Introducing CPU load Analysis from AADL Models for ROS applications : a use case - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2021

Introducing CPU load Analysis from AADL Models for ROS applications : a use case

Résumé

Robotic software can exhibit low performances even while running on allegedly powerful multi-core processor plat- forms. To find why is essential to guarantee the success of a mission. This paper presents an approach to analyze the processing resource demand of a ROS (Robotic Operating System) based robotic application. The AADL (Architecture Analysis and Design Language) language is used to model the application software, the hardware, and the deployment of the software components onto the hardware. The analysis of processor loads, and the checking of MIPS demand against MIPS capacity, is done with the OSATE2 tool, thanks to a few properties of the language. To set up those properties, a rapid profiling of the hardware and of the software components of the application is done, based on common Linux performance counting tools. The final accuracy is good enough to allow for a fast verification of the deployment options, and to help thinking or rethinking the software vs hardware architectures.
Fichier non déposé

Dates et versions

hal-03818578 , version 1 (18-10-2022)

Identifiants

Citer

Eric Senn, Lucie Bourdon. Introducing CPU load Analysis from AADL Models for ROS applications : a use case. 2021 Forum on specification & Design Languages (FDL), Sep 2021, Antibes, France. pp.01-08, ⟨10.1109/FDL53530.2021.9568386⟩. ⟨hal-03818578⟩
20 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More