Both of Us: Development of a Multi Featured Service Robot

• Luize Cunha Duarte UFPR https://orcid.org/0009-0009-6471-2393
• Luan Matheus Trindade Dalmazo UFPR https://orcid.org/0000-0002-1589-6482
• Gabriel Pontarolo UFPR https://orcid.org/0009-0001-6416-4333
• Felipe Gustavo Bombardelli UFPR https://orcid.org/0009-0004-5762-0849
• Eduardo Todt UFPR https://orcid.org/0000-0001-6045-1274

Resumo

Service robots require the integration of multiple technologies such as Simultaneous Localization and Mapping (SLAM), computer vision, and human-computer interaction to enable effective development. This paper presents the development of a modular, low-cost service robot designed for educational and research purposes. The robot integrates a range of hardware components including RGB-D cameras, LiDAR, and a mobile base with a software stack based on the Robot Operating System 2 (ROS 2). A feature is the implementation of a voice-controlled virtual assistant powered by Large Language Models (LLMs), enabling natural human-robot interaction. The system architecture emphasizes modularity, accessibility, and real-world applicability, supporting experiments in SLAM, computer vision, and multimodal interaction. The robot's construction combines off-the-shelf components, 3D-printed parts, and open-source tools, making it an effective learning platform for robotics and AI. The paper also discusses the technical and cost challenges faced during development, and outlines future improvements including enhanced sensor fusion and SLAM benchmarking.

Palavras-chave: Computer vision, Simultaneous localization and mapping, Service robots, Large language models, Virtual assistants, Operating systems, Education, Robot vision systems, Systems architecture, Sensor fusion, Service Robots, Large Language Models