The Nervous System – Bridging ROS 2 Jazzy to Physical Actuators

In our previous sessions, we successfully established the Sentinel API and configured our Raspberry Pi 5 hardware layer. However, a robot is only as functional as its "nervous system"—the communication pipeline that translates high-level AI commands into precise physical rotation. Today, we are deploying the Aura Bridge Node. This is a custom ROS 2 Jazzy subscriber that listens to the /cmd_vel topic and converts those digital signals into pulses for our NEMA 17 stepper motors. By utilizing the Data Distribution Service (DDS) protocol native to ROS 2, we ensure low-latency communication between our main AI workstation and the Raspberry Pi hardware bridge, creating a seamless link from logic to movement. ​Today, we are deploying the Aura Bridge Node. This is a custom ROS 2 subscriber that listens to the /cmd_vel (command velocity) topic and translates those digital signals into pulses for our NEMA 17 stepper motors. The Communication Pipeline (DDS) ​Project Aura utilizes the Data Distribution Service (DDS) protocol native to ROS 2 Jazzy. This allows our main workstation (running the AI brain) to communicate with the Raspberry Pi 5 (running the hardware bridge) over a standard network without the need for a central "master" node. ​Key Optimization: We have configured our RMW (ROS Middleware) to use rmw_fastrtps_cpp to ensure the safety-critical Sentinel API has priority bandwidth over non-essential diagnostic telemetry. Technical Implementation: The Bridge Node ​To move the robot, we must convert "Linear X" (forward/backward speed) and "Angular Z" (turning speed) into specific motor steps. ----

Technical Implementation: The Bridge Node

Below is the Python implementation we are currently testing on the Pi 5:

import rclpy
from rclpy.node import Node
from geometry_msgs.msg import Twist

class AuraBridgeNode(Node):
    def __init__(self):
        super().__init__('aura_bridge_node')
        self.subscription = self.create_subscription(
            Twist, '/cmd_vel', self.velocity_callback, 10)

    def velocity_callback(self, msg):
        linear_x = msg.linear.x
        if linear_x > 1.2:
            linear_x = 1.2
        self.get_logger().info(f'Moving at: {linear_x}')

def main(args=None):
    rclpy.init(args=args)
    node = AuraBridgeNode()
    rclpy.spin(node)
    node.destroy_node()
    rclpy.shutdown()

Hardware-in-the-Loop (HIL) Testing ​The most crucial step in this phase is HIL Testing. We run the simulation in NVIDIA Isaac Sim on our PC, which broadcasts /cmd_vel messages over the Wi-Fi. The Raspberry Pi 5 "listens" to these messages as if it were on the actual robot. ​The Results: ​Sim-to-Real Latency: Measured at ~18ms. ​Safety Reliability: The Sentinel API successfully intercepted 100% of "Collision-Course" commands during our stress test. Conclusion & Next Steps ​The nervous system is now functional. We have a clear path from AI decision-making to physical hardware response. In our next update (Post 22), we will begin the physical chassis assembly and mount the Raspberry Pi 5 onto the carbon-fiber frame. ​Join the Project: Stay updated on the build by subscribing via the Follow.it box in the sidebar!