Autonomous Design of a Green Sea-Cleaner Boat

. This paper addresses the urgent global issue of marine pollution, with a focus on the inadequacies of traditional cleanup methods and the promising potential of an autonomous green sea-cleaner boat. The study presents a sea-cleaner boat design with reduced wake-wash, integrated with an autonomous system using visual equipment and LiDAR sensors for precise navigation and debris identification. Parametric studies of catamaran hull configurations show that the Flat-Outside Model exhibits the lowest wave elevation, indicating reduced hydrodynamic force and wake-wash, making it an environmentally preferable option. Additionally, the sea-cleaner boat employs LiDAR for obstacle detection and a camera with a Convolutional Neural Network for efficient debris identification and collection, enhancing operational safety and efficiency. Overall, the autonomous green sea-cleaner boat represents a significant advancement in maritime environmental conservation.


Introduction
Marine pollution, particularly the menace of plastic waste, is an urgent global issue [1] [2] and [3].Traditional cleanup methods like manual beach clean-ups, vessels with nets for debris collection, and diving operations are increasingly seen as inadequate [4].They often miss microplastics, are labor-intensive, cover limited areas, and might even harm marine ecosystems [2] [5].Additionally, their effectiveness wanes in the face of adverse weather conditions, and they aren't scalable for the vastness of our oceans' pollution problem [1].
The autonomous green sea-cleaner boat marks a significant stride in maritime environmental conservation, embodying the synthesis of sustainability and advanced technology [6][7].This vessel stands out for its eco-friendliness, utilizing renewable energy sources to significantly reduce its carbon footprint, a stark contrast to traditional fossil fuel-dependent boats [4] [5].It's designed for cleaner and more sustainable maritime operations, featuring sophisticated autonomous systems equipped with sensors and artificial intelligence for precise marine debris detection and collection [6][8].The efficiency of this autonomous vessel is unmatched; it can navigate vast oceanic territories tirelessly, far surpassing the limited scope and capabilities of conventional cleanup methods [5].
In this paper, we present a comprehensive approach that involves parametric studies of various catamaran hull form configurations, particularly focusing on the flat inside and flat outside hull form models, with the primary objective of identifying the most effective design for reducing wake-wash motion in various marine conditions.To achieve this, our study entails a rigorous evaluation of three distinct vessel designs through advanced Computational Fluid Dynamics (CFD) simulations, enabling us to select the most promising design based on hydrodynamic performance.Furthermore, our approach extends to the integration of advanced robotics, artificial intelligence, and eco-friendly technologies for autonomous marine pollutant extraction vessels, highlighting the importance of efficient and sustainable solutions for addressing marine pollution.We also develop an autonomous system tailored to the chosen vessel design, ensuring optimized pollutant extraction in diverse marine conditions, thus contributing significantly to environmental conservation and ocean preservation.

Metodhology
This study begins by applying CFD to evaluate the hydrodynamic performance of the seacleaner boat with various hull form configurations.Basically hull configurations with reduced wake-wash will be selected for incorporation with the autonomous system.In the context of integrating autonomous capabilities into sea-cleaner boats for improved marine debris management, Convolutional Neural Networks (CNNs) have a fundamental role.CNNs are a specialized class of deep learning models designed to analyze visual data.Their ability to automatically learn and extract distinctive features from images, such as shapes, colors, and textures, makes them a crucial tool for marine debris detection, classification, and evaluation.Through the application of convolution layers, CNNs identify relevant patterns and features in the marine environment, while pooling layers down-sample data to focus on key areas of interest.Furthermore, CNNs excel in classifying detected debris into different categories, enabled by extensive training on labeled data.By incorporating CNNs into sea-cleaner boat systems, we empower these vessels with the capability to detect, classify, and evaluate marine debris in real-time, contributing to more effective and efficient debris removal in marine ecosystems.

Fig. 2. Detailed autonomous system
A detailed representation of the autonomous system's development is presented in Fig. 2. The development of the autonomous simulation system for the sea-cleaner boat includes several essential components.The boat model is configured based on the proposed design, focusing on reducing wake-wash attributes and assessing the feasibility of autonomous systems in challenging operational conditions.Realistic environmental simulation is vital to ensure the system functions effectively in authentic conditions, including operational areas and weather scenarios.Currently, we are configuring a simulated river environment for the testing.Furthermore, advanced visual equipment is integrated for marine debris detection, accompanied by LiDAR sensors for precise mapping and navigation.The sensors on the simplified ship models are configured to mimic the sensor suite that would be present on the real-world vessels.Sensors such as cameras and Lidar system are set up to provide crucial data for autonomous navigation, obstacle detection, and pollutant identification within the simulated environment.The autonomous control algorithm is a critical element, serving as the brain behind ship models in navigating a simulated marine environment.It integrates data from sensors, including CNN-based cameras and LiDAR, for informed decision-making, obstacle avoidance, pollutant detection, and efficient performance.Ongoing evaluation and data analysis ensure system improvement and adaptability to changing conditions.

Simulation Condition
The sea-cleaner model in Fig. 3 was initially designed by PT.Bahtera Tangguh Indonesia (BTI).The current design exhibits increased wake-wash characteristics, which can significantly harm the environment, particularly in constrained water areas like coastal regions and rivers.Consequently, we have designed two more models: The Flat-outside Model and the Flat-Inside Model, with the aim of achieving a significant reduction in wakewash as depicted in Fig 4 .we perform CFD simulations to determine the more favourable configuration, as indicated by reduced wake-wash.Ultimately, we intend to integrate the improved sea-cleaner boat design with an autonomous system to enhance its effectiveness in cleaning marine debris within specified regions.

Computational Domain and Boundary Setting Conditions
Table 2 displays the CFD simulation setup, which provides information on the positioning and types of boundary conditions in relation to the sea-cleaner boat's centre of gravity (COG).This configuration is designed to capture the fundamental flow characteristics of the boat's wake-wash across the three boat models that the CFD has the capability to capture this particular phenomenon.

4
Equations and mathematics

CFD Simulation
The predictions of the total ship resistance of the three model of the sea-cleaner boats are presented in Table 3 and Table 4.The results showed that when the forward velocities were raised from 8 knots to 10 knots, there was a substantial increase in total ship resistance, possibly increasing by a factor of two compared to the 8-knot speed.Among the three distinct hull form models, there is no noteworthy impact on the total ship resistance, as their values are nearly identical.This implies that the FIM and FOM models have a negligible impact on the total resistance of the ship when compared to the existing hull form design.As usual, the total ship resistance data is important in determining the necessary engine power.According to the prediction results, it is evident that the existing sea-cleaner boat model, with a total resistance of 10517.302N, requires approximately 72.551 Horse Power (HP).The specific predicted power requirements for each of the sea-cleaner models can be found in Table 5.
The characteristics of the wave-elevations for each model of the sea-cleaner boat are displayed in Figs. 5 to 8. The results indicated that the wave elevation of the Flat-Outside Model is the lowest, even in comparison to the existing model and the Flat-Inside Model as clearly depicted in Figs. 5 and 6.Inherently, this indicates a reduced hydrodynamic force acting in the direction of the outer hull form, as seen in Fig. 7.As a result, the magnitude of the velocity in the direction of x (forward velocity) and y (transverse velocity) has significantly reduced.The CFD simulation clearly captured this characteristic phenomenon, as illustrated in Fig. 8.The characteristic of the FOM in the sea-cleaner boat renders it a more ecologically sustainable option.Inherently, this results in a reduction in wake-wash motion, which is highly significant due to the potential for excessive wake turbulence to induce erosion and harm sediment habitats [9].In the worst scenario, it results in elevated water turbidity, which poses a threat to marine life.Increased turbidity hinders light penetration, adversely affecting the feeding and reproductive behaviour of various species.This means that the lower wake scouring motion of the Flat-Out Model of the sea-cleaner boat is a favourable characteristic for mitigating environmental damage and preserving aquatic ecosystems.

Navigation
The simulation commences with the sea-cleaner boat positioned at a designated point within the river, resembling the scenario illustrated in Fig. 9. Before embarking on its journey, the vessel deploys the LiDAR system for obstacle detection, ensuring its safe passage without encountering any obstructions.Upon confirming obstacle avoidance, the boat proceeds towards an unobstructed area.Simultaneously, the on-board camera is activated to scan for the presence of debris, as exemplified in Fig. 9(a).In the event of a successful trash detection within its field of view, the boat alters its course and navigates to the identified trash location as depicted Fig. 9(b).Upon reaching the garbage site as seen in Fig. 9(c), the boat comes to a halt to facilitate the collection of debris.

Object Detection
The utilization of a camera paired with a Convolutional Neural Network (CNN) greatly enhances the sea-cleaner boat's object detection and classification abilities.As illustrated in Figures 12 and 13, the system demonstrates its capacity to accurately classify diverse objects.For example, it can identify a bottle as "red" and a pole as "yellow."This object classification not only equips the autonomous boat with the capability to discern which items are suitable for collection as marine debris but also ensures it can effectively avoid collisions with obstacles such as poles, thus enhancing its operational safety and efficiency.

Conclussion
In conclusion, the adoption of the Flat-outside Model (FOM) for catamaran hull geometry stands out as a significant advancement in marine vessel design, specifically engineered to mitigate wake-wash motion.This feature is crucial in preserving the delicate balance of marine ecosystems and safeguarding the integrity of coastal regions from the erosive effects of conventional hull designs.The FOM design leads the path to more sustainable marine transportation by minimizing the ecological impact of nautical operations.
The integration of advanced Artificial Neural Network (ANN) and state-of-the-art Convolutional Neural Network (CNN) technologies signifies a groundbreaking achievement in maritime innovation, as it enables autonomous marine debris detection, advancing the cause of cleaner oceans and showcasing the potential of intelligent technologies in environmental conservation.

Fig. 1 .
Fig. 1.Methodology of development of autonomous sea-cleaner boat

Fig. 9 .
Fig. 9. Navigational performance of the sea-cleaner boat during operation

Table 1 .
Dimension of the boat

Table 2 .
Domain and boundary condition

Table 3 .
Domain and boundary condition

Table 4 .
Coefficient of total ship's resistance prediction