RoboSub 2025

LazerShark

Coming soon...

LazerShark is our first Autonomous Underwater Vehicle (AUV) and was engineered for the 2025 RoboSub Competition. The vehicle was designed, manufactured, and tested over the course of one year by a small team of students–most of whom had prior experience with ROVs through the MATE competition, but no background in AUVs. With limited time, resources, and team capacity, we focused on optimizing for fast iteration, simplicity, modularity, future expandability, and reliability over creating a polished final product.
We built the majority of subsystems in-house, including the chassis, electrical box, and custom PCBs. Wherever possible, we reused components, tools, and knowledge from our past ROV projects. This gave us a head start on core systems such as the electrical stack-up, enclosure design and embedded software stack.
The result is a modular AUV built to take on the challenges of autonomous underwater navigation. Although the LazerShark is not the final evolution of the CRC’s AUVs, it reflects a year-long journey of trade-offs, rapid learning, and practical engineering grounded in our team’s prior experience.


Technical Report

Chassis

The mechanical design prioritizes modularity and expandability. These qualities, along with rapid development cycles, were favored over producing a polished final product due to the limited team size and time constraints.
The vehicle chassis consists of two main components: the electrical and thruster extrusions. The thruster extrusions are secured to the electrical enclosure using four FDM-printed mounting brackets. The chassis provides mounting points for the thrusters, tooling, and a protected location for the DVL.
LazerShark's weight is balanced by the buoyancy of the electrical box's volume. For fine buoyancy tuning, FDM-printed cartridges filled with lead shot can be added to or removed from unused interior spaces in the electrical box.

LazerShark chassis CAD

Electrical Box

The electrical box houses the complete LazerShark electrical stack. It is constructed from machined MIC 6 aluminum plates that were TIG welded together. The enclosure was designed to provide easy access to all components. It includes two polycarbonate windows: a top window to view the stack and a front window for the ZED 2i camera. These windows are secured with bolts and sealed using O-rings to provide a reliable watertight seal. The box is pressure-tested before each deployment to ensure minimal risk of failure.

LazerShark electrical enclosure with labeled components
  • 1 The Nvidia Jetson Orin Nano, handles the heavy computational costs of computer vision. Compare d to a Raspberry Pi, which CRC has used religiously in the past, a Jetson Orin Nano has a much better performance for CV. The Nvidia Jetson Orin Nano was chosen for its high level of support, documentation, and performance given its low cost.
  • 2 The Pi Hat is a custom legacy PCB designed last year for MATE ROV. The Pi Hat acts as the bridge between the electronics and our ROS network, interfacing with the components directly or through the RP2040 over UART. The RP2040 runs a micro-ROS bridge, allowing it to control the Blue Robotics ESCs and interface with the BMS board. The Pi Hat also has several status LEDs for debugging purposes and cooling fans.
  • 3 Power is supplied by an off-the-shelf lithium-ion battery pack, which connects to a custom Battery Management System (BMS) board which handles all hardware-level control and protection. The BMS board includes reverse polarity protection, a dual power path system and a current shunt IC. To protect the batteries, LazerShark goes into a low power error state if the energy consumed from the battery is above a safe capacity.
  • 4 The StereoLabs Zed 2i stereoscopic camera serves as the eyes of the vehicle. With two lenses which mimic binocular vision, the Zed captures high-resolution 120° Wide-Angle FOV stereo images. These images are processed on the Jetson in real time to identify and locate objects, aid in vehicle position tracking, and create a global map of the environment.
  • 5 A Doppler Velocity Logger (DVL) is a critical component of the vehicle’s localization system, as it measures velocity relative to the bottom of the pool. CRC is grateful to Water Linked for providing us a substantial discount on their A50 Doppler Velocity Logger (DVL). The A50 interfaces with the Pi over ethernet. A custom TCP driver sends velocity data to the ROS network which is used for state estimation.
  • 6 PNI Sensor generously provided us with their NaviGuider and TargetPoint-TCM Inertial Measurement Units (IMUs). Custom drivers were written to interface with the proprietary serial protocols allowing for direct and fast integration with the ROS2 network. These high end IMUs provide the LazerShark with accurate attitude and acceleration data, which is vital to pose estimation and navigation.

Open Source Software

The LazerShark software stack is completely built on the Robot Operating System 2 (ROS2). Using ROS2 and various other open source libraries, CRC has built completely custom control, navigation, and vision systems. An Extended Kalman Filter fuses sensor data from LazerShark’s IMUs, its DVL, and vision systems to provide accurate pose estimations. A fully custom control system based on a Linear Quadratic Regulator takes pose estimates to generate inputs to stabilize the system and to bring the LazerShark to any desired pose. Utilizing the Zed 2i, the Jetson Orin Nano, and the Nvidia Nvblox library, the LazerShark can construct a 3D map of the competition environment and localize within it. These mapping and control systems enable a custom navigation system based on the D* Lite algorithm, allowing for obstacle avoidance in path generation between any two points in the competition environment.


GitHub

Acknowledgements

Cabrillo Robotics Club extends its sincerest gratitude to everyone who has supported the development of LazerShark. Our accomplishments would not have been possible without the generous donations by corporate sponsors, individual donors, and supporters. Thank you to PNI Sensor for providing us with their NaviGuider and TargetPoint-TCM IMUs, and to Water Linked for giving us a generous discount on their A50 DVL. Additionally, thank you to CITO Medical generously providing access to their facilities to machine the electrical housing, and SOLIDWORKS for providing licenses. We are grateful to AMD AMD, Intel, Cabrillo Engineering Department alumnus Mark Cowell, and many others for monetary donations. CRC recognizes the support of advisors Mike Matera and Andrew Thach. Finally, a big thank you to Marlene Coury for providing us a pool for testing. Your generosity is greatly appreciated.


Our Sponsors

Gallery