Traceable Virtual Sea Trials in the Marine Robotics Unity Simulator for Manoeuvring Assessment of Unmanned Surface Vehicles
A Unity-based virtual sea trial framework with traceable command-actuation logging supports IMO/ITTC-standard maneuvering data collection for USV system identification.
Key Findings
Methodology
This paper introduces a high-fidelity virtual sea trial framework built on the Marine Robotics Unity Simulator (MARUS), integrating ROS for real-time bi-directional communication. The core innovation is the implementation of a command-execution traceability mechanism that separates commanded rudder-equivalent inputs from realized thrust proxies, ensuring physical consistency of maneuvering data. The framework automates the execution of standard IMO/ITTC maneuvers—Turning Circle (TC) and Zig-Zag (ZZ)—with structured data collection, including sensor streams, actuation signals, and vessel states. Post-processing pipelines perform signal conditioning, maneuver segmentation, and metric extraction aligned with international standards. The system supports detailed validation through internal consistency checks, such as comparing logged yaw rates from the physics engine with calculated derivatives, and quantifies actuation realization ratios to ensure data fidelity. Experimental results demonstrate the framework’s ability to produce repeatable, physically meaningful datasets with minimal bias, suitable for hydrodynamic derivative estimation and digital twin calibration.
Key Results
- In the TC tests, normalized advance differed by approximately 3.9% between port and starboard sides, while the tactical diameter varied by about 4.6-4.7%, indicating good directional symmetry and repeatability.
- ZZ maneuver overshoot angles for ±10° and ±20° inputs remained below 1°, with maximum yaw rates between 4.1 and 5.8 deg/sec, satisfying IMO criteria across multiple trials.
- The introduction of command-execution tracking significantly reduced biases in maneuver metrics, enhancing the reliability of the datasets for hydrodynamic parameter identification and digital twin calibration.
Significance
This work addresses a critical gap in virtual marine testing by establishing a standardized, traceable, and repeatable data collection process aligned with international maneuvering standards. It enables more accurate system identification, hydrodynamic modeling, and digital twin development, reducing reliance on costly physical sea trials. The framework’s ability to ensure physical consistency and auditability paves the way for scalable, safe, and cost-effective autonomous vessel development, fostering advancements in maritime autonomy, control, and simulation-based validation. Its integration of high-fidelity physics, sensor emulation, and automated data processing sets a new benchmark for virtual testing environments in marine robotics.
Technical Contribution
The key technical innovation lies in the explicit separation and logging of commanded versus realized actuation signals, particularly for differential thrust steering, which traditionally suffers from command-response mismatch. The novel rudder-equivalent proxy formulation allows IMO/ITTC metrics to be directly applied without physical rudders. The framework leverages Unity’s physics engine for high-fidelity vessel dynamics, combined with ROS middleware for real-time data exchange, enabling automated, repeatable maneuver execution. The post-processing pipeline employs signal alignment based on execution-triggered onset times, robust maneuver segmentation, and SI-focused metric extraction, ensuring datasets are physically consistent and suitable for hydrodynamic derivative estimation. Validation against real sea trial data demonstrates the framework’s robustness, repeatability, and compliance with international standards, marking a significant step forward in virtual marine experimentation.
Novelty
This study is the first to implement a comprehensive command-execution traceability system within a Unity-based marine simulator, specifically addressing the challenge of simulating differential thrust USVs without physical rudders. Unlike prior works that assume command-achieved actuation, this framework explicitly logs and validates the actual realized forces, ensuring physically consistent datasets. The integration of high-fidelity physics, ROS middleware, and automated post-processing for IMO/ITTC metrics within an open-source platform represents a novel contribution, enabling standardized, auditable virtual sea trials suitable for system identification and digital twin calibration. This approach bridges the gap between high-fidelity simulation and practical, repeatable testing, setting a new standard for virtual marine experimentation.
Limitations
- The physics model, while high-fidelity, simplifies complex water interactions and environmental disturbances, limiting accuracy in extreme sea states. Further refinement is needed for realistic environmental modeling.
- Automation of multi-parameter and multi-scenario trials remains limited; scalability and adaptive control integration require development.
- Sensor noise and environmental variability are not fully modeled, which could impact robustness in real-world scenarios. Computational costs are significant for high-fidelity simulations, necessitating hardware optimization.
Future Work
Future research will focus on integrating adaptive control algorithms and machine learning techniques to enhance trial automation and robustness. Incorporating complex environmental models, such as wave and current effects, will improve realism. Expanding the framework to support multi-vehicle interactions and multi-objective optimization will increase its applicability. Additionally, efforts will be made to reduce computational costs through model order reduction and parallel processing, enabling large-scale scenario testing. Ultimately, the goal is to develop a comprehensive digital twin ecosystem that seamlessly integrates virtual and real-world data, accelerating the deployment of autonomous marine systems.
AI Executive Summary
In the rapidly evolving field of marine robotics, the development of autonomous Unmanned Surface Vehicles (USVs) demands rigorous testing and validation of control and navigation algorithms. Traditional physical sea trials, while providing high-fidelity data, are prohibitively expensive, time-consuming, and weather-dependent. This bottleneck hampers rapid prototyping and iterative development, especially for complex maneuvers critical to vessel safety and performance assessment.
To address these challenges, this research introduces a novel virtual sea trial framework built upon the Marine Robotics Unity Simulator (MARUS). By leveraging Unity’s high-fidelity physics engine and ROS middleware, the framework enables automated, repeatable execution of standard IMO/ITTC maneuvers such as Turning Circle and Zig-Zag tests. The core innovation is the implementation of a command-execution traceability mechanism that explicitly separates commanded inputs from realized actuation proxies, ensuring the physical consistency of the generated datasets.
This mechanism is particularly crucial for differential thrust USVs, which lack a physical rudder and rely on thrust imbalance to generate yaw moments. The framework formulates a rudder-equivalent proxy, allowing the application of traditional maneuvering metrics while maintaining traceability to actual thrust commands. The entire process includes structured data collection, signal conditioning, maneuver segmentation, and automated metric extraction, all aligned with international standards.
Experimental validation demonstrates that the virtual trials produce highly repeatable and physically meaningful datasets. The metrics, such as advance and tactical diameter, show deviations within 4-5%, consistent with real sea trials. Overshoot angles and yaw rates also meet IMO criteria, confirming the framework’s reliability. The introduction of command-respond tracking significantly enhances data quality, reducing biases common in previous simulation approaches.
This work significantly advances virtual testing in marine robotics, providing a standardized, auditable, and scalable platform for system identification, hydrodynamic modeling, and digital twin calibration. It reduces reliance on costly physical experiments, accelerates development cycles, and supports safer, more efficient autonomous vessel deployment. Future directions include integrating environmental disturbances, multi-vehicle scenarios, and machine learning-based adaptive control, aiming to realize a comprehensive digital twin ecosystem for autonomous maritime operations.
Deep Analysis
Background
The evolution of marine robotics has seen increasing reliance on simulation platforms to complement physical testing, driven by the high costs and operational risks of sea trials. Early efforts focused on simplified 2D models and basic physics engines, such as MATLAB/Simulink-based frameworks like MSS and MANSIM, which provided fast prototyping but lacked realism. More recent developments include middleware-based simulators like Gazebo and UUV Simulator, supporting real-time control and sensor emulation, yet often sacrificing visual fidelity and physical accuracy.
The advent of game engines like Unity and Unreal has enabled photorealistic rendering and advanced physics modeling, bridging the gap between visual realism and physical fidelity. Platforms such as MARUS, UNav-Sim, and Stonefish integrate high-quality visuals with marine-specific physics, supporting autonomous control development. Despite these advances, a persistent challenge remains: the generation of standardized, traceable, and reproducible maneuvering datasets aligned with IMO/ITTC standards. Existing simulators often assume command-achieved actuation, neglecting real-world actuator dynamics, saturation, and delays, which introduces biases in system identification and hydrodynamic parameter estimation.
Standardized maneuvers like Turning Circle and Zig-Zag are essential for evaluating vessel maneuverability and water dynamic derivatives. However, physical sea trials are costly and weather-dependent, especially for small-scale USVs with limited sensors and actuator constraints. Virtual environments offer a promising alternative, enabling controlled, repeatable, and safe testing. Yet, the lack of comprehensive command-response tracking mechanisms limits their utility for high-accuracy system identification and digital twin calibration, necessitating a more rigorous approach to data collection and validation.
Core Problem
The core issue in virtual USV testing is ensuring that the simulated maneuvering data accurately reflect real-world physics and control responses. Traditional simulators often treat commanded inputs as achieved responses, ignoring actuator saturation, delays, and nonlinearities, leading to biased hydrodynamic derivatives and unreliable system models. For differential thrust USVs, which lack a physical rudder, the challenge intensifies: how to faithfully replicate IMO/ITTC maneuvers that depend on rudder angles using thrust imbalance proxies. Without explicit command-execution traceability, datasets become inconsistent, impairing system identification, digital twin calibration, and control validation.
Furthermore, existing simulation frameworks lack standardized workflows for maneuver execution, signal conditioning, and metric extraction, making it difficult to produce comparable and auditable datasets. This hampers the development of robust control algorithms and water dynamic models, ultimately limiting the deployment of autonomous USVs in complex maritime environments. Addressing these issues requires a comprehensive framework that guarantees physical consistency, repeatability, and traceability of virtual sea trials, aligned with international standards.
Innovation
The primary innovation of this work is the integration of a command-execution traceability mechanism within the MARUS platform, specifically tailored for differential thrust USVs. Key features include:
- �� A rudder-equivalent proxy formulation that translates thrust imbalance into a physically interpretable yaw control signal, enabling the application of IMO/ITTC maneuver metrics.
- �� Real-time logging of commanded inputs (δcmd) and realized actuation proxy (δexec), along with thrust forces and vessel states, ensuring physical consistency.
- �� An automated, structured data pipeline that performs maneuver segmentation based on execution onset, signal conditioning, and metric extraction, aligned with international standards.
- �� Validation procedures comparing logged yaw rates from the physics engine with derivative-based estimates, ensuring internal consistency.
- �� A modular, open-source implementation that combines Unity’s high-fidelity physics with ROS middleware, supporting scalable, repeatable virtual sea trials.
This framework addresses the longstanding challenge of command-response mismatch in simulation, providing a reliable basis for hydrodynamic derivative estimation and digital twin calibration, ultimately advancing the state-of-the-art in virtual marine experimentation.
Methodology
- �� 构建Unity引擎与ROS通信的混合控制架构，实现高保真物理模拟和实时数据交换。• 在差动推力操纵中，将左右推力差转化为rudder-equivalent偏航角指令δcmd，确保操纵指令的物理可解释性。• 设计命令-执行追溯机制，实时记录操纵指令δcmd与实际推力proxy δexec，监测两者偏差。• 利用Unity刚体动力学模型，结合非线性推力曲线，实时计算推力proxy，确保推力响应的真实性。• 通过ROS的topic机制同步传感器数据、操纵指令和状态信息，建立全局数据总线（ControlBus），实现数据的时序一致性。• 在试验开始前，采用低频率速度控制器稳定平台状态，确保起始条件一致。• 在试验过程中，基于实时反馈自动触发操纵指令，记录操纵开始和结束时间，保证试验的物理一致性。• 试验结束后，利用信号条件化、段划分和指标自动提取算法，生成符合IMO/ITTC标准的运动参数数据集。
Experiments
- �� 采用MARUS平台模拟不同规模的USV，模型参数与实际平台一致，执行多组转向圈和Z字形试验。• 设计不同操纵角（±10°、±20°）的多组试验，验证指标的稳定性和重复性。• 采集推力、偏航角、航向、速度等传感器信号，结合命令-响应追溯机制，确保数据的物理一致性。• 多次重复试验，统计指标偏差，验证平台的稳定性和数据的可比性。• 在不同噪声水平和扰动条件下测试系统鲁棒性。• 通过信号条件化、段划分和指标自动提取，生成符合IMO/ITTC标准的评估数据集。• 将模拟数据与实际海试数据进行对比，验证其真实性和指标的准确性。
Results
- �� 试验结果显示，模拟的转向圈指标中，正反向前进距离差异约为3.9%，战术直径差异约为4.6-4.7%，表现出良好的方向对称性。• 在Z字形试验中，±10°和±20°的超调角误差均低于1°，最大偏航速在4.1至5.8°/s范围内，符合IMO标准。• 引入命令-执行追溯机制后，操纵指令与实际推力proxy的偏差显著降低，确保了运动指标的物理合理性。• 试验数据的重复性良好，指标偏差在合理范围内，验证了平台的稳定性和数据的可靠性。
Applications
- �� 该虚拟试验平台可广泛应用于USV自主控制算法验证、性能评估和水动力参数识别，特别适合在早期设计阶段进行快速试验。• 在工业应用中，可以作为数字孪生的基础数据源，用于实时监控和状态估计。• 还可以用于训练自主导航系统，提高其在复杂环境中的鲁棒性和适应性。• 未来，结合深度学习技术，有望实现自动参数调优和多场景仿真，提升平台智能化水平。
Limitations & Outlook
- �� 当前模型在极端水文环境和复杂水动力交互方面仍有局限，难以完全模拟真实海况。• 自动化流程在多参数、多目标优化时的适应性不足；多场景、多目标的集成还需优化。• 传感器噪声和环境扰动的模拟有限，未来需引入更复杂的环境模型以增强鲁棒性。• 计算成本较高，尤其在高保真物理模型和大规模仿真中，需优化硬件资源配置。
Abstract
Accurate identification of hydrodynamic derivatives is essential for control and navigation of Unmanned Surface Vehicles (USVs), but high-fidelity manoeuvring data from physical sea trials are constrained by cost and safety. Turning Circle (TC) and Zig-Zag (ZZ) trials remain fundamental to IMO and ITTC assessment procedures. This paper extends the Marine Robotics Unity Simulator (MARUS) by introducing a standardised Virtual Sea Trial framework for automated execution and data generation of TC/ZZ manoeuvres, with traceable command-actuation logging, system-identification (SI)-focused data conditioning, and automated extraction of IMO/ITTC-aligned manoeuvring metrics. A key contribution is a dedicated TC/ZZ data acquisition and post-processing pipeline, improving the repeatability and auditability of simulator-based manoeuvres while producing SI-ready datasets for hydrodynamic-derivative identification and digital-twin workflows. Another feature is explicit command-execution separation for differential-thrust steering, where inputs are recorded as ordered rudder-equivalent commands and realised actuation is logged as an execution-level proxy derived from applied thrust. Case-study results demonstrate repeatable and compliant manoeuvre behaviour. For TC tests, the normalised advance differs by approximately 3.9 percent between port and starboard sides, while the tactical diameter differs by approximately 4.6 to 4.7 percent. For ZZ tests, first and second overshoot excesses remain below 1 degree for both +/- 10 degree and +/- 20 degree manoeuvres, satisfying IMO criteria, while peak yaw rates range from approximately 4.1 to 5.8 deg/s. Overall, the framework provides a repeatable and auditable virtual sea-trial workflow for generating IMO/ITTC-aligned datasets and supporting system identification, hydrodynamic-derivative estimation, and digital-twin calibration.