I’m Rassil MAHJOUB

2D tower defense game developed in C++ with an object-oriented design: enemies move along the paths of a tile-based map, and towers placed along the way try to stop them.

Key points

• Tile-based 2D map with several paths followed by the enemies.
• Game entities (map, towers, enemies) modeled as classes, following object-oriented principles such as encapsulation and inheritance.
• Game loop handling enemy movement, tower attacks and rendering.

Technologies: C++ · Object-oriented programming

Three-person team project: an embedded industrial supervision system that collects sensor data in real time, stores it, displays it in a web interface and detects anomalies, as a lightweight alternative to traditional SCADA systems.

Key points

• Raspberry Pi 4 as the central unit (database, web interface, network) and an Arduino UNO for real-time acquisition over a USB serial link: temperature, humidity, light, sound and water level, plus LEDs, a buzzer and a servomotor.
• ROS 2 Humble architecture connected to the web interface through rosbridge and WebSockets; measurements stored in SQLite, with real-time charts and threshold-based anomaly detection.
• Secured remote access: SSH keys, SFTP, VNC, and HTTPS with TLS through an Nginx reverse proxy.
• My role: user interface development, ROS integration and writing of the final documentation.

Technologies: Raspberry Pi 4 · Arduino UNO · ROS 2 Humble · rosbridge · WebSocket · Python · JavaScript · SQLite · Nginx

Team course project: design, validation and critical analysis of a complete stereo vision pipeline that reconstructs 3D information from two real USB cameras.

Key points

• Stereo calibration with a chessboard (25 image pairs, reprojection error below 0.6 px per camera) and epipolar rectification.
• Image preprocessing (CLAHE contrast enhancement, bilateral filtering) and disparity estimation with OpenCV StereoSGBM, with optional WLS filtering.
• Conversion to a depth map and a colored 3D point cloud, with a comparison of the results with and without rectification.

Technologies: Python · OpenCV · Stereo calibration · Point clouds

Engineering project carried out over my three-year Robotics & Instrumentation program at Polytech Dijon: an autonomous mobile robot that disinfects hospital rooms by combining UV-C irradiation with conventional spraying. It started as a TurtleBot2 prototype and has grown into a dedicated robot designed with industrialization in mind, developed and tested in simulation under ROS 2 Jazzy and Gazebo Harmonic.

Dedicated robot (current version)

• Complete CAD design (650 × 620 × 1,960 mm, about 94.5 kg estimated with a full tank) exported to the URDF/Xacro model; automated tests keep the CAD and the simulation geometry consistent.
• Disinfection payload: 8 low-pressure UV-C lamps (720 W electrical) sized for at least 25 mJ/cm² at 2 m, and a 4-nozzle spraying system with a 15 L tank.
• Sensors and computing: Livox Mid-360 3D lidar on top of the mast, SICK safety laser scanner, two RGB-D cameras and an IMU, with an x86 main computer and an NVIDIA Jetson Orin Nano.
• Simulated hospital ward in Gazebo Harmonic (8 rooms, corridor, nursing station, charging station): 360° scan built from the Livox point cloud, SLAM with slam_toolbox, Nav2 navigation and autonomous exploration.
• Engineering file: specifications, bill of materials, power budget and battery autonomy (3 h target), stability study, electrical schematics, initial risk analysis, EU compliance file and verification & validation plan.

Operator web interface

• Robot-agnostic web interface (one YAML profile per robot) shared by the prototype, its simulation and the final robot.
• Real-time 2D and 3D map with lidar scans and point cloud, camera feed, manual driving and emergency stop.
• Programs (mapping, disinfection of a zone or of the whole map, rounds, docking), scheduling, history, reports and a supervision screen; commands protected by an operator token.

TurtleBot2 prototype (first iteration)

• Kobuki base, LDS-01 lidar and Orbbec Astra camera: SLAM (Cartographer, RTAB-Map), Nav2, frontier exploration, battery-triggered docking and a first UV-C coverage map.
• Ported from Gazebo Classic to Gazebo Harmonic, and run on the physical robot on a Jetson with ROS 2 Jazzy Docker images.
• Stopped because the required payload (UV-C mast, 15 L tank, batteries) exceeded what the TurtleBot2 chassis can carry; its lessons shaped the dedicated robot.

Technologies: ROS 2 Jazzy · Gazebo Harmonic · Nav2 · slam_toolbox · RTAB-Map · C++ · Python · URDF/Xacro · CAD · Docker · NVIDIA Jetson

Personal project focused on building a modern, generic web platform for connecting to, visualizing, diagnosing and operating ROS 2 robotic systems in real time.

Main features

• Interactive visualization of the ROS 2 graph, TF tree, topics, services, actions, parameters, logs and diagnostics.
• Live monitoring of cameras, maps, laser scans, point clouds and articulated robot models.
• Navigation tools, alerts, multi-robot fleet supervision, data analysis, export and synchronized rosbag recording and replay.
• Optimized real-time data transport and role-based security for reliable local or remote operation.

Technologies: ROS 2 · Python · FastAPI · WebSockets · React · TypeScript · Three.js · Docker