Recognition of obstacle type based on visual sensors is important for navigation by unmanned surface vehicles (USV), including path planning, obstacle avoidance, and reactive control. Conventional detection techniques may fail to distinguish obstacles that are similar in visual appearance in a cluttered environment. This work proposes a novel obstacle type recognition approach that combines a dilated operator with the deep-level features map of ResNet50 for autonomous navigation. First, visual images are collected and annotated from various different scenarios for USV test navigation. Second, the deep learning model, based on a dilated convolutional neural network, is set and trained. Dilated convolution allows the whole network to learn deep features with increased receptive field and further improves the performance of obstacle type recognition. Third, a series of evaluation parameters are utilised to evaluate the obtained model, such as the mean average precision (mAP), missing rate and detection speed. Finally, some experiments are designed to verify the accuracy of the proposed approach using visual images in a cluttered environment. Experimental results demonstrate that the dilated convolutional neural network obtains better recognition performance than the other methods, with an mAP of 88%.

Determination of calibration parameters is essential for the fusion performance of an inertial measurement unit (IMU) and odometer integrated navigation system. Traditional calibration methods are commonly based on the filter frame, which limits the improvement of the calibration accuracy. This paper proposes a graph-optimisation-based self-calibration method for the IMU/odometer using preintegration theory. Different from existing preintegrations, the complete IMU/odometer preintegration model is derived, which takes into consideration the effects of the scale factor of the odometer, and misalignments in the attitude and position between the IMU and odometer. Then the calibration is implemented by the graph-optimisation method. The KITTI dataset and field experimental tests are carried out to evaluate the effectiveness of the proposed method. The results illustrate that the proposed method outperforms the filter-based calibration method. Meanwhile, the performance of the proposed IMU/odometer preintegration model is optimal compared with the traditional preintegration models.

This paper proposes a novel speed optimisation scheme for unmanned sailboats by sliding mode extremum seeking control (SMESC) without steady-state oscillation. In the sailing speed optimisation scheme, an initial sail angle of attack is first computed by a piecewise constant function in the feed forward block, which ensures a small deviation between sailing speed and the maximum speed. Second, the sailing speed approaches to maximum gradually by extremum search control (ESC) in the feedback block. In SMESC without steady-state oscillation, a switching law is designed to carry out the control transformation, so that the speed optimisation system carries out SMESC in the first convergence phase and ESC without steady-state oscillation in the second stability phase. This scheme combines the advantages of both control algorithms to maintain a faster convergence rate and to eliminate steady-state oscillation. Furthermore, the strict stability of the speed optimisation system is proved in this paper. Finally, we test a 12-m mathematical model of an unmanned sailboat in the simulation to demonstrate the effectiveness and robustness of this speed optimisation scheme.

The civilian Automatic Identification System (AIS) has no inherent protection against spoofing. Spoofed AIS messages have the potential to interfere with the safe navigation of a vessel by, amongst other approaches, spoofing maritime virtual aids to navigation and/or differential global navigation satellite system (DGNSS) correction data conveyed across it. Acting maliciously, a single transmitter may spoof thousands of AIS messages per minute with the potential to cause considerable nuisance; compromising information provided by AIS intended to enhance the mariner's situational awareness. This work describes an approach to authenticate AIS messages using public key cryptography (PKC) and thus provide unequivocal evidence that AIS messages originate from genuine sources and so can be trusted. Improvements to the proposed AIS authentication scheme are identified which address a security weakness and help avoid false positives to spoofing caused by changes to message syntax. A channel loading investigation concludes that sufficient bandwidth is available to routinely authenticate all AIS messages whilst retaining backwards compatibility by carrying PKC ‘digital signatures’ in a separate VHF Data Exchange System (VDES) side channel.

To solve the problem of spacecraft attitude manoeuvre planning under dynamic multiple mandatory pointing constraints and prohibited pointing constraints, a systematic attitude manoeuvre planning approach is proposed that is based on improved policy gradient reinforcement learning. This paper presents a succinct model of dynamic multiple constraints that is similar to a real situation faced by an in-orbit spacecraft. By introducing return baseline and adaptive policy exploration methods, the proposed method overcomes issues such as large variances and slow convergence rates. Concurrently, the required computation time of the proposed method is markedly reduced. Using the proposed method, the near optimal path of the attitude manoeuvre can be determined, making the method suitable for the control of micro spacecraft. Simulation results demonstrate that the planning results fully satisfy all constraints, including six prohibited pointing constraints and two mandatory pointing constraints. The spacecraft also maintains high orientation accuracy to the Earth and Sun during all attitude manoeuvres.

Distance measuring equipment (DME/DME) as the main reversionary method provides alternative positioning, navigation and timing (A-PNT) services for use during a Global Navigation Satellite System (GNSS) outage. Considering the geometry limitation of DME/DME, multi-DMEs with better geometry can be used to increase the accuracy and integrity performance of positioning. This paper discusses the opportunities and challenges related to use of multi-DMEs as an alternate source of positioning, navigation and timing. To support the performance for A-PNT, the basic idea is considering the existing installed equipment. In this paper, barometer altimeter and TACAN are used to help improve the performance of A-PNT provided by multi-DMEs both in accuracy and integrity. Based on the database of EUROCONTROL, the test results demonstrate that 79⋅7% of a reference area roughly matching with the continental European locations achieve RNP 1 using multi-DMEs when the DME measurement accuracy is 0⋅2 NM (95%). When the DME measurement accuracy is 0⋅1 NM (95%), 87⋅9% of the reference area can achieve RNP 1 using multi-DMEs. The usage of barometer/TACAN measurements aided multi-DMEs improves the performance of the accuracy and integrity monitoring.

In traditional satellite navigation receivers, the parameters of tracking loop such as loop bandwidth and integration time are usually set in the design of the receivers according to different scenarios. The signal tracking performance is limited in traditional receivers. In addition, when the tracking ability of weak signals is improved by extending the integration time, negative effect of residual frequency error becomes more and more serious with extension of the integration time. To solve these problems, this paper presents out research on receiver tracking algorithms and proposes an optimised tracking algorithm with inertial information. The receiver loop filter is designed based on Kalman filter, reducing the phase jitter caused by thermal noise in the weak signal environment and improving the signal tracking sensitivity. To confirm the feasibility of the proposed algorithm, simulation tests are conducted.

The design of new rules on seaways, such as traffic restrictions, requires determining the degree of improvement in marine traffic safety beforehand by considering the occurrence of new hazardous factors. This study proposes a method to predict the future traffic behaviour and ship encounter frequency (EF) with the introduction of a new traffic rule. First, a sensitivity analysis is conducted to identify the factors affecting the EF. A method of predicting future traffic behaviour and EF is presented based on the analysis of changes in the traffic flow in an area with a temporal restriction. Results show that the method appropriately predicts the location and degree of the occurrence of encounters in the sea area. The proposed method contributes to the discussion of future traffic safety, when sailing in a specific area is restricted by new regulations, installations of new offshore wind farms and fishing reefs.

The regional ship collision risk assessment for multiple ships in restricted waters is of great significance to the early warning of ship collision risk and the intelligent supervision of maritime traffic. Given the existed method of regional ship collision risk assessment without considering the impact of ship aggregation density, this paper proposes a novel regional ship collision risk assessment method that considers the aggregation density (AD) of the clusters of encounter ships (CES) for intelligent surveillance and navigation. The effectiveness of the proposed method has been examined by the experimental case study in the waters of Xiamen, China, and analysis has been compared with other existed studies to show the advantages of the new proposed algorithm. The results show that the study method can more intuitively and effectively quantify the temporal and spatial distribution of regional collision risks in the restricted sea area. The proposed method can improve the efficiency of traffic management when monitoring the ship collision risks in macroscopic view, and assist the safety of manned and unmanned ship navigation.

The paper presents the error characteristics of a vehicle dynamic model (VDM)-based integration architecture for fixed-wing unmanned aerial vehicles. Global navigation satellite system (GNSS) and inertial measurement unit measurements are fused in an extended Kalman filter (EKF) which uses the VDM as the main process model. Control inputs from the autopilot system are used to drive the navigation solution. Using a predefined trajectory with segments of both high and low dynamics and a variable wind profile, Monte Carlo simulations reveal a degrading performance in varying periods of GNSS outage lasting 10 s, 20 s, 30 s, 60 s and 90 s, respectively. These are followed by periods of re-acquisition where the navigation solution recovers. With a GNSS outage lasting less than 60 s, the position error gradually grows to a maximum of 8⋅4 m while attitude errors in roll and pitch remain bounded, as opposed to an inertial navigation system (INS)/GNSS approach in which the navigation solution degrades rapidly. The model-based approach shows improved navigation performance even with parameter uncertainties over a conventional INS/GNSS integration approach.

Following substantial progress achieved recently, the Galileo constellation provides a considerable satellite resource for the GNSS applications. In this regard, the performance assessment of real-time single-frequency precise positioning with Galileo satellites is the main objective of this research. For this purpose, several experimental tests were conducted in this study with two single-frequency positioning models, namely single-frequency code-based positioning and code-phase combination. The results show that Galileo presents an adequate number of visible satellites sufficient for single-frequency positioning. Also, the study demonstrates that, in comparison to GPS observations, Galileo observations have a significantly lower noise level. For the single-frequency code-based positioning, Galileo presents a better positioning accuracy than GPS by 25⋅8% on average. When compared with GPS, a 9⋅4% better positioning accuracy is acquired from Galileo for the single-frequency code-phase combination, with its average convergence time shorter than GPS by a ratio of 24⋅4%.

Celestial navigation is an important means of maritime navigation; it can automatically achieve inertially referenced positioning and orientation after a long period of development. However, the impact of different accuracy of observations and the influence of nonstationary states, such as ship speed change and steering, are not taken into account in existing algorithms. To solve this problem, this paper proposes an adaptively robust maritime celestial navigation algorithm, in which each observation value is given an equivalent weight according to the robust estimation theory, and the dynamic balance between astronomical observation and prediction values of vessel motion is adjusted by applying the adaptive factor. With this system, compared with the frequently used least square method and extended Kalman filter algorithm, not only are the real-time and high-precision navigation parameters, such as position, course, and speed for the vessel, calculated simultaneously, but also the influence of abnormal observation and vessel motion status change could be well suppressed.

Several types of equipment have been developed over the years to assist ship operators with their tasks. Nowadays, navigational equipment typically provides an enormous volume of information. Thus, there is a corresponding need for efficiency in how such information is presented to ship operators. Augmented reality (AR) systems are being investigated for such efficient presentation of typical navigational information. The present work is particularly interested in an AR architecture commonly referred as monitor augmented reality (MAR).

In this context, the development of MAR systems is briefly summarised. The projection of three-dimensional elements into a camera scene is presented. Potential visual assets are proposed and exemplified with videos from a ship manoeuvring simulator and a real experiment. Enhanced scenes combining pertinent virtual elements are shown exemplifying potential assistance applications. The authors mean to contribute to the popularisation of MAR systems in maritime environments. Further research is suggested to define optimal combinations of visual elements for alternative maritime navigation scenarios. Note that there are still many challenges for the deployment of MAR tools in typical maritime operations.

Maritime video surveillance has become an essential part of the vessel traffic services system, intended to guarantee vessel traffic safety and security in maritime applications. To make maritime surveillance more feasible and practicable, many intelligent vision-empowered technologies have been developed to automatically detect moving vessels from maritime visual sensing data (i.e., maritime surveillance videos). However, when visual data is collected in a low-visibility environment, the essential optical information is often hidden in the dark, potentially resulting in decreased accuracy of vessel detection. To guarantee reliable vessel detection under low-visibility conditions, the paper proposes a low-visibility enhancement network (termed LVENet) based on Retinex theory to enhance imaging quality in maritime video surveillance. LVENet is a lightweight deep neural network incorporating a depthwise separable convolution. The synthetically-degraded image generation and hybrid loss function are further presented to enhance the robustness and generalisation capacities of LVENet. Both full-reference and no-reference evaluation experiments demonstrate that LVENet could yield comparable or even better visual qualities than other state-of-the-art methods. In addition, it takes LVENet just 0⋅0045 s to restore degraded images with size 1920 × 1080 pixels on an NVIDIA 2080Ti GPU, which can adequately meet real-time requirements. Using LVENet, vessel detection performance can be greatly improved with enhanced visibility under low-light imaging conditions.

Automatic identification system (AIS)-based ship trajectory data are important for analysing maritime activities. As the data accumulate over time, trajectory compression is needed to alleviate the pressure of data storage, migration and usage. The grating algorithm, as a vector data compression algorithm with high compression performance and low computation complexity, has been considered as a very promising approach for ship trajectory compression. This algorithm needs the threshold to be set for each trajectory which limits the applicability over a large number of different trajectories. To solve this problem, a dynamic adaptive threshold grating compression algorithm is developed. In this algorithm, the threshold for each trajectory is dynamically generated using an effective approaching strategy. The developed algorithm is tested with a complex trajectory dataset from the Qiongzhou Strait, China. In comparison with the traditional grating method, our algorithm has improved advantages in the ease of use, the applicability to different trajectories and compression performance, all of which can better support relevant applications, such as ship trajectory data storage and rapid cartographic display.

Spatial boundaries play an important role in defining spaces, structuring memory and supporting planning during navigation. Recent models of hierarchical route planning use boundaries to plan efficiently first across regions and then within regions. However, it remains unclear which structures (e.g. parks, rivers, major streets, etc.) will form salient boundaries in real-world cities. This study tested licensed London taxi drivers, who are unique in their ability to navigate London flexibly without physical navigation aids. They were asked to indicate streets they considered as boundaries for London districts or dividing areas. It was found that agreement on boundary streets varied considerably, from some boundaries providing almost no consensus to some boundaries consistently noted as boundaries. Examining the properties of the streets revealed that a key factor in the consistent boundaries was the near rectilinear nature of the designated region (e.g. Mayfair and Soho) and the distinctiveness of parks (e.g. Regent's Park). Surprisingly, the River Thames was not consistently considered as a boundary. These findings provide insight into types of environmental features that lead to the perception of explicit boundaries in large-scale urban space. Because route planning models assume that boundaries are used to segregate the space for efficient planning, these results help make predictions of the likely planning demands of different routes in such complex large-scale street networks. Such predictions could be used to highlight information used for navigation guidance applications to enable more efficient hierarchical planning and learning of large-scale environments.