Building Underwater Robots: An Open-Source Toolkit

Author: Denis Avetisyan


Researchers have developed a modular and affordable open-source toolkit designed to accelerate the development of underwater robotic systems for a range of applications.

This toolkit integrates control and power management electronics with the ROS 2 software stack to facilitate operation of Underwater Robotic Joints, encompassing both serial robots and tendon-driven soft manipulators, and extending to underactuated sampling systems.
This toolkit integrates control and power management electronics with the ROS 2 software stack to facilitate operation of Underwater Robotic Joints, encompassing both serial robots and tendon-driven soft manipulators, and extending to underactuated sampling systems.

This work details a ROS2-based framework featuring a pressure-compensated, leak-detecting joint and associated control systems for underwater manipulation.

Despite growing demand for underwater robotics in fields like marine science and subsea operations, progress is hindered by the high cost and limited accessibility of modular, research-grade hardware. This paper introduces ‘An Open Toolkit for Underwater Field Robotics’, presenting a cost-effective and fully open-source platform encompassing a depth-rated robotic joint with early leakage detection, compact control electronics, and a ROS2-based software stack. Validated through laboratory testing and field deployments-including a manipulator, soft gripper, and sediment sampler-this toolkit demonstrably lowers the barrier to entry for underwater manipulation research. Will this open approach accelerate innovation and improve reproducibility in the rapidly evolving field of underwater field robotics?


The Challenge of Underwater Precision

The ocean presents a uniquely challenging environment for robotics. Current underwater systems, while capable of basic functions, frequently struggle with the precision and dependability demanded by intricate tasks. This limitation stems from a confluence of factors including diminished visibility, strong currents, and the immense pressure at depth, all of which impede fine motor control and reliable operation. Traditional robotic joints, often relying on seals and lubrication, are particularly vulnerable to water ingress and corrosion, leading to reduced lifespan and unpredictable performance. Consequently, complex manipulations – such as delicate sample collection, infrastructure repair, or even archaeological excavation – often remain beyond the reach of autonomous underwater vehicles, necessitating continued reliance on costly and potentially dangerous human intervention.

The operational lifespan and efficacy of current underwater robotic systems are significantly hampered by persistent challenges in maintaining precise control and preventing water ingress. Traditional joint designs, often relying on seals and O-rings, degrade under the immense pressure and corrosive saltwater environments, leading to performance drift and eventual failure. Water intrusion not only compromises electrical components but also introduces friction and corrosion within the joints themselves, demanding frequent maintenance and limiting sustained operation. This issue is exacerbated at greater depths, where even minor seal imperfections can result in catastrophic failure, hindering long-term underwater exploration and intervention tasks. Consequently, the development of innovative sealing strategies and robust joint mechanisms is crucial for enabling reliable and extended underwater robotic deployments.

The pursuit of deeper and more effective underwater exploration and intervention hinges critically on advancements in robotic joint technology. Current limitations in depth and operational lifespan stem from the challenges of maintaining precision and preventing water ingress into sensitive mechanisms; a truly robust joint must overcome these hurdles. Such a design necessitates not only durable materials and innovative sealing strategies, but also adaptable control systems capable of compensating for unpredictable currents and varying payloads. Progress in this area isn’t simply about incremental improvements; it’s about unlocking access to previously unreachable environments, enabling complex tasks like deep-sea maintenance, intricate archaeological recovery, and the detailed study of fragile marine ecosystems – all requiring consistent, reliable performance at extreme depths and under immense pressure.

This assembly illustrates the core mechanical components of the underwater robotic joint, excluding wiring and the DHT11 humidity sensor.
This assembly illustrates the core mechanical components of the underwater robotic joint, excluding wiring and the DHT11 humidity sensor.

Resilient Design: Core Components

The Underwater Robotic Joint employs a Dynamixel XM430-W350 smart servo motor to achieve accurate and controlled articulation. This motor provides both position and velocity control, allowing for complex movement patterns essential for underwater manipulation and navigation. The XM430-W350 features a multi-turn resolution encoder and a closed-loop control system, enabling precise positioning even under load and in challenging environmental conditions. Its integrated features minimize the need for external control hardware, contributing to the joint’s compact design and simplifying system integration.

The underwater robotic joint utilizes an aluminum canister to provide physical protection for sensitive internal components. This cylindrical housing is constructed from aluminum alloy 6061-T6, selected for its high strength-to-weight ratio and corrosion resistance in saltwater environments. The canister’s wall thickness is 8mm, providing structural integrity against external pressures and potential impacts during operation. Finite element analysis was conducted to verify the canister’s ability to withstand a minimum of 3 bar of external pressure, corresponding to a depth of 30 meters. The design incorporates a threaded end cap with a torque specification of 20 Nm to ensure a secure and watertight seal when combined with the O-ring.

The Underwater Robotic Joint maintains operational integrity at depth through a combination of sealing technologies. An O-Ring seal provides the primary static barrier against water ingress, while a Blue Robotics penetrator establishes a watertight connection for all electrical conductors. This dual-layer approach has been rigorously tested, demonstrating reliable functionality in field deployments to 30 meters and confirmed performance up to 40 meters under controlled hyperbaric chamber conditions. These validation tests confirm the system’s ability to protect sensitive electronics from hydrostatic pressure and maintain consistent electrical connectivity during submerged operation.

Hyperbaric chamber pressure is maintained while a static load is applied to the underwater robotic joint for testing.
Hyperbaric chamber pressure is maintained while a static load is applied to the underwater robotic joint for testing.

Proactive Leak Detection: Ensuring Longevity

An Early Leakage Detection System was integrated directly into the joint assembly. This system employs both a capacitive humidity sensor and a thermistor to identify the presence of water ingress. The capacitive humidity sensor measures changes in dielectric constant caused by moisture, while the thermistor detects temperature variations associated with evaporating water. Data from both sensors is processed to provide an indication of potential leakage, enabling proactive maintenance and reducing the risk of component failure. The combined sensor approach increases the reliability of leak detection by cross-referencing data and minimizing false positives.

The integrated Early Leakage Detection System functions by continuously monitoring for the ingress of water, providing alerts at the initial stages of potential leakage. This proactive approach enables operators to implement preventative maintenance measures, such as resealing or component replacement, before minor issues escalate into significant structural failures. By addressing leaks promptly, the system minimizes the risk of costly downtime, repairs, and complete joint compromise, thereby extending the overall operational lifespan and ensuring sustained performance of the robotic system. Timely intervention, facilitated by the early warning system, is critical in preventing catastrophic failures that could result from prolonged exposure to moisture and subsequent material degradation.

The implemented Early Leakage Detection System exhibits sufficient sensitivity to identify even trace amounts of moisture ingress within the joint, a capability crucial for proactive maintenance and extended operational life. Concurrent testing demonstrated the integrated tendon-driven soft gripper’s robust grasping performance, achieving a maximum pull-out force of 37 N with a standard deviation of 3.7 N. This combination of sensitive leak detection and reliable grasping ensures both preventative maintenance and continued functionality under operational loads.

Following water ingress, the humidity sensor rapidly detects and registers a corresponding increase in moisture levels, as evidenced by the simultaneous changes in temperature and humidity over time.
Following water ingress, the humidity sensor rapidly detects and registers a corresponding increase in moisture levels, as evidenced by the simultaneous changes in temperature and humidity over time.

Expanding Capabilities: Integrated Systems and Control

The robotic joint’s architecture centers on a seamless integration with the Robot Operating System 2 (ROS 2) software framework, a critical design choice that unlocks sophisticated capabilities for underwater manipulation. This connection facilitates advanced control algorithms, allowing for precise and coordinated movements despite the challenges of the aquatic environment. Moreover, the ROS 2 integration streamlines data acquisition from integrated sensors – including force, torque, and position – and enables robust communication with external systems. By leveraging the modularity and extensive tooling of ROS 2, developers can readily implement and test new control strategies, perception algorithms, and communication protocols, fostering rapid innovation and adaptation for diverse underwater tasks. The standardized messaging and service interfaces provided by ROS 2 also promote interoperability with other robotic systems and software tools, expanding the potential for collaborative underwater operations.

The robotic architecture facilitates the creation of specialized tools for diverse underwater tasks. Notably, a tendon-driven soft gripper allows for delicate manipulation of fragile objects, adapting to irregular shapes without causing damage – a crucial capability for marine archaeology or biological sample collection. Complementing this is an underactuated sediment sampler, designed to efficiently collect seafloor samples with minimal power consumption and mechanical complexity. This design prioritizes robustness and ease of use in challenging underwater conditions. By providing a flexible platform for tool integration, the system significantly expands the range of operations achievable in environments ranging from shallow coastal areas to depths of 30 meters, offering capabilities beyond those of traditional rigid robotic systems.

The newly developed robotic joint’s capacity for intricate underwater tasks is exemplified by a three-degree-of-freedom RRR serial manipulator, a device capable of complex motions within challenging aquatic environments. This manipulator isn’t merely a demonstration of kinematic possibility; it represents a fully validated toolkit, rigorously tested and proven reliable for operation at depths reaching 30 meters. Such performance signifies a substantial advancement in underwater robotics, opening possibilities for detailed inspection, precise manipulation of objects, and delicate sample acquisition in previously inaccessible or hazardous locations. The successful validation at this depth confirms the robustness of the system’s design and materials, suggesting a pathway toward even deeper-water applications and prolonged operational lifespans.

During gait, joint feedback successfully controls the RRR-manipulator's position while tracking a semi-elliptical trajectory with its end-effector.
During gait, joint feedback successfully controls the RRR-manipulator’s position while tracking a semi-elliptical trajectory with its end-effector.

The presented toolkit prioritizes modularity and affordability in underwater robotics. This approach mirrors a fundamental principle of efficient design: simplification. As John von Neumann stated, “It is possible to carry out any operation on any object, but it is not always desirable to do so.” The system’s focus on open-source hardware and software, coupled with early leakage detection, directly addresses the complexities inherent in underwater environments. Abstractions age, principles don’t; a robust, adaptable framework-like the one detailed-outlives fleeting technological trends. Every complexity needs an alibi, and this toolkit offers a clear justification for each design choice, fostering a system that is both powerful and understandable.

The Simplest Solution

The presented work, in its earnest pursuit of modularity and affordability, highlights a perennial truth: the most significant challenges lie not in building complexity, but in eliminating it. A successful robotic system, particularly one submerged, should ideally require no explanation. The fact that extensive software frameworks and pressure compensation schemes are still necessary suggests a fundamental failure to approach the problem with sufficient parsimony. Each component added, each line of code written, represents a potential point of failure, a demand for maintenance, an admission of insufficient elegance.

Future effort should not center on extending the toolkit, but on contracting it. The true metric of progress will be the number of features discarded, not added. Leakage detection, while pragmatic, is a symptom of inadequate sealing – a tacit acknowledgement of inherent unreliability. The goal is not a robot that reports its impending failure, but one that is incapable of failing in the first place.

The current focus on software frameworks, while offering flexibility, risks obscuring a more fundamental problem: the unnecessary complexity of the hardware itself. Clarity is courtesy; a truly open toolkit will be defined not by its extensibility, but by its emptiness-a space for ingenuity unburdened by pre-conceived limitations. A system that needs instructions has already failed.


Original article: https://arxiv.org/pdf/2512.15597.pdf

Contact the author: https://www.linkedin.com/in/avetisyan/

See also:

2025-12-18 23:26