Getting Started
Overview
This section guides you through the basic requirements and preparation needed to work with Pixel Robot. By the end of this section, you will understand the system requirements, robot variants, and how to choose the correct configuration for your use case.
System Requirements

Pixel Robot is designed to run natively on Ubuntu 22.04 LTS with ROS 2 Humble Hawksbill.
Supported Operating System
Required Software
- ROS 2 Humble Hawksbill (Desktop or Desktop Full)
- Python 3.10+
It is recommended to install the full ROS 2 desktop version to ensure compatibility with visualization and navigation tools.
Supported Robot Variants
Pixel Robot is available in three hardware variants. Each variant builds on the previous one, adding new sensing and perception capabilities.
V1 – LIDAR-Based Navigation
Primary Use Case:
Basic autonomous navigation and SLAM
Features: - Differential drive mobile base - 2D LIDAR for mapping and obstacle detection - Wheel encoder–based odometry - ROS 2 Navigation Stack support - Keyboard or joystick teleoperation
Typical Applications: - Learning ROS 2 navigation - Mapping and localization experiments - Indoor autonomous navigation
V2 – LIDAR + Vision (Pi Camera)
Primary Use Case:
Autonomous navigation with visual object detection
Additional Features over V1: - Raspberry Pi Camera module - Vision-based object detection - Camera integration with ROS 2 image pipeline
Typical Applications: - Object detection and tracking - Vision-based robotics projects - AI perception experiments
V3 – LIDAR + Depth Vision (Intel RealSense)

Primary Use Case:
Advanced perception and 3D-aware navigation
Additional Features over V2: - Intel RealSense depth camera - RGB-D perception - Depth-based obstacle understanding - Enhanced perception for autonomy
Typical Applications: - 3D perception research - Depth-aware navigation - Advanced autonomous robotics projects
Choosing the Right Variant
| Variant | LIDAR | Camera | Depth | Use Case |
|---|---|---|---|---|
| V1 | ✓ | ✗ | ✗ | Navigation & SLAM |
| V2 | ✓ | Pi Camera | ✗ | Navigation + Object Detection |
| V3 | ✓ | RealSense | ✓ | Advanced Perception & Autonomy |
Choose the variant that best matches your learning goals or project requirements.
You’re now ready to begin working with Pixel Robot 🚀