This site has different types of tech blogs, including AR, CV, SLAM, Math, Coding and software developments. Each blog will be updated from time to time.

Index of main posts are as following.

• SLAM
  • Basis
    • General Math
    • State Estimation
    • Photogrammetric Computer Vision
    • Multi-View Geometry in Computer Vision
  • Theory
    • Visual SLAM
    • Visual Inertial SLAM
  • Architecture Survey
    • Visual SLAM Survey
    • Visual Inertial SLAM Survey
• AR
  • Introduction
  • Microsoft Mixed Reality
    • HoloLens Concepts
    • Development Path
    • Windows App development
• C++
  • Introduction
  • Grammar
• Data Structure and Algorithm
  • Summary of common questions
  • Summary of typical questions

More posts could be found in categories tab.

Multilayer Neural Network

• Inherit tf.keras.Model can built own neural network
  • Every neural have several weights and a bias, and can handle a full input
• Use tf.keras.Sequential()

Convolutional Neural Networks (CNN)

• Can consider the relationship among several pixels
• Special Layers:
  • Convolutional Layer
    • Won’t change the scale of the input if the stride is 1 * 1
  • Pooling Layer
    • Will generates more features but won’t increase the calculation

Recurrent Neural Networks (RNN)

• Works well in the time-based classification
• Make sense of what they’ve seen before, both short and long-term memory
• Scenarios:
  • Video Processing
  • Speech Processing
  • Sentiment Analysis
  • Time series Data
• Key Implements
  • What long term memory should be kept or removed
  • What new memory should be remembered as the long term memory
  • What new memory should be remembered as the short term memory

Literature review for SLAM papers.

Majority content of this post came from the book Basic Knowledge on Visual SLAM: From Theory to Practice.

SLAM (Simultaneous Localization and Mapping) is the technique to establish the nearby environment and localize the moving object inside it at the same time according to sensor measurements. SLAM can be used in augmented reality, autonomous vehicles, and robot’s navigation. The mathematic theory of SLAM is the state estimation.

Visual SLAM is a kind of SLAM that depends on on visual sensors (cameras), which also involves photometric and geometric computer vision.

The SLAM algorithm based on traditional computer vision geometry and state estimation has been mature. There are lots of open source SLAM architecture, and different commercial products have their own implementation. Every SLAM implementation has its pros and cons. This post is to provide the high level guidance for how to design a practical SLAM system for products.

Talks about the future development directions of SLAM.

Notes of Operation System course (2021 Autumn, Tsinghua University)

General introduction of C++ related knowledge.

This post talks about the fundamental math for SLAM, including Homogeneous Coordinates, Transformation in 3D space, Probability, SVD, Least Square Optimization, Lie Group and Lie Algebra and etc.

This post talks about the theory for Visual Inertial SLAM. Notes of book SLAM in Autonomous Driving.