CartoGrasp

English | 简体中文 | 繁體中文

An open-source indoor mobile manipulation robot project focused on small indoor object handling, room-scale mapping, bounded navigation, and user-owned local AI workflows.

Table Of Contents

• CartoGrasp
  • Table Of Contents
  • What This Project Is
  • What It Tries To Do
  • What This First Version Focuses On
  • Current Status
  • Repository Structure
  • Quick Start
    • 1. Clone The Repository
    • 2. Read The Overview
    • 3. Follow The Documentation
    • 4. If You Want To Contribute
  • Contributing
  • Safety Note
  • License

What This Project Is

This repository is the home of the CartoGrasp project.

The long-term goal is to build an open-source indoor robot that combines:

• room-scale mapping and localization
• bounded navigation between named locations
• a lightweight mobile arm for pickup, transport, and placement
• local perception workflows for a small supported object set
• reproducible documentation for a user-owned deployment path

The README files describe the project scope, usage direction, and contribution entry points.

What It Tries To Do

CartoGrasp is aimed at a practical indoor workflow:

• map a room or apartment
• navigate to a specified named destination
• identify a supported object near the target area
• pick up a selected object with a robot arm
• move it to a low-height drop box and place it down safely

The project also intends to support a local data workflow for new objects. When the robot encounters an unsupported object, operators should be able to trigger a recording flow, upload the captured data to a local host, and use it later for dataset expansion and model updates.

What This First Version Focuses On

The first version is intentionally narrow. It is mainly focused on:

• one indoor mobile robot
• one ROS 2-based system architecture
• one practical sensing path centered on LiDAR, onboard vision, and local perception
• one manipulation path for light household objects
• one user-owned local network workflow for data upload and model update

That also means the first version is intentionally not trying to do everything at once. For now, it does not aim to cover:

• arbitrary object handling without scope limits
• cloud-dependent operation as the baseline path
• unrestricted open-domain voice interaction
• end-to-end reinforcement learning as the main control architecture
• broad claims before hardware and software baselines are stable

Current Status

Planning / prototyping

Right now the repository is focused on:

• project identity
• repository structure and documentation
• repository structure
• contribution workflow
• early planning placeholders

Concrete subsystem implementations will be added incrementally as the project moves from planning into reproducible prototyping.

Repository Structure

.
|-- .github/                  # Issue templates, PR template, CI workflows
|-- README.md                 # English overview
|-- README.zh-CN.md           # Simplified Chinese overview
|-- README.zh-TW.md           # Traditional Chinese overview
|-- CONTRIBUTING.md
|-- CODE_OF_CONDUCT.md
|-- SECURITY.md
|-- SUPPORT.md
|-- LICENSE
`-- NOTICE

Quick Start

1. Clone The Repository

git clone https://github.com/Zw-awa/CartoGrasp.git
cd CartoGrasp

2. Read The Overview

Start with:

• README.md
• README.zh-CN.md
• README.zh-TW.md

3. Follow The Documentation

Start here:

• Docs
• Overview
• Usage
• Safety

4. If You Want To Contribute

Read:

• CONTRIBUTING.md
• Issue templates
• SUPPORT.md

Contributing

Contributions across robotics, embedded systems, perception, navigation, manipulation, documentation, and reproducibility are welcome. Start with CONTRIBUTING.md.

For large changes, open an issue first so the project can keep its technical assumptions coherent.

Safety Note

This repository targets a moving electromechanical robot prototype. Even early indoor robots can pinch fingers, collide with furniture, drop objects, or damage hardware if brought up without proper safeguards. Treat motion, power, battery, and manipulator testing as hardware safety work.

License

This project is licensed under the Apache License 2.0. Copyright 2026 Zw-awa. Additional attribution details are recorded in NOTICE.