Benchmarking Quantized Edge LLMs for Natural Language Mobile Robot Control
Applied Sciences · 20.07.2026 · 16 (14) · 7267
DOI: 10.3390/app16147267 ↗Kurz gefasst
Fachartikel in der Zeitschrift Applied Sciences, online erschienen am 20.07.2026. Verfasst von Burak Aggul und 2 weiteren Autor:innen. Die Arbeit nennt 35 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Istanbul Topkapi University.
Abstract (Originalsprache)
Deploying large language models on limited edge hardware remains an open challenge for natural language robot control. This study compares three quantized sub-1.5B models, namely SmolLM2-135M, Gemma3-1B, and TinyLlama-1.1B, for velocity-command generation on a TurtleBot3 Burger in ROS/Gazebo using llama.cpp. The evaluation uses a two-tier protocol: an atomic benchmark with 900 inference trials (50×6×3) measuring parse success, semantic accuracy, and latency, and a mission benchmark with 150 multi-step navigation runs and 870 total steps measuring waypoint, heading, and endpoint error from Gazebo ground-truth pose. A clear model-load cold-start effect is observed: the first warmup inference after loading is roughly 6–9× slower than steady-state warm-start inference because of KV-cache initialization and memory-mapped weight paging. Warm-start latencies differ significantly across models (Kruskal–Wallis H=682.3; p
Förderung
- Istanbul Topkapi University
Zitieren
APA
Aggul, B., Erdemir, G., & Arik, K. (2026). Benchmarking Quantized Edge LLMs for Natural Language Mobile Robot Control. Applied Sciences, 16(14), 7267. https://doi.org/10.3390/app16147267
Vancouver
Aggul B, Erdemir G, Arik K. Benchmarking Quantized Edge LLMs for Natural Language Mobile Robot Control. Applied Sciences. 2026;16(14):7267. doi:10.3390/app16147267
BibTeX
@article{Aggul2026,
author = {Aggul, Burak and Erdemir, Gökhan and Arik, Kaan},
title = {Benchmarking Quantized Edge LLMs for Natural Language Mobile Robot Control},
journal = {Applied Sciences},
year = {2026},
volume = {16},
number = {14},
pages = {7267},
doi = {10.3390/app16147267}
}
EUVIMED