Benchmarking data logging strategies in ROS-integrated multi-sensor robots under network constraints

Saadi, Nour Elhouda Ben and Istenes, Zoltán (2025) Benchmarking data logging strategies in ROS-integrated multi-sensor robots under network constraints. In: Proceedings of the International Conference on Formal Methods and Foundations of Artificial Intelligence. Eszterházy Károly Katolikus Egyetem Líceum Kiadó, Eger, pp. 188-200. ISBN 9789634963035

Preview

Text fmfai2025_pp188-200.pdf - Published Version Download (7MB) | Preview

Abstract

This work benchmarks multiple logging strategies, including Java- Script Object Notation (JSON) serialization, JSON Lines (JSONL), Web- Socket based rosbridge, native Robot Operating System (ROS) Transmission Control Protocol (TCP)/User Datagram Protocol (UDP) transports, and NFS-based recording, on a multisensor robot over a 100 Mbps constrained network. We evaluate dropped messages, resource usage, and storage footprint across high-throughput data streams, such as LiDAR, Red-Green-Blue plus Depth (RGB-D) images, Inertial Measurement Unit (IMU), and Global Navigation Satellite System (GNSS) data. Results show that TCPROS provides the highest reliability, while UDPROS, rosbridge, and JSON-based methods incur significant losses when network capacity is saturated. Network File System (NFS) recording performs well, provided that network stability is maintained. Our findings reveal key trade-offs between transport guarantees, network limitations, and message size fragmentation.

Actions (login required)

Edit Item