changh95/slam_lecture_codes

Code exercises for the SLAM courses

What it solves

Learning SLAM (Simultaneous Localization and Mapping) is hard because it combines C++, computer vision, point cloud processing, optimization, and sensor fusion. This repository gives learners a set of concrete, runnable code exercises that follow along with two FastCampus lecture series, so they can practice the concepts instead of just watching videos.

How it works

The repository is organized into two main course folders. The first is a comprehensive zero-to-hero SLAM course that goes from C++ and Python basics, through 3D geometry, then into SLAM front-end (feature detection, tracking, point clouds, ICP) and back-end (graph optimization, bundle adjustment), and ends with advanced AI-integrated SLAM and hardware optimization. The second course focuses on autonomous driving, covering similar topics but with a stronger emphasis on computer vision, LiDAR processing, sensor fusion, and final projects like ORB-SLAM and Cartographer. Each chapter has theory sections plus hands-on code examples that use libraries like OpenCV, Eigen, Ceres, PCL, and ROS. The repository also provides a base Docker image with many of these libraries pre-installed.

Who it’s for

Students, developers, and researchers who want to learn SLAM through practical coding. It’s especially useful for people following the FastCampus lectures, but anyone with basic programming knowledge can use the exercises to build SLAM skills. The content spans from beginner-friendly C++/Python to advanced topics like GPU acceleration and modern SLAM systems.

Highlights

  • Two complete course tracks: a full zero-to-hero SLAM series and a dedicated autonomous driving series
  • Over 60 hands-on code exercises covering all major SLAM topics
  • Uses real-world libraries: OpenCV, Eigen, Sophus, Ceres, g2o, GTSAM, PCL, and more
  • Includes a base Docker image with all major libraries already set up
  • Covers both classical and modern SLAM, including AI-integrated methods (NeRF, Gaussian Splatting) and hardware acceleration (CUDA, SIMD)
  • Final projects in scenarios like autonomous driving, drones, robots, and AR/VR

Related

  • Project
    SS47816/lidar_obstacle_detector

    A real‑time, CPU‑only 3D LiDAR obstacle detector and tracker for ROS 1/2, with ground segmentation, box fitting, multi‑object tracking, and occupancy grids. Optional GPU mode adds neural‑network labels. MIT licensed.

  • Project
    vmayoral/robot_hacking_manual

    A comprehensive introductory series and manual on robot cybersecurity, providing case studies and tutorials to highlight vulnerabilities and promote security-first robot development.

  • Project
    hwan0806/Awesome-LiDAR-Mapping

    A curated index of peer-reviewed LiDAR mapping papers that provide open-source code, covering odometry, inertial fusion, dynamic object removal, and map merging to help researchers and engineers quickly find usable implementations.

  • Project
    Xbotics-Embodied-AI-club/Xbotics-Embodied-AI-Job

    A community-run, free referral list for embodied AI (robotics) jobs, aggregating long-term and rolling openings from companies, labs, and universities, plus learning resources and career support.