Nathan Barranco-Janniaux

In this position, I have contributed to the definition and the architecture of a software framework acting as a cloud inside the train. The goal of the framework is to provide application hosting and also network connection with the train. Here are some key points on which I have contributed:

In this role, I had the privilege of contributing to the cutting-edge autonomous train project, focusing on developing the Perception Environment, including the Signaling Detection System and Obstacle Detection System. :

•&: Authored the comprehensive system specification for a complex test bench, encompassing need definition and the design of system and technical architecture tailored to these needs.

•: Utilized decision matrices and Proof of Concept (PoC) validations to address and eliminate technical uncertainties, ensuring system reliability and efficiency.

•:

- Automated creation of virtual test scenarios using a 3D engine, combined with field data from prototype locomotives to generate "recorded" scenarios.

- System deception testing through the execution of scenarios in SiL (Software in the Loop), TiL (Target in the Loop), and HiL (Hardware in the Loop) environments.

- Automated generation of test reports for performance evaluation of the system under test. &:

• & : Docker, Kubernetes, Azure Cloud

• : GigEVision, GeniCam, ROS, millisecond-level data stream synchronization

• & : Python, C++, HTML, CSS, JavaScript, Vue JS, Java

• : Efficient handling and storage of over 300TB of data, leveraging robust networking and database solutions. :

• Provided technical support to multidisciplinary teams involved in the test bench development.

• Contributed to the definition of the validation strategy following the V-Cycle process to ensure quality assurance and compliance with project standards.

- Optimisation de l'autonomie d'un boitier contenant une Raspberry Pi permettant de faire de la captation vidéo sur mouvement.

- Amélioration d'un logiciel de traitement d'image via l'intelligence artificielle.

- Réseaux de neurones à convolution

- Comptage de véhicules sur vidéo via machine learning (YOLO, DEEP SORT)