Sim-to-Real Virtual Guidance for Robot Navigation

It is an effective, easy-to-implement, and low-cost modular framework for robot navigation tasks

Links

• GitHub repository

Introduction

• 1. Introduction
  • 1.1. Intro.
  • 1.2. Features
• 2. How It Works
  • 2.1. Perception Module
  • 2.2. Localization Module
  • 2.3. Planner Module
  • 2.4. Control Policy Module
  • 2.5. Details

Installation

• 1. ROS
  • 1.1. Prerequisites
  • 1.2. Install through Script
  • 1.3. Install Manually
  • 1.4. Set up ROS
  • 1.5. Create a Workspace
• 2. ZED camera
  • 2.1. Install cv-bridge
• 3. ORB-SLAM2
  • 3.1. Install Dependency Packages
  • 3.2. Install Eigen3
  • 3.3. Install OpenCV 3.2
  • 3.4. Install Pangolin
  • 3.5. Install Doxygen
  • 3.6. Install ORB-SLAM2
  • 3.7. Set up Configuration File
• 4. Control Policy Virtual Environment
  • 4.1. Prerequisites
  • 4.2. Create a Virtual Environment
• 5. Perception Environment
  • 5.1. Prerequisites
  • 5.2. Install Preception module via pip
• 6. Navigation Package
  • 6.1. Installation Packages

Usage

• 1. How to Build a Map
  • 1.1. Run Localization Module
  • 1.2. Fine-tune
• 2. Camera
  • 2.1. Launch ZED Camera
• 3. Perception Module
  • 3.1. For Indoor Environment
  • 3.2. For Outdoor Environment
• 4. Localization Module
  • 4.1. Set up Map File
  • 4.2. Setup Environment
  • 4.3. Run ORB-SLAM2
  • 4.4. Run pose convertor
• 5. Planner Module
  • 5.1. Launch planner with Rviz
• 6. Control Policy Module
  • 6.1. Launch RL model
  • 6.2. Launch Goal-Receiver
  • 6.3. Launch AGV Controller
• 7. Robot Control
  • 7.1. Launch the AGV
  • 7.2. Use Joystick To Control AGV

Trainning Procedure

• 1. How to Train an RL Agent
  • 1.1. Why in Virtual Environment
  • 1.2. Training Environment Set-Up
  • 1.3. Training method
  • 1.4. Computational Platform
• 2. How to Train a Segmentation Model
  • 2.1. Installation

Indices and tables

• Index

• Search Page