Software

The ROS 2 stack our robots run

We develop on ROS2, running on Ubuntu. Our stack spans perception, navigation, manipulation and high-level task execution, built and tested across our GitHub repositories .

Perception

  • vision_system - ROS 2 packages for our vision system, processing the sensor data our robots need to interact with the environment.
  • yolotino - training pipeline for our YOLO object-detection models; inference then runs on-robot.
  • alvar - our fork of the AR-marker library ALVAR, adapted for OpenCV 3/4.
  • robotino_navigation - our navigation stack, built on ROS 2’s Navigation2 framework.
  • laser_scan_integrator - fuses two laser scanners into one, accounting for the robot’s footprint.
  • motor-move - precise, obstacle-avoidance-free positioning used specifically for approaching production stations (MPS), as opposed to general navigation.

Manipulation

Task execution and reasoning

Our robots’ decision-making combines a CLIPS-based rule engine with newer behavior-tree components:

  • clips_executive and clips_vendor - the ROS 2 CLIPS executive and its packaging of the underlying CLIPS rule engine.
  • clips_rcsml_agent - our CLIPS-executive-based agent for the league (per its own description: “for the RoboCup Logistics League, soon to be RoboCup Smart Manufacturing League”).
  • expertino-rcll - our agent for central goal reasoning.
  • robotino-bts - behavior trees for the Robotino platform.

Simulation and monitoring

  • rcll_simulation_webots - our Webots simulation world for the league, used for repeatable testing without physical robots.
  • monitino - monitoring tooling for the stack.
  • zenoh_config - our configuration for rmw_zenoh, the Zenoh-based ROS 2 middleware.

Legacy and migration

For years, our competition-winning software was built on the Fawkes Robot Software Framework (“Fawkes for the RCLL Carologistics Flavor”). As the league moves toward Smart Manufacturing, our active development has shifted to the ROS 2 stack described above; Fawkes remains an important part of our technical history and is still maintained.

Repositories are updated frequently - check github.com/carologistics for the current state of any specific package.