About Me
I deployed the full autonomous navigation stack for EPFL Xplore's Mars rover solo: a LiDAR-inertial SLAM pipeline, 9-axis IMU integration, wheel odometry, and a custom EKF, all running reliably in GPS-denied outdoor terrain. The rover won the European Rover Challenge 2025.
I am now building a custom 3D SLAM system from scratch in C++ using GTSAM factor-graph optimization — LiDAR-inertial fusion, place recognition, and loop closure. I have also tuned production-grade systems (LIO-SAM, FAST-LIO2, GLIM) on real hardware, implemented EKFs and UKFs across multiple platforms, and built a custom 6-layer AHRS PCB with full firmware running the VqF attitude filter.
My focus: robust localization and state estimation for robots operating in GPS-denied, unstructured environments.
Projects
Featured — SLAM & State Estimation
★ ERC 2025 Winner
LiDAR-Inertial SLAM — Mars Rover Navigation Stack
EPFL Xplore | Team Leader Autonomous Navigation | Sep 2024 — Aug 2025
Deployed the full autonomous navigation stack solo for a 4-wheeled Mars rover competing in the European Rover Challenge 2025 — GPS-denied outdoor terrain, no fallback.
SLAM & Localization:
- LiDAR-inertial SLAM pipeline (Ouster 3D LiDAR + 9-axis IMU)
- Custom Extended Kalman Filter fusing wheel odometry, IMU, and LiDAR-inertial odometry
- Global pose corrections via triangulation and trilateration with convex optimization (CVXPY/ECOS)
- Sub-15 cm accuracy in GPS-denied outdoor environments
- Production SLAM systems tuned on hardware: LIO-SAM, FAST-LIO2, GLIM
Planning & Control:
- Nav2 stack with Hybrid A* global planner
- Pure Pursuit path tracking with double-Ackermann kinematics
- Dynamic obstacle avoidance
Technologies:
C++ Python ROS2 OpenCV Docker Gazebo Arduino maxon EPOS
★ Ongoing
Custom 3D SLAM System — GTSAM Factor Graphs
Personal Project | Fall 2025 — Ongoing
Building a complete LiDAR SLAM system from scratch in C++ using GTSAM factor-graph optimization as the back-end. No black-box libraries — every component designed and implemented from the ground up.
Architecture:
- Front-end: LiDAR scan-to-map matching for incremental odometry; IMU preintegration for inter-frame constraint generation
- Back-end: GTSAM incremental smoothing (iSAM2) for real-time factor graph optimization
- Place Recognition: descriptor-based loop closure detection
- Back-end trigger: pose-graph optimization on detected loop closures
Benchmarked against production systems (LIO-SAM, FAST-LIO2, GLIM) on the same hardware and datasets.
Technologies:
C++ GTSAM ROS2 PCL Eigen LiDAR
★ Hardware + Firmware
Custom 6-Layer AHRS PCB & Drone Visual-Inertial Odometry
Personal Project | Fall 2025 — Ongoing
AHRS Hardware: Designed and built a custom 6-layer PCB in KiCad implementing a full AHRS (Attitude and Heading Reference System). Firmware runs the VqF (Versatile Quaternion-based Filter) algorithm — a state-of-the-art sensor fusion filter for robust attitude estimation from 9-axis IMU data (accelerometer + gyroscope + magnetometer).
Visual-Inertial Odometry: The AHRS PCB serves as the IMU unit for an FPV drone VIO pipeline — camera + IMU tight coupling for GPS-denied state estimation. Implements EKF-based fusion of visual feature tracks and inertial measurements.
Technologies:
KiCad C STM32 VqF OpenCV EKF/UKF ROS2
Other Projects
EPFL Xplore — Current Rover (Software Systems Engineer)
Sep 2025 — Ongoing
Leading end-to-end software architecture for the 2025/26 competition rover: autonomous navigation (ROS2/Nav2), robotic arm control (MoveIt), real-time wireless communication, and sensor fusion (IMU, LiDAR, cameras). Coordinating hardware-software integration across a multidisciplinary team on NVIDIA Jetson platforms.
Technologies:
C++ Python ROS2 MoveIt Docker maxon EPOS
STM32 RTOS Autonomous Mobile Robot
Academic Project | Spring 2025
Real-time autonomous navigation on the e-puck 2 platform using ChibiOS RTOS. Extended Kalman Filter for localization, real-time obstacle detection and mapping, efficient RTOS task scheduling, sensor fusion with IMU and proximity sensors.
Technologies:
STM32 ChibiOS C EKF Embedded Systems
Thymio Autonomous Mobile Robot
Academic Project | Fall 2025
Autonomous navigation with EKF localization and ArUco tag triangulation & trilateration using convex optimization (CVXPY/ECOS). Global path planning and real-time obstacle detection.
Technologies:
Python EKF OpenCV CVXPY
Solar Tracking Solar Oven
Personal Project | Fall 2025 — Ongoing
2-DoF sun-tracking system with custom DC-DC Buck converter PCB, ESP32 on FreeRTOS, lux sensor array, PID motor control, and mechanical design in Fusion 360.
Technologies:
ESP32 FreeRTOS KiCad Fusion 360 PID
Direction of Arrival — LibreSDR
Personal Project | Fall 2025
MUSIC and Root-MUSIC algorithm implementations for direction-of-arrival estimation on the LibreSDR platform.
Technologies:
Python SDR Signal Processing
Technical Skills
Experience
EPFL Xplore — Software Systems Engineer
Leading end-to-end software architecture of the 2025/26 competition rover: autonomous navigation (ROS2/Nav2), robotic arm control (MoveIt), and real-time wireless communication. Coordinating perception, planning, and control integration across a multidisciplinary team on Jetson platforms. Containerized deployment with Docker; sensor fusion across IMU, LiDAR, and cameras.
C++ Python ROS2 Docker Jetson maxon EPOS
EPFL Xplore — Team Leader, Autonomous Navigation
Led the development of the full navigation subsystem for the ERC 2025 rover. Solo deployment of the complete SLAM and navigation stack — LiDAR-inertial odometry, 9-axis IMU integration, wheel odometry, and custom EKF, running in GPS-denied outdoor terrain. Won 1st place at the European Rover Challenge 2025.
C++ Python ROS2 OpenCV Docker Gazebo Arduino
EPFL Xplore — Software Engineer
Designed and implemented ROS2-based manual and autonomous navigation for an outdoor rover. 2D LiDAR SLAM integration, low-level PID motor controller on Arduino with custom wheel odometry. Collaborated with mechanical and electrical engineers on system integration.
C++ Python ROS2 Arduino
ETML — Machining Intern
Hands-on manual metal machining: turning, milling, drilling, sawing, tapping, and brazing.
Education
EPFL — Master of Science in Robotics
2025 — Ongoing
Relevant Coursework:
- Manipulation & Computer Vision
- Autonomous Navigation
- Sensor Fusion and State Estimation
- Machine Learning & Convex Optimization
- Multivariable and Non-Linear Control, Model Predictive Control
EPFL — Bachelor of Science in Microengineering
GPA: 5.38 / 6 | 2022 — 2025
Relevant Coursework:
- Electronics I & II, OOP, Digital System Design
- AVR Microcontrollers & Embedded Systems
- Control Systems, Signals and Systems
- Real and Complex Analysis, Linear Algebra
- Introduction to PCB Design & Manufacturing
École Européenne Luxembourg II
Baccalauréat Scientifique | 95.02 / 100 | 2022
- Secretary of the BAC Committee · Yearbook Committee
Awards
European Rover Challenge 2025 — 1st Place
Led the autonomous navigation subsystem (SLAM, localization, path planning) that earned EPFL Xplore first place in the international Mars rover competition.
Luxembourg Informatics Olympiad — Semi-Finalist
Applied optimization and path-finding algorithms in competitive programming.
Languages
Native
Fluent — TOEFL 112/120
Basic
Basic
Interests
Detailed Project Showcases
ERC 2025 — LiDAR-Inertial SLAM Navigation Stack
Problem
Design and deploy a complete autonomous navigation system for a Mars rover operating in GPS-denied, rough outdoor terrain — with no fallback localization source.
Localization & State Estimation
Sensor suite: Ouster 3D LiDAR + 9-axis IMU (accelerometer, gyroscope, magnetometer) + wheel encoders
- LiDAR-inertial odometry as the primary odometry source — high-frequency, drift-bounded
- Custom EKF fusing wheel odometry, IMU, and LiDAR-inertial poses; adaptive noise covariance tuning for outdoor terrain
- Double-Ackermann kinematics model for accurate motion prediction during tight turns
- Triangulation + trilateration for absolute pose correction using visual landmarks — solved as a Second-Order Cone Program (SOCP) with CVXPY + ECOS
Production SLAM systems tuned on hardware: LIO-SAM, FAST-LIO2, GLIM — used as references and for benchmarking the custom EKF stack.
Perception & Planning
- 3D Ouster LiDAR for obstacle detection and costmap generation
- Camera-based landmark detection (OpenCV)
- Nav2 Hybrid A* global planner + Pure Pursuit path tracking
- Dynamic obstacle avoidance with local planner
Results
| Metric | Result |
|---|---|
| Competition | European Rover Challenge 2025 — 1st Place |
| Localization accuracy | Sub 15 cm in GPS-denied outdoor terrain |
| Deployment | Solo, full stack on NVIDIA Jetson in Docker |
Landmark-based global pose correction: solved as a Second-Order Cone Program (SOCP) using CVXPY and the ECOS solver.
Contact
Open to internship opportunities, research collaborations, and robotics projects — particularly in SLAM, state estimation, and autonomous systems.
📧 Email: arno.laurie@epfl.ch
🔗 LinkedIn: linkedin.com/in/arno-laurie
💻 GitHub: github.com/Netfluxx
📍 Location: Lausanne, Switzerland
Last updated: June 2026