Bridging Python Agents to ROS

Connecting AI Logic to ROS Controllers

One of the most powerful features of ROS 2 is its Python client library, rclpy. This allows us to write modern AI agents (using PyTorch, TensorFlow, or LangChain) in Python and directly interface them with the robot's hardware.

The rclpy Library

rclpy provides the standard ROS 2 interface for Python. It allows your Python scripts to become first-class citizens in the ROS graph.

Example: AI Subscriber Node

Here is an example of a node that subscribes to a camera topic and processes the image (placeholder for AI logic).

# See examples/module-1/bridge_node.py for the full runnable code
import rclpy
from rclpy.node import Node
# ... (See file for imports)

class AI_Perception_Node(Node):
    # ... (See file for implementation)
    def run_ai_inference(self, data):
        # Simulation of an AI model prediction
        prediction = "Detected Object: Box"
        self.get_logger().info(f'AI Inference Result: {prediction}')

Hands-on: Running the Bridge

We have created a mock perception node in examples/module-1/bridge_node.py.

python examples/module-1/bridge_node.py

(In another terminal, you can publish to sensor_data_topic to see it react)

Integration Patterns

1. Direct Integration: The AI model runs directly inside the ROS node (as shown above). Good for low-latency.
2. Service-Based: The AI model runs as a separate service (e.g., a Docker container exposing a REST API) and a ROS node acts as a client. Good for heavy models.
3. Bridge Node: A dedicated node translates between ROS messages and the AI model's custom input/output format.