two lowest endpoints fall below 25%, with the largest gaps appearing under joint constraints and multi-aircraft coordination. These results expose a persistent execution gap between static validity and simulation-passing behavior, motivating simulation-in-the-loop evaluation for LLM-based decision making in safety-critical aviation settings.</abstract>
<identifier type="citekey">cai-etal-2026-atc</identifier>
<identifier type="doi">10.18653/2026.findings-emnlp.1552</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1552/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>33032</start>
<end>33055</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T ATC-Bench: A Simulation-Grounded Benchmark for Evaluating LLM-Generated Air Traffic Control Instructions
%A Cai, Jinjun
%A Li, Jiaxing
%A Wang, Xingyue
%A Zheng, Weihuang
%A Kong, Youyong
%A Ding, Hui
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F cai-etal-2026-atc
%X Air traffic control (ATC) requires structured, safety-critical decision making whose effects unfold through aircraft trajectories, so text-level correctness alone is insufficient for evaluating large language models (LLMs) in this domain. We introduce ATC-Bench, a simulation-grounded benchmark for ATC instruction generation with Chinese-facing scenario descriptions and controller-style instruction fields. Given a structured Chinese traffic situation, a model generates a JSON control plan that is translated into BlueSky commands and evaluated through closed-loop simulation against separation, no-fly-zone (NFZ) avoidance, and route-recovery constraints. The benchmark spans four task families that form a controlled complexity gradient: single-aircraft NFZ avoidance, pairwise conflict resolution, joint conflict-airspace constraints, and multi-aircraft conflict graphs. Experiments with 11 representative LLM endpoints show that while most models produce format-valid instructions, task success varies widely across models and task families: the highest-scoring endpoint under our fixed protocol achieves 63.5% macro-average task success rate (TSR), while the two lowest endpoints fall below 25%, with the largest gaps appearing under joint constraints and multi-aircraft coordination. These results expose a persistent execution gap between static validity and simulation-passing behavior, motivating simulation-in-the-loop evaluation for LLM-based decision making in safety-critical aviation settings.
%R 10.18653/2026.findings-emnlp.1552
%U https://aclanthology.org/2026.findings-emnlp.1552/
%U https://doi.org/10.18653/2026.findings-emnlp.1552
%P 33032-33055
Markdown (Informal)
[ATC-Bench: A Simulation-Grounded Benchmark for Evaluating LLM-Generated Air Traffic Control Instructions](https://aclanthology.org/2026.findings-emnlp.1552/) (Cai et al., Findings 2026)
ACL