RDML BlueROV2 Deployment

This repository hosts the onboard software stack used to deploy the Robotic Decision Making Lab's (RDML) BlueROV2.

---

Hardware

ROV Platform

• BlueROV2 (Blue Robotics) -- Heavy Configuration with 150 m tether
• Navigator Flight Controller (with Raspberry Pi 4, 8 GB)
• Alpha 5 (Reach Robotics)
• Modifications: payload bay equipped with additional compute and sensing capabilities

Compute

Device	Operating System
Raspberry Pi 4, 16 GB (bluerov_pi)	Raspberry Pi OS Lite (Bookworm)
Raspberry Pi 5, 16 GB (autonomy_pi)	Ubuntu 26.04 Server
Teensy 4.0	Teensyduino
NVIDIA Jetson Orin Nano	Ubuntu 24.04

Additional Sensors

• Nucleus 1000 DVL (Nortek)
• SeaTrac Lightweight USBL (Blueprint Subsea)
• Oculus Sonar (Blueprint Subsea)

Cameras

• stellarHD (Deep Water Exploration)

Topside

• See our corresponding topside repository

---

System Architecture

flowchart LR
    subgraph ElectronicsBottle["Electronics Bottle"]
        FC["Navigator Flight Controller\n(ArduSub)"]

        subgraph BlueROVPi["bluerov_pi (Raspberry Pi 4)"]
            BlueOS["BlueOS"]
            MAVROS["mavros_node\n(Docker)"]
        end
    end

    subgraph AutonomyBottle["Autonomy Electronics Bottle"]
        subgraph AutonomyPi["autonomy_pi (Raspberry Pi 5)"]
            Agent["teensy_driver"]
            Bar30["bar30_router"]
            DVL["nucleus_driver"]
            BME["bme680_driver"]
            EKF["robot_localization"]
            RC["ros2_control\n(control stack)"]
            Coord["controller_coordinator"]
        end

        subgraph Teensy["Teensy 4.0"]
            IMU["BNO08x driver"]
        end
    end

    FC -- MAVLink --> BlueOS
    BlueOS -- MAVLink --> MAVROS
    MAVROS -- "/uas1/mavlink_source" --> Bar30
    IMU -- "serial, 921600 baud" --> Agent
    Agent -- "/vehicle/imu01" --> EKF
    DVL -- "/nucleus_driver/twist" --> EKF
    Bar30 -- "/vehicle/depth" --> EKF
    EKF -- "/vehicle/odometry/filtered" --> RC
    Coord -- "controller_manager/switch_controller" --> RC
    RC -- "mavros/rc/override" --> MAVROS

Loading

---

Networking

Device	IP Address
Raspberry Pi 4 (bluerov_pi)	192.168.2.2
Raspberry Pi 5 (autonomy_pi)	192.168.2.3
NVIDIA Jetson Orin Nano	192.168.2.4
Nortek Nucleus 1000	192.168.2.201
Barlus Underwater Camera (New)	192.168.2.20
Barlus Underwater Camera (Old)	192.168.2.11

---

Wiring & GPIO

Teensy 4.0

Signal	Pin / value
I2C SDA	18
I2C SCL	19
BNO08x INT	2
BNO08x RST	3
I2C addr	0x4A
UART → autonomy_pi	Serial1, 921600 baud → Pi UART0 (/dev/ttyAMA0)
USB debug	Serial, 9600 baud

Autonomy Pi GPIO

Device	GPIO
arm1	24
arm2	25
sonar	26
dvl	27
PLD (power-loss detect, pld_monitor)	6 — no debounce, 1 low sample = shutdown

---

Firmware and Software

BlueOS Configuration

• ArduSub: vX.X.X
• BlueOS's MAVLink router is configured with an additional UDP endpoint (UDP Client), in the MAVLink Endpoints page, targeting 127.0.0.1:14755. This mirrors the flight controller's MAVLink stream to the mavros_node container (network_mode: host), which listens on that port per its fcu_url in mavros.yaml.

ROS 2 Configuration

• ROS 2 Lyrical
• MAVROS is loaded by default on the bluerov_pi via service
• The core autonomy stack, including transforms, state estimation, and control are loaded by default on the autonomy_pi via service

---

Getting Started

Each device is provisioned by its own install script, which sets up the ROS 2 workspace, system dependencies, and systemd services:

• bluerov_pi: modules/bluerov_pi/scripts/install.sh
• autonomy_pi: modules/autonomy_pi/scripts/install.sh
• jetson: modules/jetson/scripts/install.sh

The Teensy firmware is built and flashed with PlatformIO:

cd modules/teensy
pio run -t upload

---

Citation

This repository has been used in the following papers:

@article{palmer2026stochastic,
  title         = {{Stochastic Physics-Informed Neural Networks on Lie Groups for Learning Underwater Vehicle Dynamics}},
  author        = {Palmer, Evan F. and Hatton, Ross L. and Hollinger, Geoffrey A.},
  journal       = {arXiv preprint arXiv:2608.08356},
  year          = {2026},
  eprint        = {https://doi.org/10.48550/arXiv.2608.08356},
  archivePrefix = {arXiv},
  primaryClass  = {cs.RO},
}

License

This project is licensed under the MIT License. Any documentation included from an external source retains its original licensing.