## Dark blue vision SLAM lesson 5 homework

1. Content
The operation has been completed. Here are some codes to share.
2. T2
1. ORB extraction
void computeAngle(const cv::Mat &image, vector<cv::KeyPoint> &keypoints) {
int half_patch_size = 8;
// int half_boundry = 16;
int bad_points =0; //A point in a corner where an angle cannot be calculated
for (auto & ...

Added by **balistic** on *Thu, 10 Mar 2022 06:10:16 +0200*

## [step through the pit for you] 04 + ROS noetic + opencv3 runs vins fusion successfully

In Ubuntu 20 04 installing ros and running vins fusion encountered many problems and stepped on many pits. A summary is sent here. The installation of ROS Noetic, Ceres solver, eigen and other libraries is omitted. In git
vins-fusion After direct compilation, there will be various errors, which is caused by the difference between the opencv4 ...

Added by **tylrwb** on *Tue, 22 Feb 2022 15:15:03 +0200*

## SLAM exercises - point cloud fusion, filtering, smoothing and gridding

Computer vision life learn SLAM learning notes from scratch
The following topics are from computer vision life. Learn SLAM series from scratch
Point cloud fusion
subject
Title: point cloud fusion experiment. Given the RGB + depth images taken by 3 frames (discontinuous) RGB-D camera and the transformation matrix between them (taking t ...

Added by **ndorfnz** on *Sat, 19 Feb 2022 23:36:32 +0200*

## Original SLAM exercise - rigid body rotation in three-dimensional space

Handwritten SLAM notes from zero
Rotation mode conversion
Title 1: the known rotation matrix is defined as rotating 45 ° along the Z axis. Please initialize the rotation vector, rotation matrix, quaternion and Euler angle according to this definition. Please program:
1. The mutual conversion relations of the above four expressions ...

Added by **jonasr** on *Sat, 19 Feb 2022 13:20:47 +0200*

## ORB-SLAM2 from theory to code implementation: ORB feature extraction and matching theory and code explanation

1. Theoretical knowledge
Feature points are composed of key points and descriptors. ORB feature points (Oriented FAST and Rotated BRIEF) are composed of Oriented FAST corner points and BRIEF (binary robot independent elemental features) descriptors. Its calculation speed is 100 times that of sift feature points and 10 times that of surf featur ...

Added by **frans-jan** on *Fri, 18 Feb 2022 12:23:25 +0200*

## Nvidia Xavier NX install and test Cartographer

preface
Recently, we successfully installed and tested Google's open source laser slam algorithm Cartographer in Nvidia Xavier NX and Nvidia Xavier AGX, and recorded some pits in the process.
setup script
First, install ROS. Note that Nvidia Xavier NX and Nvidia Xavier AGX are installed Ubuntu 18 04 system, so install the corresponding ...

Added by **nahla123** on *Wed, 09 Feb 2022 15:29:53 +0200*

## SLAM learning notes -- coordinate transformation and eigen Library

Vector basic knowledge
slightly
coordinate transformation
Basic concepts of coordinate transformation
I recommend taking a look at his linear algebra series 3blue1brown
The following formula represents the most basic transformation, Euclidean transformation (only rotation + translation), and R is an orthogonal matrix. Left multiplying ...

Added by **jmicozzi** on *Mon, 31 Jan 2022 04:26:37 +0200*

## Introduction to ROS learning_ Topic communication mechanism

1, Principle of topic communication
The implementation model of topic communication is complex, as shown in the figure below. The model involves three roles:
ROS Master (Manager) Talker (publisher) Listener (subscriber) The ROS Master is responsible for keeping the registered information of Talker and Listener, matching Talker and Listener wi ...

Added by **adyre** on *Thu, 27 Jan 2022 09:06:24 +0200*

## Laser SLAM: Lidar motion distortion compensation -- auxiliary method of wheel speed odometer

preface
In the previous blog, I combed ALOAM: motion distortion compensation of lidar link
In ALOAM, only lidar is used as the sensor, so the position and attitude matching between frames can be used to compensate the data of the next frame For example, the pose transformation from k-1 frame to K frame is obtained through inter frame matc ...

Added by **franklyn** on *Tue, 25 Jan 2022 06:26:28 +0200*

## Three derivation methods of ceres solver

Nonlinear optimization involves the derivation of the objective function and iterative optimization. Ceres Solver provides three derivation methods: automatic derivation, numerical derivation and analytical derivation.
1. Automatic derivation
Automatic derivation is to define an imitation function and then pass it to AutoDiffCostFunction, s ...

Added by **cabaz777** on *Mon, 24 Jan 2022 09:36:38 +0200*