Content uploaded by Ke Zhang
Author content
All content in this area was uploaded by Ke Zhang on Jan 13, 2025
Content may be subject to copyright.
A real-time multi-ship collision avoidance decision-making system for autonomous ships considering ship motion uncertainty
Abstract
Ship autonomous collision avoidance has attracted increasing attention in recent years. However, more attention is paid to the scenario in which the target ship keeps its course and speed. Less attention is paid to the development of a collision avoidance decision-making system under the uncertainty of the target ship's movement in a complex multi-ship encounter situation. Based on the idea of model predictive control (MPC), this paper proposes an autonomous ship collision avoidance decision-making system (CADMS) suitable for the uncertainty of ship motion. The CADMS includes four modules: collision risk analysis module, control and execution module, ship trajectory prediction module and collision avoidance decision-making model. The proposed model can be implemented in the collision avoidance decision-making system for safe navigation or it can be included in the ship autonomous navigation process. The decision-making model is achieved from a risk identification-motion prediction-ship control-scheme implementation perspective, and the dynamic and uncertainty features of the ship action (i.e., alter course or change speed) are considered in the modelling process. The real-time rolling update of ship collision avoidance decisions is realised based on the time series rolling method. Four scenarios are designed to demonstrate the collision avoidance decision-making system's performance. The results show that the proposed collision avoidance decision-making system is a reasonable and effective system for collision avoidance, particularly in multi-ship encounter situations of target ships suddenly altering course or change speed.
... Rong et al. (2019) decomposes ship motion prediction into lateral, where lateral motion uncertainty is described by a data-driven non-parametric Bayesian model based on a Gaussian Process or the longitudinal uncertainty results from the uncertainty on the ship acceleration along the route. Liu et al. (2022) use a convolutional LSTM network to predict spatial -temporal collision risks, while Zhang et al. (2023) address the uncertainty of ship motion in decision-making systems. Although these methods incorporate some aspects of uncertainty, they often fail to dynamically adapt to real-time changes in ship behaviours, especially in high-density traffic environments. ...
... Considering the COLREGs and good seamanship, the model for the classification of ship encounter situations is presented refer to (Zhang et al. 2023). As shown in Table 1, the Time to the Closest Point of Approach (TCPA) indicates how soon the two ships will reach their closest point if no action is taken. ...
... As demands on the operational environment and shipping efficiency continue to grow, intelligent ships must not only possess advanced autonomous navigation capabilities 2 of 21 but also be capable of perceiving, predicting, and making real-time decisions in complex maritime environments [4]. This requires efficient processing of multi-source data and a deep understanding of dynamic navigation conditions [5]. ...
... In converting the course to vector components, a northward direction is taken as 0 degrees, increasing clockwise. Subsequently, the relative position vector between the vessels is calculated as specified in Equation (4). Using these vectors, TCPA can be computed by the formula shown in Equation (5), which calculates the time point when the rate of change in relative position between the vessels is at its minimum. ...
Intelligent ships are a key focus for the future development of maritime transportation, relying on efficient decision-making and autonomous control within complex environments. To enhance the perception, prediction, and decision-making capabilities of these ships, the present study proposes a novel approach for constructing a time-series knowledge graph, utilizing real-time Automatic Identification System (AIS) data analyzed via a sliding window technique. By integrating advanced technologies such as knowledge extraction, representation learning, and semantic fusion, both static and dynamic navigational data are systematically unified within the knowledge graph. The study specifically targets the extraction and modeling of critical events, including variations in ship speed, course changes, vessel encounters, and port entries and exits. To evaluate the urgency of encounters, mathematical algorithms are applied to the Distance to Closest Point of Approach (DCPA) and Time to Closest Point of Approach (TCPA) metrics. Furthermore, the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering algorithm is employed to identify suitable docking berths. Additionally, multi-source meteorological data are integrated with ship dynamic data, providing a more comprehensive representation of the maritime environment. The resulting knowledge system effectively combines ship attributes, navigational status, event relationships, and environmental factors, thereby offering a robust framework for supporting intelligent ship operations.
... Their experimental results demonstrate that both methods could generate safe, efficient, and reliable collision avoidance strategies in both time-sequenced dynamic obstacle and mixed environments. Zhang et al. (2023) proposed an autonomous Collision Avoidance Decision-Making System (CADMS) for ships. The system is realized from the perspective of risk identificationmotion prediction-ship control-scheme implementation, and the dynamic and uncertain characteristics of ship actions (i.e., changing course or changing speed) are considered in the modeling process to realize real-time rolling updates of ship collision avoidance decisions. ...
The Artificial Potential Field (APF) algorithm has been widely used for collision avoidance on unmanned ships. However, traditional APF methods have several defects that need to be addressed. To ensure safe navigation with good seamanship and full compliance with the Convention on the International Regulations for Preventing Collisions at Sea, 1972 (COLREGS), this study proposes a dynamic collision avoidance method based on the APF algorithm. The proposed method incorporates a ship domain priority judgment encounter situation, allowing the algorithm to perform collision avoidance operations in accordance with actual operational requirements. To address path interference and unreachable target issues, a new attractive potential field function is introduced, dividing the attractive potential field of the target point into multiple segments simultaneously. Additionally, the repulsive force on the own ship is reduced when close to the target point. The results show that the proposed method effectively resolves path oscillation problems by integrating the potential field based on traditional APF with partial ideas from the Dynamic Window Approach (DWA). In comparison with traditional APF algorithms, the overall smoothing degree was improved by 71.8%, verifying the effectiveness and superiority of the proposed algorithm.
... Incorporating realtime environmental and operational data, such as wave, wind, currents, and maritime traffic, will be prioritized to improve prediction accuracy. Advanced ship maneuvering prediction models (Gil et al., 2024;Zhang et al., 2023a) can be further explored by employing hybrid deep learning techniques, enabling the generation of precise and safe maneuvering commands while accounting for complex multi-ship interactions under real-world operational conditions. The integration of these models is expected to yield a robust system capable of accurately predicting collision probabilities and damage consequences based on realistic ship encounter situations at sea (Montewka et al., 2010;Zhang et al., 2021a,b). ...
Ship collisions can result in catastrophic outcomes, necessitating effective real-time collision risk assessment methods for proactive risk management. These methods need to rapidly evaluate both the probability of collision and the potential damage dimensions (length, height, and penetration) in real conditions. Existing frameworks often underestimate collision damage consequences during operational risk assessments. This paper presents a hybrid deep learning approach for the real-time prediction of collision damage dimensions under real ship operation conditions. Collision scenarios are identified using Automatic Identification System (AIS) data, with damage extents simulated through the Super Element (SE) method. A comprehensive database of collision scenarios and corresponding damage assessments is developed, sourced from realistic operational data of Ro-Pax ship in the Gulf of Finland. The deep learning model is trained and validated using this dataset, ensuring the model's relevance and practical applicability. Extensive comparative analyses and generalization tests demonstrate the high accuracy of the model in predicting ship collision damages in diverse ship operational conditions. In addition, traditional simulation methods for evaluating damage dimensions require approximately 10 min, whereas the trained deep learning model reduces the time to less than 0.1 s, enabling real-time potential collision consequence assessment in real operational conditions. The proposed model may provide significant insights for ship operators, enhancing ship safety and supporting intelligent decision-making in ship operations.
... In the context of ship trajectory prediction, the future motion characteristics of a ship are intricately linked to its historical and present motion patterns. Within the framework of long short-term memory (LSTM), the historical motion patterns of a ship can be encapsulated through cell states, thereby establishing a long-term memory [13][14][15][16][17]. The LSTM architecture employs sigmoid and tanh activation functions for its two input gates, where input data passes through these gates to extract features that are subsequently directly combined and fed into the cell state [18]. ...
Due to the rapid economic development of modern society, the demand for cargo in the shipping industry has experienced unprecedented growth in recent years. The introduction of a large number of ships, especially large, new, and intelligent ships, has made shipping networks more complex. Controlling transportation risks has become more challenging than ever before. Ship trajectory prediction based on automatic identification system (AIS) data can effectively help identify abnormal ship behaviors and reduce maritime risks such as collisions, grounding, and contacts. In recent years, with the rapid development of deep learning theories, recurrent neural network models (long short-term memory and gated recurrent unit) have been widely used in ship trajectory prediction due to their powerful ability to capture hidden information in time-series data. However, these models struggle with tasks involving high complexity of trajectory features. To address this issue, this paper introduces a switching-input mechanism based on LSTM, constructing a ship trajectory prediction model based on the SI-LSTM model. The switching-input mechanism enables the model to adjust its processing of important information according to dynamic changes in input data, effectively capturing local features of complex trajectories. The experimental section, which includes eight cases of complex trajectories, demonstrates the competitive generalization ability and prediction accuracy of SI-LSTM.
... These methods can be broadly classified into two categories: traditional methods and advanced methods based on machine learning and deep learning. To better understand the main findings in the field and identify the Traditional methods integrate physical models and data-driven models for prediction, encompassing the Kalman filter (KF) [16,17], Gaussian mixture model (GMM) [18], support vector machine (SVM) [19][20][21], and other technologies to enhance prediction accuracy. Due to the restricted capacity of these methods in handling uncertain factors, nonlinear motion, and non-static characteristics, traditional methods relying solely on motion equations are unable to achieve precise predictions. ...
Accurate ship trajectory prediction is crucial for real-time vessel position tracking and maritime safety management. However, existing methods for ship trajectory prediction encounter significant challenges. They struggle to effectively extract long-term and complex spatial–temporal features hidden within the data. Moreover, they often overlook correlations among multivariate dynamic features such as longitude (LON), latitude (LAT), speed over ground (SOG), and course over ground (COG), which are essential for precise trajectory forecasting. To address these pressing issues and fulfill the need for more accurate and comprehensive ship trajectory prediction, we propose a novel and integrated approach. Firstly, a Trajectory Point Correlation Attention (TPCA) mechanism is devised to establish spatial connections between trajectory points, thereby uncovering the local trends of trajectory point changes. Subsequently, a Temporal Pattern Attention (TPA) mechanism is introduced to handle the associations between multiple variables across different time steps and capture the dynamic feature correlations among trajectory attributes. Finally, a Great Circle Route Loss Function (GCRLoss) is constructed, leveraging the perception of the Earth’s curvature to deepen the understanding of spatial relationships and geographic information. Experimental results demonstrate that our proposed method outperforms existing ship trajectory prediction techniques, showing enhanced reliability in multi-step predictions.
... Over the years, methods allowing for the automatization of the different tasks performed during the ship operation, such as navigation and maneuvering, mooring and anchoring, have been developed. Recent approaches for solving the collision avoidance problem of autonomous ships include the application of the Velocity Obstacle (VO) algorithm combined with the Model Predictive Control (MPC) [1], the algorithm utilizing the isochrone method [2] and many approaches based on machine learning, such as the Deep Reinforcement Learning (DRL) [3], the Inverse Reinforcement Learning (IRL) [4], Multi-Agent Reinforcement Learning (MARL) [5] and fuzzy logic [6], [7]. Other recently proposed methods were based on swarm intelligence, such as the Beetle Antennae Search (BAS) algorithm [8] and the hybrid method based on the Artificial Potential Field (APF) and the Ant Colony Optimization (APF-ACO) algorithm [9]. ...
The paper presents the results of a research on the development of algorithms for ship’s safe ship trajectory calculation and automatic control of the ship along the determined trajectory. These methods are intended for use in the autonomous navigation system of an Unmanned Surface Vehicle (USV) or a Maritime Autonomous Surface Ship (MASS). The work presents the consecutive stages of a research that must be carried out in order to develop a system that will perform the task of multidimensional ship control in a port. The safe trajectory is calculated with the use of the Ant Colony Optimization (ACO) method. The ship motion control is based on the Linear Matrix Inequalities (LMI) method, implemented using the Mathworks and Yalmip libraries. The results of controlling the ship’s motion in accordance with the calculated trajectory are presented in the paper. With the use of the developed system, the ship can move autonomously based on the information from the DGPS system and a gyrocompass. The presented results concerned a computer simulation of maneuvers in the port of the Blue Lady ship from the Foundation for Shipping Safety and Environmental Protection.
... Over the years, methods allowing for the automatization of the different tasks performed during the ship operation, such as navigation and maneuvering, mooring and anchoring, have been developed. Recent approaches for solving the collision avoidance problem of autonomous ships include the application of the Velocity Obstacle (VO) algorithm combined with the Model Predictive Control (MPC) [1], the algorithm utilizing the isochrone method [2] and many approaches based on machine learning, such as the Deep Reinforcement Learning (DRL) [3], the Inverse Reinforcement Learning (IRL) [4], Multi-Agent Reinforcement Learning (MARL) [5] and fuzzy logic [6], [7]. Other recently proposed methods were based on swarm intelligence, such as the Beetle Antennae Search (BAS) algorithm [8] and the hybrid method based on the Artificial Potential Field (APF) and the Ant Colony Optimization (APF-ACO) algorithm [9]. ...
Despite advancements in encryption technology, the issue of covert data transmission remains relevant. Researchers are actively enhancing established methods of concealing information with software tools that utilize mathematical models and developing new data entry methods. Steganographic techniques can be used with cryptographic methods to enhance the confidentiality of information within a container or as a standalone software tool. This article examines the steganographic integrity of concealing confidential information within audio files. The steganographic technique was selected as the phase coding method, which involves substituting the audio element’s phase with the secret message’s relative phase. An implementation of the chosen method was developed for the experiment. During the experiment, secret data of varying lengths were used to assess the impact of the length of the concealed information on the distortion of the audio file that may be perceptible to the listener. The study also investigated the resilience of the concealed information to noise attacks, allowing for the determination of the threshold of its robustness. The findings provide insight into the reliability of this steganographic method for hiding classified information within media objects.
This chapter explores the risk assessment for Maritime Autonomous Surface Ships (MASS), highlighting the balance between reducing human error and introducing new risks from autonomous technology. It argues the limitations of traditional risk assessment methods like Failure Mode and Effect Analysis (FMEA), Hazard and Operability Studies (HAZOP), Fault Tree Analysis (FTA), and Event Tree Analysis (ETA), which may not adequately address the complexities of MASS. To overcome those limitations, the chapter introduces novel methodologies such as System-Theoretic Process Analysis (STPA) and Real-time Risk Analysis, offering dynamic and comprehensive approaches to risk assessment in autonomous maritime operations. Additionally, this chapter reviews the international guidelines and the development of the MASS Code by the International Maritime Organization (IMO), emphasizing a flexible regulatory framework to accommodate various risk assessment techniques. This flexibility is expected to promote the use of advanced risk assessment methods or the development of tailor-made risk assessment methods for MASS operations.
A novel real-time collision avoidance method for autonomous ships based on modified velocity obstacle (VO) algorithm and grey cloud model is proposed. A typical VO algorithm is used to judge whether there is a collision risk for ships in the potential collision area (PCA). Then, in order to quantify the collision risk of ships in different encounter situations within the PCA and trigger a prompt warning of danger of collision, this study sets up a novel collision risk assessment method based on asymmetric grey cloud model (AGC). It can effectively consider the randomness, ambiguity, and incompleteness of the information in the ship collision risk evaluation process. Moreover, reachable collision-free velocity sets under different encounter situations and optimal steering angle model are constructed. A real-time collision avoidance method based on modified VO algorithm and manoeuvring motion characteristics of vessels is put forward. In this model, various constraints are considered including the International Regulations for Preventing Collisions at Sea (COLREGs), ship manoeuvrability, and ordinary practice of seaman. Finally, several case studies are carried out to verify the performance and reliability of the collision avoidance model. The results show that the proposed method can not only effectively identify and quantify the collision risk in real-time but also offer proper collision-free solutions for autonomous ships.
Collision prevention is critical for navigational safety at sea, which has developed rapidly in the past decade and attracted a lot of attention. In this article, an improved velocity obstacle (IVO) algorithm for intelligent collision avoidance of ocean-going ships is proposed in various operating conditions, taking into count both a ship’s manoeuvrability and Convention on the International Regulations for Preventing Collisions at Sea (COLREGs). An integrated model combines a three-degree-of-freedom manoeuvring model with ship propeller characteristics to provide a precise prediction of ships in various manoeuvring circumstances. In the given case, what is different to present studies, this improved algorithm allows for decision-making in two ways: altering course and changing speed. The proposed technique is demonstrated in a variety of scenarios through simulation. The findings reveal that collision-avoidance decision-making can intelligently avoid collisions with the target ships (TSs) in multi-ship situations.
In the last few years, autonomous ships have attracted increasing attention in the maritime industry. Autonomous ships with an autonomous collision avoidance capability are the development trend for future ships. In this study, a ship manoeuvring process deduction-based dynamic adaptive autonomous collision avoidance decision support method for autonomous ships is presented. Firstly, the dynamic motion parameters of the own ship relative to the target ship are calculated by using the dynamic mathematical model. Then the fuzzy set theory is adopted to construct collision risk models, which combine the spatial collision risk index (SCRI) and time collision risk index (TCRI) in different encountered situations. After that, the ship movement model and fuzzy adaptive PID method are used to derive the ships’ manoeuvre motion process. On this basis, the feasible avoidance range and the optimal steering angle for ship collision avoidance are calculated by deducting the manoeuvring process and the modified velocity obstacle (VO) method. Moreover, to address the issue of resuming sailing after the ship collision avoidance is completed, the Line of Sight (LOS) guidance system is adopted to resume normal navigation for the own ship in this study. Finally, the dynamic adaptive autonomous collision avoidance model is developed by combining the ship movement model, the fuzzy adaptive PID control model, the modified VO method and the resume-sailing model. The results of the simulation show that the proposed methodology can effectively avoid collisions between the own ship and the moving TSs for situations involving two or multiple ships, and the own ship can resume its original route after collision avoidance is completed. Additionally, it is also proved that this method can be applied to complex situations with various encountered ships, and it exhibits excellent adaptability and effectiveness when encountering multiple objects and complex situations.
A novel collision avoidance (CA) algorithm was proposed based on the modified artificial potential field (APF) method, to construct a practical ship automatic CA system. Considering the constraints of both the International Regulations for Preventing Collisions at Sea (COLREGS) and the motion characteristics of the ship, the multi-ship CA algorithm was realized by modifying the repulsive force model in the APF method. Furthermore, the distance from the closest point of approach-time to the closest point of approach (DCPA-TCPA) criterion was selected as the unique adjustable parameter from the perspective of navigation practice. Collaborative CA experiments were designed and conducted to validate the proposed algorithm. The results of the experiments revealed that the actual DCPA and TCPA agree well with the parameter setup that keeps the ship at a safe distance from other ships in complex encountering situations. Consequently, the algorithm proposed in this study can achieve efficient automatic CA with minimal parameter settings. Moreover, the navigators can easily accept and comprehend the adjustable parameters, enabling the algorithm to satisfy the demand of the engineering applications.
Maritime Autonomous Surface Ships (MASSs) are attracting increasing attention in recent years as it brings new opportunities for water transportation. Previous studies aim to propose fully autonomous system on collision avoidance decisions and operations, either focus on supporting conflict detection or providing with collision avoidance decisions. However, the human-machine cooperation is essential in practice at the first stage of automation. An optimized collision avoidance decision-making system is proposed in this paper, which involves risk appetite (RA) as the orientation. The RA oriented collision avoidance decision-making system (RA-CADMS) is developed based on human-machine interaction during ship collision avoidance, while being consistent with the International Regulations for Preventing Collisions at Sea (COLREGS) and Ordinary Practice of Seamen (OPS). It facilitates automatic collision avoidance and safeguards the MASS remote control. Moreover, the proposed RA-CADMS are used in several encounter situations to demonstrate the preference. The results show that the RA-CADMS is capable of providing accurate collision avoidance decisions, while ensuring efficiency of MASS maneuvering under different RA.
As the number of ships for marine transportation increases with the advancement of global trade, encountering multiple ships in marine traffic becomes common. This situation raises the risk of collision of the ships; hence, this paper proposes a novel Fuzzy-logic based intelligent conflict detection and resolution algorithm, where the collision courses and possible avoiding actions are analysed by considering ship motion dynamics and the input and output fuzzy membership functions are derived. As a conflict detection module, the Collision Risk (CR) is measured for each ship by using a scaled nondimensional Distance to the Closest Point of Approach (DCPA) and Time to the Closest Point of Approach (TCPA) as inputs. Afterwards, the decisions for collision avoidance are made based on the calculated CR, encountering angle and relative angle of each ship measured from others. In this regard, the rules for the Fuzzy interface system are defined in accordance with the COLREGs, and the whole system is implemented on the MATLAB Simulink platform. In addition, to deal with the multiple ship encounters, the paper proposes a unique maximum-course and minimum-speed change approach for decision making, which has been found to be efficient to solve Imazu problems, and other complicated multiple-ship encounters.
In this paper, a collision avoidance guidance and control (CAGC) scheme for fully autonomous unmanned surface vehicle (USV) according to the guidance, navigation and control (GNC) architecture is studied. The CAGC scheme is divided into guidance layer and control layer. To avoid the possibility of collision in tracking guidance, the proposed guidance layer needs remove the tracking guidance and realize the direct guidance to the control layer, which is designed according to the theory of finite control set model predictive control (FCS-MPC) with control ability. Next, to balance the contradiction between calculation and guidance accuracy, the methods of delay compensation and control set normal distribution are designed. In addition, in the control layer, the vector propulsion USV as the control object is non-minimum phase system because of its small size and significant sway thrust, so the virtual control point is used to solve this system, and the relationship between virtual control point and vessel speed is determined by zero-dynamics stability analysis. And the control layer adopts adaptive sliding mode control theory to resist the influence of strong disturbances. Finally, simulation results are provided to illustrate the efficacy of the CAGC scheme.
The ability to avoid shorelines, reefs and moving ships or obstacles is the premise of automatic navigation. For achieving the goal, conventional A star algorithm is improved to plan route globally. The safety distance is set between planning route and obstacles. Additionally, the planning route is further optimized to remove unnecessary waypoints, and only vital waypoints are retained for conforming to navigation practice and track keeping control. Besides, minimum course alteration algorithm (MCA) is proposed to avoid moving ships or obstacles constrained by COLREGs. The proposed algorithms are validated in simulation. The results show that the algorithms are credible in two and multi ships encounter situations, even the targets ships take unexpected course alteration and the own ship could well avoid collision with shorelines, reefs and moving ships or obstacles under external disturbance.
Ship motion prediction is applied to the shipboard stabilized platform to keep the equipment on the platform stable all the time, which is of great practical significance to the safety and efficiency of shipboard equipment operation. Long Short-term Memory (LSTM) Network is a classic time series prediction method that has made remarkable achievements in this field. However, the dynamic frequency range of single LSTM in ship motion prediction is insufficient to meet the stabilized platform with higher precision requirements. To improve the performance of LSTM in ship motion prediction, this paper presents a novel method named as multiscale attention-based LSTM. At first, wavelet transform is employed to decompose ship motion signals into several frequency scales, which makes LSTM to capture the inherent law of ship motion from each frequency scale. And then the weights of different scales are obtained by attention mechanism, which promote the sensitivity of the whole system by paying more attention to significant information and suppress the interference of noise signals. Both of the steps form a multiscale attention mechanism, which promote the adaptability and improve the performance of the LSTM. In addition, to avoid being trapped in local optimization, the two-stage training mechanism is designed for model training based on the model structure. Ship motion data are used to evaluate the feasibility and effectiveness. The experiments show that the proposed method achieves better performance compared with other popular methods.
The wider use of electronic devices in shipping has led to the introduction of new navigation control systems and decision support tools for collision avoidance. The application of these systems improves the data utilisation and allows for the prediction of ship's performance in various operational conditions. In this study, a decision support methodology for collision avoidance of ocean going vessels is developed, taking into account both the ship's manoeuvrability and propulsion system performance whilst employing a new formulation for the estimation of the collision probability indicator. An integrated model that simulates the ship's manoeuvrability and propulsion system performance is employed to predict the required ship's hull and propulsion plant dynamic response. The integrated model couples a 3-DOF manoeuvring model with a mean value engine model, providing a fast but accurate prediction of ship and her propulsion system characteristics during specific manoeuvring scenarios. The derived ship trajectories populate a database that is employed as input to the developed decision support for investigating various encounter situations depending on the target ship's initial position, approach angle, trajectory and speed. The developed decision support provides the collision probability indicator as function of the ordered rudder angle and engine speed as the available control options to avoid the collision. This study provides support to the officer of the watch to make decisions on the ship and propulsion system control parameters during encountering situations, thus contributing to the safer maritime operations.