ROS 2 integration (quantum_mind_ros)

A ROS 2 (ament_python) package with one node, human_model_node, that serves Quantum Mind’s HumanModelService: it collects people’s answers, refits the human-model ensemble, and answers queries with a predicted answer distribution, its uncertainty and an ask-or-act decision.

Topic	Type	Content
/human_model/answers (subscribed)	std_msgs/String	JSON {"order": "AB", "answers": [1, 0]} (1 = yes, in the order asked)
/human_model/query (subscribed)	std_msgs/String	JSON {"order": "AB", "ask_cost": 1, "error_cost": 3} (fields optional)
/human_model/decision (published)	std_msgs/String	JSON reply: prediction, p_first_yes, uncertainty, weights, action, reason

Only standard messages are used, so no interface package has to be built.

# in a ROS 2 workspace (Humble or later), with Quantum Mind installed in the same Python environment
cp -r integrations/ros2/quantum_mind_ros ~/ros2_ws/src/
cd ~/ros2_ws && colcon build --packages-select quantum_mind_ros && source install/setup.bash
ros2 launch quantum_mind_ros human_model.launch.py
ros2 topic pub --once /human_model/answers std_msgs/String "{data: '{\"order\": \"AB\", \"answers\": [1, 0]}'}"
ros2 topic pub --once /human_model/query std_msgs/String "{data: '{\"order\": \"AB\"}'}"
ros2 topic echo /human_model/decision

Testing status. The service logic (quantum_mind.applications.robotics.HumanModelService) is tested in the package’s CI, and the node’s callbacks are tested there with stand-in rclpy and std_msgs modules (tests/test_ros2_node.py). The package has not yet been built and run in a ROS 2 installation; please report problems via the issue tracker.