YOU CAN BUILD AN OPEN-SOURCE UNDERWATER RESEARCH ROBOT FOR 850 DOLLARS
An underwater robot built from 3D-printed cradles, off-the-shelf thrusters and a Raspberry Pi Compute Module, for 850 to 2,000 dollars depending on thrusters and printing.
by Scott Mayberry, Jinzhi Cai, Ruochu Yang, Junkai Wang, Fumin Zhang
RoboticsOpen-hardware
Built withRaspberry PiTeensy3D printing
- difficulty
- ●●●●●
- time
- a month of weekends
- license
- MIT (software), CERN-OHL-P-2.0 (hardware)
- repo
- repo FINISHED0 stars
●●●●● · a month of weekends · MIT (software), CERN-OHL-P-2.0 (hardware) · 0 stars · repo FINISHED
WHAT YOU’LL NEED
Jump to section
COMPAREE VERDICT
MUR is a research platform published by a university robotics lab, not a weekend kit. The paper and its GitHub/Zenodo repository give you everything you need to replicate it — CAD, PCBs, code, sensor choices, thruster options and assembly videos — but this is a build for someone who has already assembled a quadcopter or a robot arm and wants to take the same approach underwater. With six generic off-the-shelf thrusters and self-printed parts it comes to about 860 dollars; with Blue Robotics T200 thrusters it is closer to 1,900, or around 2,000 with commercially printed parts. The single hardest part is not the printing or the wiring, it is the ROS (ROS1) software and sensor fusion, so the IMU, pressure sensor and other inputs agree on where the robot is. If you have never worked with ROS, that will take longer than the mechanical build. It is a shallow-water platform: the authors recommend 0-10 m. What you get for the cost is a modular, repairable, fully open platform that a university lab or a maker collective can build, break, fix and improve — the opposite of a sealed commercial unit that ships back to the manufacturer for repair.
IN THE REPO
GOOD TO KNOW
- —Full CAD for the 3D-printed cradles, ballast box and thruster mounts is on Zenodo and GitHub.
- —Bill of materials names every sensor, thruster option, ESC, camera and connector.
- —ROS stack is complete: sensor fusion, camera drivers, thruster control and telemetry over Ethernet.
- —The paper is the documentation, with the BOM, assembly sections with photos, wiring and PCB descriptions, cost options (Table 4) and validation results.
- —Licence: MIT for software, CERN Open Hardware Licence v2 Permissive for hardware — both allow commercial use.
- —Recommended depth is 0-10 m; the authors tested position holding in a test tank at varying depths over a 15-minute trial.
Parts to buy
11 itemsFrom our check of the build. Exact quantities and part numbers are in the creator’s BOM.
Can I build this?
Build at your own risk. Projects involve tools, electronics and sometimes mains voltage — follow the creator’s safety notes.
More builds like this
All projectsGallery
Start here
Navigation into the creator’s own docs — we don’t rewrite the guide, we route you to the source.
- 1.Read the HardwareX paper (The paper is the build guide, with the bill of materials, assembly photos, PCB descriptions, the four cost options and validation results.)
- 2.Download the CAD and code (Files are on GitHub (github.com/scottmayberry/MUR) and Zenodo (doi.org/10.5281/zenodo.14968240): CAD for the printed parts, PCB files, the ROS stack and setup guides.)
- 3.Decide on thrusters(Six generic off-the-shelf thrusters bring the build to about 860 dollars; six T200s push it to about 1,900. The paper lists both options in its cost table.)
- 4.Set up the ROS environment(MUR currently runs ROS1 (a ROS 2 port is in progress), on the ground station and on the Raspberry Pi. If you have never run a ROS workspace, do a tutorial project first.)
KNOWN ISSUES
- The cost range is wide - 850 dollars to 2,000 dollars - because the thruster choice dominates it: six T200s cost about 1,200 dollars against about 180 for six generic thrusters. The paper compares them only on price, so decide whether a shallow-water platform needs the premium thrusters before you order.
- It is a shallow-water platform: the authors recommend 0-10 m, and their station-keeping test was done in a test tank. Going deeper needs extra waterproofing and pressure-rated parts.
- Tuning the sensor fusion and control stack takes longer than the mechanical build if you are new to ROS. MUR runs on ROS1 today, so use a ROS1 setup rather than a current ROS 2 distribution.
- The sensor connectors follow the Pixhawk standard, which is good for compatibility but means you need the right crimping and pin-removal tools the first time.
- Documentation is split between the paper and the repository's README files, setup guides and assembly videos; there is no single printed manual, so keep both open while building.
- Assembly videos exist in the repository (hardware/Assembly/assembly_videos) and on YouTube — watch them alongside the paper before you start gluing and potting.
Can I use this in a pool?
Yes, that is exactly the environment it is designed for — shallow water testing, pool trials, pond surveys. The lack of a depth rating means open ocean or lake diving below 10 metres is not recommended.
Do I need the Raspberry Pi Compute Module 4 or can I use a regular Pi?
Yes, the CM4. It plugs into the custom Compute Module Mini board, which also carries the sensors, a three-port Ethernet switch, USB ports and CAN interfaces. A regular Pi 4 would mean redesigning that board and the compute cradle.
What if I have never used ROS?
Start with a smaller ROS tutorial project first - a wheeled robot or a sensor rig - and note that MUR currently runs ROS1, so learn on ROS1. The MUR stack assumes you already know topics, launch files and parameter tuning.
How long does the battery last?
The authors measured about 80 minutes on a standard 4S 3200 mAh pack under moderate thruster use. Expect less if you run the thrusters hard.
Can I add more cameras?
The design supports it: the Compute Module Mini has six USB ports and the camera streaming service works with any USB camera, matched by its ID. You would print or adapt a cradle for the extra camera and check that the tether link keeps up with more streams.
Community builds
No community builds yet — be the first, we feature the best ones.
Discussion1
FROM THE COMPAREE TEAM
The paper's build comes in at roughly 860 to 1,880 dollars depending on thrusters, and it is rated for 0 to 10 metres. Where would you actually deploy it: pool testing, pond surveys, or something else?
Scott Mayberry, Jinzhi Cai, Ruochu Yang, Junkai Wang, Fumin Zhang
MUR was developed by a research team at Georgia Institute of Technology and the Hong Kong University of Science and Technology. The paper published in HardwareX positions it as a low-cost alternative to commercial ROVs for shallow water research, where the usual requirement for a boat, a budget and a support team limits who can deploy underwater robotics.
DISCLAIMER
- Comparee is not the author of the projects featured here. All rights to each project belong to its creator — every page links to the original source, and we never host creators’ files.
- Information is provided without warranty and may become outdated as projects evolve. Prices are indicative bands only — always check the creator’s parts list for current costs.
- Building and operating any project is at your own responsibility. Protective equipment, safe workshop practice and compliance with local regulations are the builder’s responsibility.
CompareeTEAM22d agoedited
Practical notes from our verification: the software and design files are on GitHub (github.com/scottmayberry/MUR) and archived on Zenodo, and the HardwareX paper is the main documentation, with a detailed bill of materials. There are also official intro and thruster assembly videos. The single biggest decision is the thruster choice: the paper's four build options run from about 850 dollars (generic thrusters, self-printed) to about 2,000 (Blue Robotics T200 thrusters, commercially printed parts). The authors measured about 80 minutes of runtime on a 4S 3200 mAh pack. The recommended operating depth is 0 to 10 metres, so this is a shallow-water platform. The control stack is ROS1, with a ROS 2 port in progress, so check that fits your setup before you start. If you need something rated for deeper dives, this is not it. Correction (4 October 2026): we re-checked this page line by line against the project's own repository, documentation and videos, and fixed errors in earlier versions.