Rico's Nerd Cluster

「离开世界之前 一切都是过程」

RGBD SLAM - What Deteriorates Its Performance

Lessons Learned From My RGBD SLAM Project. Updates Come In Anytime

“Yeah It’s on. But wait, it doesn’t look good.” Motion Blur Motion blur could cause lower number of features and feature mismatching. These will further cause huge errors in PnP solving. Belo...

Posted by Rico's Nerd Cluster on July 15, 2024

RGBD SLAM Bundle Adjustment Part 2

RGBD SLAM Backend Introduction

If you haven’t, please check out the previous article on how to formulate SLAM as an optimization problem Why Graph Optimization Frontend and Backend: the front end gives the transform from one f...

Posted by Rico's Nerd Cluster on July 12, 2024

RGBD SLAM - Bundle Adjustment From Zero To Hero, Part 1, Theories

RGBD SLAM Backend Introduction

What is Optimization? \[\begin{gather*} min F(x) \end{gather*}\] Above is an optimization problem. Any optimization needs: a cost function, target variables, and constraints. In SLAM, most proble...

Posted by Rico's Nerd Cluster on July 11, 2024

Robotics - Gauss Newton (GN) and Levenberg Marquardt (LM) Optimizers

Newton's Method, GN, LM optimizers

Newton’s Method Solving An Equation To find an arbitrary equation’s root $f(x) = 0$, We start from an arbitrary point $x_0$ that’s hopefully close to the solution, $x_s$ The main idea of Ne...

Posted by Rico's Nerd Cluster on July 11, 2024

RGBD SLAM - The PnP Problem

Solving the PnP problem - turning pixels into 3D positions!

Intro The Perspective-n-Point (PnP) problem is a very important technique in RGBD SLAM. In RGBD SLAM, it’s quite common to see PnP as a front end , and bundle adjustment as the backend. In 2D-2D M...

Posted by Rico's Nerd Cluster on July 9, 2024

More On Motion Estimation

5 Point Algorithm, How to solve DLT, and 8 point algorithm

Why Do We Still Need Multi-View Geometry, In 2024? LLMs are everywhere now. However, they are rarely seen in 3D scene reconstruction except for those on depth learning from single images. Dong et ...

Posted by Rico's Nerd Cluster on July 5, 2024

Motion Estimation From Epipolar Constraints

This is an Introduction of Epipolar Constraints and 2D-2D Motion Estimation

Intro This blog is inspired by this great great book: 14 Lectures in Visual SLAM, authored by Dr. Gao Xiang et al. This article will walk you through 8-point algorithm when features lie on differe...

Posted by Rico's Nerd Cluster on July 4, 2024

Docker Compose

Environment Variable Reading Through .env: TODO Through inline env variables 1 2 3 4 5 6 7 8 9 10 11 12 version: '3.8' services: app: image: your-image:latest environment: - USER...

Posted by Rico's Nerd Cluster on June 5, 2024

Autonomous Vehicle Introduction

General Project Flow What the robot sees - Perception localization + mapping Navigation - Motion Planning Planning + control Autonomy Levels from SAE (Society of Automotive Engineers) ...

Posted by Rico's Nerd Cluster on June 4, 2024

Docker Image Building

Dockerfile

Construct Of An Docker Image And Its Building Process A Docker image is like a box of Oreo. Its final image is like multiple Oreos (layer) stacked on top of each other. A layer could be built from...

Posted by Rico's Nerd Cluster on June 3, 2024

FEATURED TAGS
Math Deep Learning Robotics C++ Python Linux RGBD Slam Computer Vision ROS2 Hands-on ROS Docker Build Systems CUDA Electronics Hands-On ROS1 Web Devel
ABOUT ME

要做一个有 swag 的机器人工程师
Robotics @ Northwestern University

FRIENDS