fastcampus_slam_codes

This repository contains code exercises for the following lecture series provided by @changh95 at FastCampus:

• 'Computer Vision, LiDAR processing, and Sensor Fusion for Autonomous Driving'
• 'SLAM Zero-to-Hero series for Physical AI and 3D Computer Vision'

Actively reworking the repository now. Stay tuned, because A LOT OF NEW TUTORIALS are on the way!

Zero-to-Hero SLAM lectures for Physical AI and 3D Computer Vision

The course can be found here.

The course content is essentially a superset of 'Computer Vision, LiDAR processing, and Sensor Fusion for Autonomous Driving', but with a more general focus within robotics, drones, AR/VR, autonomous driving.

Table of Contents

• Chapter 1: Introduction to SLAM
  • 1.1 Lecture introduction
  • 1.2 Mobile robotics
  • 1.3 What is SLAM?
  • 1.4 Hardware used in SLAM
  • 1.5 Types of SLAM
  • 1.6 Applications of SLAM
  • 1.7 Tips for studying SLAM
  • 1.8 C++ and SLAM
  • 1.9 Basic C++ programming
  • 1.10 Building C++ libraries
  • 1.11 C++ CPU profiler
  • 1.12 C++ memory profiler
  • 1.13 Python basics
  • 1.14 Basic Python programming
  • 1.15 PyBind
  • 1.16 ROS fundamentals
  • 1.17 Rotation and translation in 3D space
  • 1.18 Homogeneous coordinates
  • 1.19 Lie Group
  • 1.20 Basics of Lie algebra
  • 1.21 Eigen + Sophus library hands-on
  • 1.22 Continuous-time representation
  • 1.23 Camera basics for robotics
  • 1.24 Camera models
  • 1.25 LiDAR basics
  • 1.26 IMU basics
  • 1.27 Radar basics
  • 1.28 Forward/Inverse kinematics
  • 1.29 Sensor calibration
  • 1.30 Kalibr package hands-on
• Chapter 2: Dive into SLAM (Front-end)
  • 2.1 Part 2 introduction
  • 2.2 Local feature detection
  • 2.3 Classical local feature detection hands-on
  • 2.4 Deep local feature detection hands-on
  • 2.5 Feature tracking basics
  • 2.6 Advanced feature tracking in practice
  • 2.7 Feature tracking hands-on
  • 2.8 Global feature detection
  • 2.9 Global feature detection hands-on
  • 2.10 Deep global feature detection hands-on
  • 2.11 Epipolar geometry
  • 2.12 Epipolar geometry hands-on
  • 2.13 Homography
  • 2.14 Homography hands-on
  • 2.15 MonoVO hands-on
  • 2.16 Triangulation
  • 2.17 Triangulation hands-on
  • 2.18 Perspective-n-points
  • 2.19 Perspective-n-points hands-on
  • 2.20 RANSAC
  • 2.21 Advanced RANSAC
  • 2.22 RANSAC hands-on
  • 2.23 M-estimator & MAXCON
  • 2.24 What is point cloud?
  • 2.25 Introduction to PCL library
  • 2.26 Point cloud preprocessing
  • 2.27 Point cloud preprocessing hands-on
  • 2.28 ICP
  • 2.29 ICP hands-on
  • 2.30 Advanced ICP hands-on
  • 2.31 Octree, Octomap, Bonxai hands-on
• Chapter 3: Dive into SLAM (Back-end)
  • 3.1 Part 3 introduction
  • 3.2 Factor graph
  • 3.3 Nonlinear least squares
  • 3.4 Nonlinear optimization
  • 3.5 Optimization on manifolds
  • 3.6 Graph-based SLAM
  • 3.7 Schur complement
  • 3.8 Auto-diff
  • 3.9 Continuous-time optimization
  • 3.10 Sparsity in SLAM
  • 3.11 Bundle adjustment
  • 3.12 Nonlinear solvers
  • 3.13 g2o hands-on
  • 3.14 GTSAM hands-on
  • 3.15 Ceres-solver hands-on
  • 3.16 SymForce hands-on
  • 3.17 SLAM systems
  • 3.18 Various map representations
  • 3.19 VSLAM system architecture
  • 3.20 LiDAR SLAM system architecture
  • 3.21 RADAR SLAM system architecture
  • 3.22 Event SLAM system architecture
  • 3.23 Inertial odometry basics
  • 3.24 Leg odometry basics
  • 3.25 Sensor fusion
• Chapter 4: Classical SLAM
  • 4.1 Part 4 introduction
  • 4.2 Feature-based VSLAM
  • 4.3 Direct VSLAM
  • 4.4 Visual-inertial odometry
  • 4.5 2D LiDAR SLAM
  • 4.6 3D LiDAR SLAM
  • 4.7 Sensor fusion SLAM
  • 4.8 ORB-SLAM 2
  • 4.9 Basalt-VIO
  • 4.10 Cartographer
  • 4.11 KISS-SLAM
  • 4.12 GLIM
  • 4.13 FAST-LIO2
  • 4.14 FAST-LIVO2
• Chapter 5: Advanced SLAM - AI Integration and Hardware Optimization
  • 5.1 Part 5 introduction
  • 5.2 SLAM + Object detection + Segmentation
  • 5.3 SLAM + Depth estimation
  • 5.4 SLAM + Camera pose regression
  • 5.5 SLAM + Deep feature matching
  • 5.6 SLAM + Deep optical flow / scene flow
  • 5.7 SLAM + Differentiable bundle adjustment
  • 5.8 SLAM + Feed-forward 3D transformer
  • 5.9 SLAM + NeRF / Implicit neural field
  • 5.10 SLAM + Gaussian Splatting
  • 5.11 SLAM + Video generation
  • 5.12 SLAM + VLM/VLA
  • 5.13 SLAM + 3D Scene graph
  • 5.14 SLAM + Certifiably optimal algorithm
  • 5.15 SLAM + Auto-encoder / diffusion
  • 5.16 SLAM + Graph processor
  • 5.17 DSP-SLAM
  • 5.18 Kimera
  • 5.19 ConceptFusion
  • 5.20 Gaussian Splatting SLAM
  • 5.21 MASt3r-SLAM
  • 5.22 PIN-SLAM
  • 5.23 Suma++
  • 5.24 Differences between desktop, server, and embedded boards
  • 5.25 Characteristics of real-time SLAM
  • 5.26 Characteristics of auto-labeling / data-crunching SLAM
  • 5.27 C++ build configuration optimization
  • 5.28 SIMD acceleration and CPU optimization techniques
  • 5.29 SIMD acceleration hands-on
  • 5.30 Introduction to NVIDIA Jetson
  • 5.31 CUDA acceleration hands-on
• Final projects
  • Project 1: SLAM for autonomous driving
  • Project 2: SLAM for drones
  • Project 3: SLAM for mobile scanner systems
  • Project 4: SLAM for quadruped robots
  • Project 5: SLAM for humanoid robots
  • Project 6: SLAM for VR/AR headsets

Libraries in Base Docker Image

Computer Vision, LiDAR processing, and Sensor Fusion for Autonomous Driving

The course can be found here.

This course contains the following contents:

• Chapter 1: Introduction to SLAM
  • 1.1 Lecture introduction
  • 1.2 What is SLAM?
  • 1.3 Hardware for SLAM
  • 1.4 Types of SLAM
  • 1.5 Applications of SLAM
  • 1.6 Before we begin...
  • 1.7 Basic C++ / CMake
• Chapter 2: Introduction 3D Spaces
  • 2.1 3D rotation and translation
  • 2.2 3D rotation and translation, using Eigen library
  • 2.3 Homogeneous coordinates
  • 2.4 Lie Group
  • 2.5 Basic Lie algebra
  • 2.6 Lie Group and Lie algebra, using Sophus library
  • 2.7 How cameras work
  • 2.8 How LiDARs work
• Chapter 3: Image processing
  • 3.1 Local feature extraction & matching
  • 3.2 Local feature extraction & matching, using OpenCV library
  • 3.3 Superpoint and Superglue, using C++ and TensorRT
  • 3.4 Global feature extraction
  • 3.5 Bag of Visual Words, using DBoW2 library
  • 3.6 Learning-based global feature extraction, using PyTorch and Tensorflow libraries
  • 3.7 Feature tracking
  • 3.8 Optical flow, using OpenCV library
• Chapter 4: Point cloud processing
  • 4.1 Introduction to point cloud processing
  • 4.2 Introduction to point cloud processing, using PCL library
  • 4.3 Point cloud pre-processing
  • 4.4 Point cloud pre-processing, using PCL library
  • 4.5 Iterative closest point
  • 4.6 Iterative closest point, using PCL library
  • 4.7 Advanced ICP methods
  • 4.8 Advanced ICP methods (G-ICP, NDT, TEASER++, KISS-ICP), using PCL library
  • 4.9 Octree, Octomap, Bonxai, using PCL/Octomap/Bonxai libraries
• Chapter 5: Multiple view geometry
  • 5.1 Epipolar geometry
  • 5.2 Essential and Fundamental matrix estimation, using OpenCV library
  • 5.3 Homography
  • 5.4 Bird's eye view (BEV) projection, using OpenCV library
  • 5.5 Simple monocular visual odometry, using OpenCV library
  • 5.6 Triangulation
  • 5.7 Triangulation, using OpenCV library
  • 5.8 Perspective-n-Points (PnP) and Direct Linear Transform (DLT)
  • 5.9 Fiducial marker tracking, using OpenCV library
  • 5.10 RANSAC
  • 5.11 Advanced RANSAC methods (USAC)
  • 5.12 RANSAC and USAC, using OpenCV and RansacLib libraries
  • 5.13 Graph-based SLAM
  • 5.14 Least squares
  • 5.15 Schur complement
  • 5.16 Bundle adjustment
  • 5.17 Bundle adjustment, using Ceres-Solver library
• Chapter 6: Visual-SLAM
  • 6.1 Overview of feature-based VSLAM
  • 6.2 Overview of direct VSLAM
  • 6.3 Overview of visual-inertial odometry (VIO)
  • 6.4 Spatial AI
  • 6.5 ORB-SLAM2, ORB-SLAM3
  • 6.6 DynaVINS
  • 6.7 CubeSLAM
• Chapter 7: LiDAR SLAM
  • 7.1 Overview of 2D LiDAR SLAM
  • 7.2 Overview of 3D LiDAR SLAM and LiDAR-inertial odometry
  • 7.3 HDL-Graph-SLAM
  • 7.4 KISS-ICP
  • 7.5 SHINE-Mapping
• Chapter 8: CI/CD for SLAM
  • 8.1 TDD and tests
  • 8.2 CI/CD
  • 8.3 CI agents
  • 8.4 CI/CD for Python SLAM projects
  • 8.5 CI/CD for C++ SLAM projects
• Final projects:
  • DSP-SLAM
  • Suma++
  • Cartographer-KITTI