A CDC–ANFIS-Based Model for Assessing Ship Collision Risk in Autonomous Navigation

Optimizing anti-collision strategy for MASS: A safe reinforcement learning approach to improve maritime traffic safety

In this paper, an Adaptive Neuro-Fuzzy Inference System (ANFIS) is proposed for low-cost navigation of ground vehicles. The ANFIS structure is proposed for fusing the measurements of the Global Positioning System (GPS) and inertial measurement unit (IMU). Low-cost GPS receivers provide the position and velocity of the vehicle at a lower rate than the IMU sampled vehicle dynamics. Additionally, the GPS measurements might be missed/delayed due to receiver's inability to lock on the signal or due to obstruction from neighboring vehicle's or infrastructures. Therefore, the variable delay between the IMU and GPS signals is selected as an additional input to the ANFIS architecture. The performance of the proposed ANFIS architecture is demonstrated through experimental results. The instrumentation system used in this experiment consists of a microcontroller system, a GPS receiver, and an IMU.

A novel ship collision risk evaluation algorithm based on the maximum interval of two ship domains and the violation degree of two ship domains

Accurate forecasting of potential collision risk in dense traffic is addressed by a framework grounded in multi-vehicle trajectory prediction. A spatio-temporal fusion architecture, STGAT-EDGRU, is proposed. A Transformer encoder learns temporal motion patterns from each vehicle’s history; a boundary-aware graph (GAT) attention network models inter-vehicle interactions; and a Gated Multimodal Unit (GMU) adaptively fuses the temporal and spatial streams. Future positions are parameterized as bivariate Gaussians and decoded by a two-layer GRU. Using probabilistic trajectory forecasts for the main vehicle and its surrounding vehicles, collision probability and collision intensity are computed at each prediction instant and integrated via a weighted scheme into a Collision Risk Index (CRI) that characterizes risk over the entire horizon. On HighD, for 3–5 s horizons, average RMSE reductions of 0.02 m, 0.12 m, and 0.26 m over a GAT-Transformer baseline are achieved. In high-risk lane-change scenarios, CRI issues warnings 0.4–0.6 s earlier and maintains a stable response across the high-risk interval. These findings substantiate improved long-horizon accuracy together with earlier and more reliable risk perception, and indicate practical utility for lane-change assistance, where CRI can trigger early deceleration or abort decisions, and for risk-aware motion planning in intelligent driving.

Due to the inherent highly nonlinear vehicle state error dynamics obtained from low-cost inertial navigation system (INS) and global positioning system (GPS) along with the unknown statistical properties of these sensors, the optimality/accuracy of the classical Kalman filter for sensor fusion is not guaranteed. Therefore, in this paper, low-cost INS/GPS measurements integration is optimized based on different Neural Networks (NN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) architectures to achieve high-accuracy vehicle state estimates. The proposed approaches involve the use of NN based architectures as well as ANFIS architectures with overlapping windows for delayed input signals. Both the NN approaches and the ANFIS approaches are used once with overlapping position windows as the input and once with overlapping position and velocity windows as the input. Experimental tests are conducted to evaluate the accuracy of the proposed AI approaches. The obtained results are presented and discussed. The study concludes that using ANFIS, with both position and velocity as input, provides the best estimates of position and velocity in the navigation system. This Input Delayed ANFIS (IDANFIS) approach is further analyzed at the end of the paper.

Regional modeling of the ionosphere using adaptive neuro-fuzzy inference system in Iran

In order to study the Automatic Collision-Avoidance (ACA) actions of vessels in overtaking situation accord with rules of International Regulation for Preventing Collision at Sea (COLREGS 72) and ordinary practice of seaman, stags of vessel’s encounter process were separated by Collision Risk Index (CRI) and Time to Immediate Danger (TID).The reason is correct Collision-Avoidance (CA) actions in different stages are different. CRI and TID are derived from defined Situation Elements (SE). Then computing models of them and ACA actions in different stages were presented. The basis of research is center moved ship’s ellipse domain theories and MMG three-free dimensional digital model that is calculated from ship’s hydrodynamic equations. The following were proved by simulations: The mathematic models presented ensured rapid and reliable convergence of SE computing; new CRI model is much more accord with the thinking process of navigators and practice at sea than before; ACA action accord with COLREGS rules and ordinary practice of seaman, can be produced by program. The research may make big advance to the final achievement of ACA of ship in overtaking situation.

Considering that collision accidents happen sometimes, it is necessary to predict the collision risk to ensure navigation safety. With the information construction in maritime and the popularity of automatic identification system application, it is more convenient to obtain ship navigation dynamics. How to obtain ship encounter dynamic parameters through automatic identification system information, assess ship collision risk, find out dangerous target ships, and give early warning and guarantee for ship navigation safety, is a problem that scholars have been studying. As an index to measure the degree of ship collision risk, CRI, namely, collision risk index, is usually obtained by calculating ship encounter parameters and comprehensive analysis. There are many factors that affect CRI, and the values of many parameters depend on expert judgment. The corresponding CRI has nonlinear and complex characteristics, which is highly correlated with the time sequence. In order to enhance the prediction accuracy and efficiency, PSO-LSTM neural network is applied in the paper to predict CRI. Experiments show that PSO-LSTM neural network can effectively predict collision risk and provide a reference for navigation safety.

Ship predictive collision avoidance method based on an improved beetle antennae search algorithm

The integrated navigation system ensures maritime autonomous surface ships (MASSs) to safely, efficiently, and autonomously complete various operations in different complex navigation environments. Investigating robust algorithms for integrated navigation is crucial for enhancing the fault tolerance of the system and ensuring the stable and continuous output of the ship’s motion state. However, existing research primarily focused on optimising particular filtering algorithms or examining the foundations of information allocation within a predetermined integrated navigation structure. As such, strategies for enhancing the robustness of the MASS integrated navigation system and the design of subsystems for federated filters in the event of navigation sensor failures have not been sufficiently investigated for complex maritime navigation scenarios. Consequently, this research introduces an observability sharing factor accounting for both system characteristics and state estimation performance in integrated navigation systems, employing nonlinear sampling filtering. Subsequently, a robust integrated navigation framework with distributed federal filter is developed. Within this framework, an adaptive federated filtering integrated navigation algorithm is proposed based on the observability sharing factor to allocate information in federated filtering. Finally, both the theoretical correctness and effectiveness of the algorithm were verified through simulations and real-ship experiments to assist with the development of accurate and fault-tolerant maritime navigation systems.

Due to the inherent highly nonlinear vehicle state error dynamics obtained from low-cost inertial navigation system (INS) and Global Positioning System (GPS) along with the unknown statistical properties of these sensors, the optimality/accuracy of the classical Kalman filter for sensor fusion is not guaranteed. Therefore, in this paper, low-cost INS/GPS measurement integration is optimized based on different artificial intelligence (AI) techniques: Neural Networks (NN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) architectures. The proposed approaches are aimed at achieving high-accuracy vehicle state estimates. The architectures utilize overlapping windows for delayed input signals. Both the NN approaches and the ANFIS approaches are used once with overlapping position windows as the input and once with overlapping position and velocity windows as the input. Experimental tests are conducted to evaluate the performance of the proposed AI approaches. The achieved accuracy is presented and discussed. The study finds that using ANFIS, with both position and velocity as input, provides the best estimates of position and velocity in the navigation system. Therefore, the dynamic input delayed ANFIS approach is further analyzed at the end of the paper. The effect of the input window size on the accuracy of state estimation is also discussed.

The primary function of a collision risk index is to determine the time when ships take action to avoid a collision. In this study, based on the complex non‐linear relationship between the collision risk degree and its influencing factors, classification and regression trees (CARTs) are applied to construct a prediction model for ship collision risk. The fuzzy comprehensive evaluation method is used to evaluate the risk of ship encounter samples to build a collision risk identification library containing expert collision avoidance experience information. The authors’ proposed CART regression model is trained using the samples in this identification library to develop a collision risk prediction model based on the CART. Their experimental results show that their proposed CART prediction model is better that the existing ship collision risk prediction model in terms of prediction accuracy and prediction speed when the feature dimension is low and the sample size is small.

Versions of this paper were first presented at the Royal Institute of Navigation GNSS Vulnerabilities and Solutions Conference held at Baska, Croatia in September 2008 and the Royal Institute of Navigation NAV 08 Conference held at Church House, Westminster, London in October 2008.The US Global Positioning System (GPS) is currently the primary source of Position, Navigation and Timing (PNT) information in maritime applications, whether stand-alone or augmented with additional systems. This situation will continue in the future with GPS, possibly together with other Global Navigation Satellite Systems (GNSS) e.g. Galileo, being the core PNT technology for e-Navigation – the future digital maritime architecture. GPS signals, measured at the surface on the Earth, are very weak. As such, the system is vulnerable to unintentional interference and jamming, resulting in possible denial of service over large geographical areas. The result of such interference could be the complete failure of the mariner's GPS receiver or, possibly worse, the presentation to the mariner of hazardously misleading information (HMI) for navigation and situational awareness, depending on how the GPS receiver reacts to the jamming incident. Recognising this, the General Lighthouse Authorities of the United Kingdom and Ireland (GLA), in collaboration with the UK Ministry of Defence (MOD) Defence Science and Technology Laboratory (DSTL), have conducted a series of sea-trials with the aim of identifying the full effects of GPS jamming on safe navigation at sea.This paper presents the key findings of these trials and provides important information on the effect of GPS denial. The GLA are playing a pivotal role in the establishment of eLoran as an independent source of PNT, taking advantage of eLoran's complementary nature, having dissimilar failure modes to GPS and the future GNSS. This paper provides information on the performance of an eLoran receiver in an area of GPS service denial. The paper presents the rationale for the work, details the system architecture employed, the data gathering efforts and finally the data analysis procedures, results and conclusions.

Intelligent data fusion system for predicting vehicle collision warning using vision/GPS sensing

A need exists to safely integrate Unmanned Aircraft Systems (UAS) into the United States' National Airspace System. Replacing manned aircraft's see-and-avoid capability in the absence of an onboard pilot is one of the key challenges associated with safe integration. Sense-and-avoid (SAA) systems will have to achieve yet-to-be-determined required separation distances for a wide range of encounters. They will also need to account for the maneuver performance of the UAS they are paired with. The work described in this paper is aimed at developing an understanding of the trade space between UAS maneuver performance and SAA system performance requirements, focusing on a descent avoidance maneuver. An assessment of current manned and unmanned aircraft performance was used to establish potential UAS performance test matrix bounds. Then, near-term UAS integration work was used to narrow down the scope. A simulator was developed with sufficient fidelity to assess SAA system performance requirements. The simulator generates closest-point-of-approach (CPA) data from the wide range of UAS performance models maneuvering against a single intruder with various encounter geometries. Initial attempts to model the results made it clear that developing maneuver performance groups is required. Discussion of the performance groups developed and how to know in which group an aircraft belongs for a given flight condition and encounter is included. The groups are airplane, flight condition, and encounter specific, rather than airplane-only specific. Results and methodology for developing UAS maneuver performance requirements are presented for a descent avoidance maneuver. Results for the descent maneuver indicate that a minimum specific excess power magnitude can assure a minimum CPA for a given time-to-go prediction. However, smaller amounts of specific excess power may achieve or exceed the same CPA if the UAS has sufficient speed to trade for altitude. The results of this study will support UAS maneuver performance requirements development for integrating UAS in the NAS. The methods described are being used to help RTCA Special Committee 228 develop requirements.