Lucas Carpentier

• - Software development for autonomous drones in a highly reliable environment (defense sector): C++ design of perception, navigation, and control modules under ROS2.

• - Multi-sensor data fusion (sonar, radar, IMU) for perception and localization in complex environments.

• - Deep learning pattern recognition: design and integration of detection models applied to sensor data.

• - Simulation and validation: development of test scenarios in a simulated environment before integration on the real target.

• - Embedded real-time systems: implementation and optimization of critical algorithms, integration on target, and validation in real conditions.

• - Real-time classification of streams on a network gateway: integration under OpenWRT in C++ of a machine learning model to classify various network streams (voice, streaming, IoT data, etc.), initial tests in simulation then deployment of the firmware on the gateway

• - Development of C++ modules for data extraction and preprocessing from embedded systems.

• - Software lifecycle management on GitLab: code review, unit and integration tests, continuous integration and deployment (CI/CD) pipelines.

• - Integration of the cuRobo framework under ROS 2 for Doosan M1013 robot: trajectory planning optimized by neural networks; C++/Python modules for control logic and autonomous behavior; obstacle avoidance with Intel RealSense D405 camera and NVIDIA Nvblox.

• - Simulation and validation scenarios under NVIDIA Isaac Sim with GPU optimization via CUDA. Tests in a virtual environment before deployment on the real robot.

– Integration of the cuRobo framework under ROS 2 for Doosan M1013 robot: trajectory planning optimized by neural networks; C++/Python modules for control logic and autonomous behavior; obstacle avoidance with Intel RealSense D405 camera and NVIDIA Nvblox.

– Simulation and validation scenarios under NVIDIA Isaac Sim with GPU optimization via CUDA. Tests in a virtual environment before deployment on the real robot.