How derivedfunction/mdc-ubuntu works
Project type and purpose
• This is a robotics full-stack application built on ROS 2 (Humble) that fuses sensor data (Azure Kinect), user input (Xbox controller or keyboard), and vehicle actuation (miniature car) with optional SLAM (rtabmap).
Main components
a. Hardware
• Azure Kinect camera (RGB-D sensor)
• Laptop (runs user-input node)
• On-car computer (runs car controller node, optionally SLAM)
• Car actuators via Maestro servo controller
b. ROS 2 Nodes / Software Modules
• laptop: subscribes to Xbox/keyboard, publishes geometry_msgs/Twist or custom control topic
• mdc_car: subscribes control topic, issues servo commands via maestro.py
• (Optional) rtabmap node: subscribes to camera topics, builds map, writes/reads database file
c. Middleware and Services
• ROS 2 DDS (pub/sub layer carrying topics)
• rtabmap SQLite/DB (rtabmap_data)
Relationships & Data Flows
• (Optional) SLAM node (rtabmap) subscribes to camera topics → publishes /map, /odom
• laptop node reads Xbox or keyboard → publishes /cmd_vel or /steering_cmd
• mdc_car node subscribes to control topic → drives car via local Maestro API
• All nodes communicate over ROS 2 DDS (via localhost or network)
Architectural Patterns & Design Principles
• Modular ROS 2 nodes following single-responsibility principle
• Topic-based pub/sub decoupling sensor, control, and actuation
• Clear separation of concerns: sensing (driver), planning/mapping (rtabmap), input (laptop), actuation (mdc_car)
• Scripts orchestrate builds and launches, but do not alter runtime architecture