Related Concept Videos
Language
Corballis and Suddendorf (2007) and Tomasello and Rakoczy (2003) highlight the role of language in...
Components of Language
Language Development
The critical period for language acquisition suggests that the ability to acquire language is at its peak early in life. As people age, this proficiency decreases. Language development begins very...
Language and Cognition
Higher Mental Functions of the Brain: Language
Language formation and comprehension take place in the dominant hemisphere. The dominant hemisphere is responsible for understanding the meaning of spoken, written, or sign language, as well as the ability to communicate. For most people, the left hemisphere is the dominant one. The right hemisphere, then, gives tone and emotional context to the...
Self-Awareness and Its Effects
You might also read
Related Articles
Articles linked to this work by shared authors, journal, and citation graph.
Enhancing Unsupervised Multi-Source Domain Adaptation for Person Re-Identification via Mixture of Experts and Graph-Based Relation.
Development of an Instrumented Glove for Palmar Pressure Assessment in Kayakers.
Related Experiment Video
Updated: Jan 29, 2026

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
Published on: December 6, 2024
Latency-Aware Benchmarking of Large Language Models for Natural-Language Robot Navigation in ROS 2.
Murat Das1, Zawar Hussain1,2, Muhammad Nawaz1,3
1Sydney Polytechnic Institute, Sydney, NSW 2000, Australia.
This study introduces a new framework for natural language robot navigation using Large Language Models (LLMs) and Robot Operating System 2 (ROS 2). It benchmarks LLMs and planners, revealing trade-offs between response speed and navigation accuracy.
More Related Videos
Area of Science:
- Robotics and Artificial Intelligence
- Human-Robot Interaction
- Natural Language Processing
Background:
- Mobile robotics often relies on complex interfaces, hindering accessibility for non-expert users.
- Integrating Large Language Models (LLMs) offers a path towards more intuitive, conversational robot control.
- Existing systems lack comprehensive benchmarking for LLM-driven navigation, especially considering latency.
Purpose of the Study:
- To develop and evaluate a latency-aware benchmarking framework for natural-language robot navigation.
- To assess the performance of various contemporary LLMs and local planners within the ROS 2 Navigation 2 stack.
- To provide insights into the trade-offs between LLM size, response latency, and navigation accuracy.
Main Methods:
- Developed a benchmarking framework integrating multiple LLMs with the ROS 2 Nav2 stack.
- Utilized a simulated TurtleBot4 platform in Gazebo Fortress for standardized indoor navigation scenarios.
- Benchmarked LLMs (e.g., GPT-4, Gemini 2.5, LLaMA-3.3) with local planners (DWB, TEB, RPP), measuring latency, accuracy, path quality, and success rate.
Main Results:
- Demonstrated clear trade-offs between LLM size, response latency, and navigation accuracy.
- Smaller LLMs offered faster responses but weaker spatial reasoning; larger LLMs showed better navigation intent but higher latency.
- Identified specific performance characteristics of different LLM-planner combinations in standardized tasks.
Conclusions:
- The proposed framework enables reproducible, multi-LLM, multi-planner evaluations for natural-language robot navigation within ROS 2.
- The findings support the development of intuitive, latency-efficient conversational interfaces for mobile robots.
- This work lays the groundwork for more accessible and adaptable robot control systems.

