Determining alternative destinations for ships

Abstract

Methods, systems, and computer-readable storage media for determining an alternative target for a vessel. A route is obtained from the vessel, the route comprising a plurality of edges from a current location of the vessel to at least a desired target. Characteristics of at least one alternative target including historical ship density data associated with the at least one alternative target are obtained and used by a trained machine learning model to predict the ship density of the ship at an estimated ship arrival time. A desired alternative target for the vessel is determined based at least on a predicted vessel density, and the route is updated such that it comprises a plurality of edges from the current position of the vessel to the desired alternative target. The route is then output to a control system associated with the vessel.

Description

TECHNICAL FIELD

[0001] The present invention relates to a method, system, and computer-readable storage medium for determining an alternative target for a vessel. BACKGROUND

[0002] In transportation and logistics, determining the best route and target for a vessel is both time consuming and expensive. Automation of these determinations has been met with many difficulties due to the number of factors to be considered in determining alternative routes and corresponding alternative targets.

[0003] Factors such as adverse weather conditions, congestion, and delays in acquiring resources can all result in changes in the determined route and delays in reaching a given destination. Furthermore, changes in each of those conditions throughout the journey can result in the need for further calculations to determine alternative routes and / or destinations.

[0004] Performing this task for a fleet of large vessels is both time consuming and expensive, and while efforts have been made to attempt to automate certain aspects of this process, they are particularly resource intensive and computationally complex. Therefore, it is desirable to reduce the computational complexity and resource requirements for determining alternative routes and / or destinations for a given vessel or fleet of vessels. SUMMARY

[0005] According to aspects of the present disclosure, methods, computer program products such as non-transitory storage media including instructions for performing the method, and systems including control systems, storage devices, remote servers, and processors configured to perform the method are provided.

[0006] The method is a method of determining an alternative target for a vessel, the method comprising the steps of: obtaining at least a route from the vessel, the route comprising a plurality of edges from a current location of the vessel to at least a desired target; obtaining characteristics of at least one alternative target from a remote server, the characteristics comprising at least historical vessel density data associated with the at least one alternative target; predicting, using a trained machine learning model, a vessel density at the at least one alternative target at an estimated vessel arrival time of the vessel at the at least one alternative target based on the characteristics; determining a desired alternative target for the vessel based on at least the predicted vessel density at the at least one alternative target; updating the route such that the route comprises a plurality of edges from the current location of the vessel to the desired alternative target; and outputting the route to a control system associated with the vessel.

[0007] This enables efficient computation of alternative targets for a given route based on factors that can occur during transit or other events that are outside the control of the ship operator that result in increased ship density at the destination (in turn meaning that delays are likely to increase). Machine learning methods are used to enable more accurate computation of ship density based on a number of factors such as weather, resource availability and likelihood of industrial action to be taken into account when determining suitable alternative targets. By outputting an updated route, the autonomous ship’s autopilot or other control system can be kept up to date with any changes to the route, further reducing disruption and delays.

[0008] Optionally, a route graph is obtained from a storage device, the route graph comprising a plurality of nodes representing real world locations, and a plurality of edges between the plurality of nodes, and wherein the route is based on the route graph. Furthermore, the at least one alternative target can be one of the plurality of nodes of the route graph. This enables an accurate route graph to be provided that can be efficiently altered and adjusted based on a large amount of data such as AIS data while also taking into account a number of other factors such as ship characteristics. This also enables autonomous components such as autopilot systems to be updated and facilitates navigation of the ship along the route. As the route is transmitted to the control system by efficiently processing the route and route graph, resources are more efficiently utilised, especially when transmitting the route and / or route graph over a network, as such network requirements are reduced as the route can be generated based on information available to / associated with the control system stored in memory.

[0009] Optionally, the characteristics of the at least one alternative target comprise at least one of: weather data associated with the at least one alternative target at an estimated ship arrival time; environmental characteristics associated with the at least one alternative target at an estimated ship arrival time; historical waiting times associated with the at least one alternative target; and delay inducing factors associated with the at least one alternative target at an estimated ship arrival time. This enables a number of factors to be taken into account when determining alternative targets, such as the weather at the alternative target, where if the visibility information is below a given threshold, the ship can not be able to access. Such examples include a port where the ship cannot dock when the visibility is poor. Other conditions such as ocean currents and / or wind speeds near the ship’s determined route can also affect the transit of the ship along the route, and so by taking these properties into account when calculating alternative targets, the reliability of the desired alternative target is improved, and whether it is worth adjusting / updating the route to include the alternative target.

[0010] The method can further comprise the step of determining whether to select the desired alternative target based on the characteristics of the desired alternative target. By taking into account whether to select the desired alternative target based on the characteristics, it can be determined whether the alternative target is more efficient and / or more desirable than the original target of the route, ensuring that the most efficient and desirable route is provided to the control system. BRIEF DESCRIPTION OF DRAWINGS

[0011] Embodiments of the present application will now be described, by way of example only, with reference to the accompanying drawings in which:

[0012] Figure 1 is a schematic representation of a machine learning architecture of a neural network according to an example;

[0013] Figure 2 is a representation of AIS data according to an example;

[0014] Figure 3 is a flowchart illustrating a method for determining an alternative target for a vessel according to an example;

[0015] Figure 4 is a representation of AIS data applied to a plurality of nodes according to an example;

[0016] Figure 5 is a representation of a route according to an example; and

[0017] Figure 6 is a schematic representation of a system for determining an alternative target for a vessel according to an example. DETAILED DESCRIPTION

[0018] Accurately determining an alternative destination / target and route for a vessel, such as a seagoing vessel, is a complex and time-consuming process, especially when an event occurs during transportation between a starting point and an original destination. For example, weather conditions can have a large impact on the chosen route, especially for vehicles such as ships and airplanes. Being able to react quickly and efficiently to such changes is crucial to ensure smooth passage along the route or effective detour to an alternative destination.

[0019] As automated systems, such as autopilot devices and / or autonomous control for watercraft, develop, accurate route planning that can compensate for real-world changes, such as weather or increased congestion, is increasingly important. In handling such watercraft, it is particularly important to enable analysis of such changes and to compensate for those changes on the route, such as changes to the destination or goal of the route. Allowing those watercraft to intuitively and intelligently update a central control system or enabling the central control system to provide updates to the watercraft facilitates fast and efficient travel along the route to the destination. Similarly, providing the control system with the information can also aid in reprogramming of such automated systems, thereby improving the efficiency of the transportation between the starting location and the destination. In some examples, the control system can include a display for indicating a user, such as an operator of the watercraft, and / or the control system can include additional processors or processing capabilities to automatically adjust and / or alter the route of the watercraft, such as to a new destination or goal. Further, in the logistics industry, such updates enable customers and clients to be timely informed of the estimated time of arrival of goods, as well as to be informed of any changes to the route. Thus, when the route of the watercraft is adjusted by the control system, the control system can notify one or more clients of the updated destination.

[0020] There are several aspects of a system for determining routes for watercraft and for determining updates to those routes. In many examples, machine learning methods and apparatus are used to quickly and efficiently analyze large amounts of information and provide accurate outputs for the control system of the watercraft for further processing.

[0021] Machine learning model

[0022] Figure 1 An example of a machine learning model 100 is shown that can be used to determine a new destination for a watercraft based on a route and other factors.

[0023] The machine learning method 100 is a neural network configured to receive a plurality of inputs 110a, 110b, analyze the plurality of inputs, and produce an output 150. The analysis of the inputs is in a number of fully connected layers, however, it should be understood that in some examples, other types of neural networks can be used, such as deep sequential models, and the methods discussed below are not limited to using a particular type of neural network.

[0024] Machine learning method 100 includes multiple layers 130, each layer configured to receive at least one input directly or from a previous layer and process the at least one input to produce an output. The output can then be provided to subsequent layers, such as between layer 1 and layer 2, or the output can be used as the output 150 of machine learning model 100. Each of the layers 130 includes at least one neuron 130a, 130b configured to process at least one input and generate an output. Neurons 130a, 130b are typically grouped into layers based on the different operations they perform on their inputs. The output is then transmitted along connection 130ab to other neurons 130a, 130b. In some examples, each neuron 130a, 130b transmits a signal along connection 130ab to another neuron 130a, 130b only if a threshold is met.

[0025] Such as Figure 1 The neural network of the machine learning model 100 is trained based on training data arranged to refine the processing performed at each neuron 130a, 130b. During training, the weights associated with each neuron 130a, 130b and connection 130ab are adjusted based on the provided inputs 110a, 110b, such that the strength of the connection increases or decreases, thereby affecting whether a signal is transmitted along connection 130ab to other neurons 130a, 130b. Each layer can be traversed multiple times according to the weights to ensure the most accurate overall output 150.

[0026] In some examples, the machine learning model 100 may include shortcuts 140, sometimes referred to as residual neural networks, which help skip layers, thereby simplifying the learning process, accelerating learning, and reducing the likelihood of encountering vanishing gradients. This can be achieved by grouping inputs 110a, 110b that are strongly linearly correlated with the output. For example, there is a strong correlation between the length of the route and the time required to traverse it.

[0027] Figure 1 The machine learning model 100 is configured to receive multiple different inputs 110a, 110b. Inputs may include data such as ship characteristics (i.e., length, manufacturing data, type), current time information (i.e., day, month, year), journey characteristics (i.e., starting position, destination, previous destination, remaining distance, predicted route, average time taken by other ships taking this route), and other computational information such as planned arrival time. Furthermore, in some examples, data may be obtained from one or more other sources. One such example is that input may be obtained from an Automatic Identification System ('AIS'), which uses transceivers on the ship to obtain current position and ship information, similar to the transponder system used by aircraft. AIS data includes information such as the ship's current position, speed, weight, heading, draft, and other characteristics.

[0028] The inputs 110a, 110b can be classified as qualitative inputs 110a or quantitative inputs 110b. Examples of qualitative inputs 110a include destination, vessel type and size, while quantitative inputs 110b include latitude / longitude and vessel speed.

[0029] To improve the number of training instances, the qualitative inputs 110a can be pre-processed by feature learning 120. This enables certain aspects of training of these features prior to the main training of the machine learning model 120. By performing this pre-processing, the qualitative inputs 110a can be processed to quantify and classify similarities between such qualitative inputs 110a. This thereby speeds up the training of the main machine learning model 100.

[0030] The output 150 can include more than one piece of information for use in the concepts described below. For example, when determining an estimated time of arrival, the output 150 can include both an estimated time of arrival prediction and a confidence estimate. The confidence estimate portion of the output 150 can utilize Bayesian dumping over several predictions to determine a probability distribution and thereby indicate whether a given prediction is likely to be more accurate than another prediction. Similarly, previous predictions can be used to indicate an average prediction error to indicate whether a particular output based on given inputs is more or less likely to be correct, or within a given allowed range. Thus, if new criteria / information is provided to the machine learning model 100, the output can indicate whether this is likely to increase or decrease the confidence in the output 150. This additional level of confidence can be provided to the control system and analyzed along with the main output to determine whether a given action should be taken based on this information. For example, where the confidence output indicates a low confidence, given the uncertainty in the output 150, further predictions and / or analysis can be performed before the control system indicates a particular action. In one example, further analysis can be performed, such as at a later time during the transportation of the vessel, before an action is taken, which can result in a more accurate output and thus a higher confidence. Conversely, where the confidence output indicates a high confidence, the control system can perform a particular action as the model indicates that the output 150 is likely to be very accurate.

[0031] The machine learning model 100 can be implemented as part of the invention described below with respect to Figures 2 to 6 but it will be appreciated that the machine learning model can be implemented with respect to any number of other inventions whereby it is necessary to obtain an estimated time of arrival or any other estimated time based on a plurality of inputs. Similarly, while the example shown has four layers, it will be appreciated that the machine learning model can have any number of layers. Figure 1

[0032] Determining an alternative vessel destination / goal

[0033] ​Determining a route for a vessel between two points, a starting location and a destination, is a complex task and relies on multiple inputs. For example, the type of vessel, the size of the vessel, and congestion in and around potential routes can all influence the optimal route. While the examples described below relate to the determination of a route for a vessel such as a ship, it will be appreciated that the described methods and systems can be used to determine an alternative destination for any type of vehicle such as a car, a delivery truck, or an airplane.

[0034] Determining a route for a vessel is not always a simple task due to the nature of the available routes, for example, sea routes for ships. The great circle distance, which accounts for the curvature of the Earth over long distances, is not always applicable, and simply dividing the Earth into a grid does not scale due to different high-density and low-density areas. That is, a large number of grid points are generated for areas on the Earth that the vessel will never travel, such as the North Pole and most of the Pacific Ocean, or in the case of a ship, for land.

[0035] Figure 2 A representation 200 of AIS data obtained in an example in the Atlantic Ocean is shown. The AIS data includes a plurality of points 210a, 210b, 210c, 210d each representing a location of a ship at a given time. For example, a first ship, Ship A, has at least two data points within the AIS data. A first data point 220a indicates a location of Ship A on January 1, 2021 at 13:12, and at least a size characteristic. A second data point 220a indicates a location of Ship A on January 9, 2021 at 14:27. Similarly, a second ship, Ship B, also has at least two data points within the AIS data, a first data point 230a indicates a location of Ship B on February 21, 2021 at 02:54, and at least a size characteristic. A second data point indicates a location of Ship B on February 28, 2021 at 05:36. It will be appreciated that there can be other data points representing locations of Ship A and Ship B in the dataset between the selected two data points. Each data point can also include other attributes such as a heading, a width, a length, a draft, a gross tonnage, and a current destination.

[0036] In some examples, the size characteristic associated with each data point can be used to group the data points for filtering. For example, in the case that the vessel for which a route is being determined exceeds a given size and is classified as a New Panamax ship, then data points including vessels that are able to pass through the Panama Canal can be excluded as they cannot be used to determine a route for the vessel (as the vessel is too large to pass through the Panama Canal).

[0037] There are many methods available for filtering and / or reducing datasets to multiple nodes. For example, the K-means clustering algorithm can be applied to historical data, where K is set to the desired number of nodes. In some examples, K can be set to 50,000 to provide a sufficient number of nodes while achieving efficient processing. It should be understood that other methods for clustering nodes can be used, such as distribution- or density-based clustering methods.

[0038] Figure 3 A method 300 for determining alternative targets for ships is shown, which overcomes the problems mentioned above. In step 310, method 300 obtains a route. This route may include information based on the above regarding... Figure 2 The described AIS data 200 (or other data) is connected to multiple edges of multiple nodes. In some examples, AIS 200 (or other data) can be used to generate multiple nodes. Figure 4 Representation 400 shows nodes 410a, 410b, and 410c identified / generated for a ship based on AIS data 200. The density of nodes 410a, 410b, and 410c is based on the cross-regional distribution of data points in AIS data 200. For example, nodes in areas near the coastline, such as those indicated in group 420, are closer together than nodes in the middle of the ocean, such as those indicated in group 430. The route (such as route 510 shown in representation 500) may include multiple nodes based on AIS data (e.g.,...). Figure 4 (as shown), and includes the edges between these nodes.

[0039] Based on the route obtained from the vessel, characteristics of one or more potential alternative destinations are obtained in step 320. Potential alternative destinations may be one or more nodes near the coastline and located within a given vicinity of the destination on the obtained route. In some examples, potential alternative destinations may also be based on the transit time from the vessel's current position to the alternative destination. For example, nodes near the coastline (such as those shown in group 420) may represent suitable alternative destinations and indicate ports capable of receiving vessels whose original destination is the port indicated by node 440.

[0040] For each of the potential alternative targets (such as those nodes forming part of group 420), characteristics can be obtained. For example, characteristics of a given alternative target may include weather data associated with that alternative target. The obtained weather data may be based on weather forecasts for a given time period. Other characteristics that can be obtained include environmental characteristics associated with the alternative target, including but not limited to port characteristics (such as depth, which, when combined with known vessel size, can indicate whether a vessel can berth), available facilities for loading / unloading cargo, and even characteristics such as available fuel types (e.g., diesel or liquefied natural gas).

[0041] In addition to characteristics representing conditions relating to weather and location of potential alternative targets, other delay inducing factors can be considered, for example whether there are industrial actions at the port associated with the alternative target, whether there is a lack of available resources such as cargo containers and / or a lack of space for temporary storage of cargo.

[0042] It will be appreciated that other factors associated with current characteristics of potential alternative targets can also be obtained. Furthermore, characteristics associated with previous times can also be obtained. For example, historical data associated with potential alternative targets can be obtained, which can include weather data, industrial action data or any other type of data relevant to determining whether it is likely that a ship would be impacted by arriving at that alternative target. Such data can include previous waiting times for ships to enter the port after arriving at the associated node, in yet further examples the waiting times can be categorised based on ship characteristics such as size, weight and type of cargo.

[0043] Characteristics of alternative targets can be obtained from storage (for example in the case of historical information) which is local to the ship or external, or from a remote server (for example such as a weather server) as will be described in further detail below with reference to Figure 6

[0044] Once the characteristics of at least one of the alternative targets have been obtained, a ship density for at least one of the alternative targets is predicted at step 330. The ship density for the at least one alternative target can be determined using a trained machine learning model (for example the trained machine learning model 100 described above with reference to Figure 1 The trained machine learning model 100 can be configured to output a ship density for a given estimated time of arrival of a ship at the at least one alternative target. The trained machine learning model 100 can receive a plurality of characteristics obtained in relation to the at least one alternative target, as well as other characteristics and real world factors.

[0045] In some examples, the trained machine learning model 100 can be configured to receive a plurality of alternative targets and their associated characteristics and provide a ship density at each of the alternative targets. Furthermore, it can be desirable to incorporate a number of other features into the determination of the ship density, such as a delay (whether positive or negative) to determine whether the ship density can be more favourable if the ship were to speed up or slow down for a given alternative target.

[0046] Similarly, in some examples, the trained machine learning model 100 can be configured to receive characteristics associated with the original target of the route and determine a ship density at the time of arrival of the ship at the original target based on those characteristics.

[0047] ​Once the vessel density for at least one of the alternative targets for the vessel has been determined, a desired alternative target is determined at step 340. The desired alternative target can be based solely on the vessel density determined with respect to step 330, and in some examples, can also be based on a number of other factors, such as the vessel density calculated with respect to the original target. It will be appreciated that the determination can be based on any number of factors, and not just the more preferred vessel density at the alternative target. For example, the vessel density can be used to filter a list of potential alternative targets, however, other factors such as the distance and / or time to each of the potential alternative targets can also be considered as part of the determination.

[0048] Once the desired alternative target has been determined, at step 350, the route is updated so that it includes a plurality of edges from the current location of the vessel to the desired alternative target. Figure 5 A schematic representation 500 is shown that includes an example route 510 having a plurality of edges between nodes, such as the nodes described above with respect to Figure 4 The route 510 includes a plurality of edges between nodes, and has an original target represented by node 520. After the prediction by the trained machine learning model 100 at step 330 and the determination of the desired alternative target at step 340, the route 520 is updated to include an edge 530 from the current location of the vessel to a node 530a representing the desired alternative target. In some examples, it can not be desirable to select the desired alternative route over the original route. In such examples, the processor 640 can include a target selection module 650 configured to determine whether it is more advantageous to select an alternative target or to maintain the route with the original target.

[0049] After the update to the route 510, at step 360, the route is output to a control system of the vessel. The control system can include a display or other processing components, such as an autonomous driving system capable of providing quick and efficient updates of the route to customers and / or operators.

[0050] In some embodiments, the method 300 can be implemented as part of a system. Figure 6 is a schematic example of a system 600 for determining an alternative target for a vessel. The system includes: a storage device 610 for storing a plurality of nodes forming part of a route map, such as the nodes described above with respect to Figure 4The system comprises: a route (described at the node); a remote server 620; at least one processor 640 for determining alternative targets for the vessel; and a control system 630 associated with the vessel for receiving updated routes. In some examples, the control system 630 associated with the vessel may be able to send routes, such as route 520, including multiple edges, to the processor 640. The storage device 610, remote server 620, at least one processor 640, and control system 630 may all be interconnected as part of a single system or a remotely connected system. Interconnection may be via a system bus 660 or other wired or wireless connections that enable components of the system to communicate over a network such as the Internet. For example, the storage device 610 may be a remote storage device, such as a cloud storage device, or a hard disk drive forming part of a remote server.

[0051] The remote server 620 is configured to provide characteristics of at least one alternative target of the vessel. The remote server 620 may include storage devices for storing historical data and may also include a connection to a network (such as the Internet) to obtain more information (e.g., weather forecasts), information related to industrial operations, or to another service provider, for example, to obtain information related to port capacity, available space for storing containers and / or cargo, and worker availability, as described above with respect to method 300.

[0052] Processor 640 includes prediction module 642 for predicting ship density at at least one alternative target at an estimated arrival time. This can be achieved using a trained machine learning model (e.g., the one mentioned above). Figure 1 The described trained machine learning model 100 determines the ship density of at least one alternative target, such that the trained machine learning model 100 is configured to output the ship density at a given estimated arrival time of the at least one alternative target. The trained machine learning model 100 may receive multiple features associated with the at least one alternative target obtained from a remote server 620, as well as other features and real-world factors. The prediction module 642 or the entire processor 640 may be a machine learning processor, or a specific neural processing unit configured and / or optimized for machine learning processing.

[0053] Following the prediction, a determination module 644 associated with the processor 640 determines a desired alternative target for the vessel based on the predicted vessel density generated by the prediction module 642. The desired alternative target can be based solely on the vessel density determined by the prediction module 642, however in some examples it can also be based on a number of other factors, such as the vessel density calculated with respect to the original target. It will be appreciated that this determination can be based on any number of factors, and not just a more preferred vessel density at the alternative target. For example, the vessel density can be used to filter a list of potential alternative targets, however other factors such as distance and / or time to each of the potential alternative targets can also be considered as part of the determination.

[0054] An update module 646 associated with the processor 640 then updates the route based on the desired alternative target determined by the determination module 644. The route is updated such that it includes a plurality of edges from the current position of the vessel to the desired alternative target. Figure 5 A schematic representation 500 is shown that includes an example route 510 having a plurality of edges between nodes, such as the nodes described above with respect to Figure 4 The route 510 includes a plurality of edges between nodes, and has an original target represented by node 520. The route 520 is updated by the update module 646 to include edges 530 from the current position of the vessel to a node 530a representing the desired alternative target.

[0055] Following the update of the route 510, the route is output by an output module 648 associated with the processor 640 to a control system 630 associated with the vessel. The control system can include a display or other processing components, such as an autonomous driving system capable of providing quick and efficient updates of the route to customers and / or operators.

[0056] In some examples, it can not be desirable to select the desired alternative route over the original route. In such examples, the processor 640 can include a target selection module 650 configured to determine whether it is more advantageous to select an alternative target or to maintain the route with the original target.

[0057] Conclusion

[0058] Reference is made herein to Figures 1-6At least some aspects of the embodiments described are implemented as computer processes for execution by a processing system or processor. However, in some examples, the present disclosure also extends to computer programs specially adapted for execution on or in a device, particularly a device on or in which the processes according to the present disclosure are performed. The program can be in the form of a non-transitory source code, object code, an intermediate form of code such as a partially compiled form, or any other non-transitory form suitable to implement the processes according to the present disclosure. The device can be any entity or apparatus capable of carrying out the program. For example, the device can comprise a storage medium such as a solid state drive (SSD) or other semiconductor-based RAM; a ROM, for example a CD ROM or semiconductor ROM; a magnetic recording medium such as a floppy disk or hard disk; an optical storage device generally; and the like.

[0059] It will be appreciated that, although certain aspects of the above-disclosed are directed to the use of cloud computing, the implementations described are not limited to a cloud computing environment. Rather, embodiments of the present disclosure are capable of implementation in conjunction with any other type of computing environment.

[0060] In the foregoing description of implementations, numerous specific details are set forth to provide an understanding of the implementations disclosed herein. However, implementations according to this disclosure can be practiced without resorting to the details specifically set forth herein. In other instances, well-known methods, procedures, components, and circuits have not been described in detail as not to unnecessarily obscure aspects of the implementations.

[0061] The above embodiments should be understood as illustrative examples of the disclosure. Other embodiments of the disclosure are contemplated. It should be understood that any feature described with respect to any one embodiment can be used alone, or in combination with other features described, and can also be used in combination with one or more features of any other of the embodiments, or any combination of any other of the embodiments. Furthermore, equivalents and modifications as would be apparent to one of ordinary skill in the art are intended to be within the scope of this disclosure as defined by the appended claims.

Claims

1. A method for determining an alternative target for a vessel, the method comprising the steps of: obtaining at least a route from the vessel, the route comprising a plurality of edges from a current location of the vessel to at least a desired target; obtaining characteristics of at least one alternative target from a remote server, the characteristics comprising at least historical vessel density data associated with the at least one alternative target; using a trained machine learning model, predicting a vessel density at the at least one alternative target at an estimated vessel arrival time of the vessel at the at least one alternative target based on the characteristics; determining a desired alternative target for the vessel based on at least the predicted vessel density at the at least one alternative target; updating the route such that the route comprises the plurality of edges from the current location of the vessel to the desired alternative target; and outputting the route to a control system associated with the vessel.

2. The method for determining an alternative target for a vessel according to claim 1, further comprising obtaining a route graph from a storage device, the route graph comprising a plurality of nodes representing real world locations, and a plurality of edges between the plurality of nodes, and wherein the route is based on the route graph.

3. The method for determining an alternative target for a vessel according to claim 2, wherein the at least one alternative target is one of the plurality of nodes of the route graph.

4. The method for determining an alternative target for a vessel according to any preceding claim, wherein the characteristics of the at least one alternative target comprise at least one of: weather data associated with the at least one alternative target at the estimated vessel arrival time; environmental characteristics associated with the at least one alternative target at the estimated vessel arrival time; historical wait times associated with the at least one alternative target; and delay inducing factors associated with the at least one alternative target at the estimated vessel arrival time.

5. The method for determining an alternative target for a vessel according to any preceding claim, further comprising the step of determining whether to select the desired alternative target based on characteristics of the desired alternative target.

6. A system for determining an alternative target for a vessel, the system comprising: a control system associated with the vessel configured to receive a route comprising the alternative target for the vessel; a storage device for storing at least one route, the route comprising a plurality of edges from a current location of the vessel to at least a desired target; a remote server configured to provide characteristics of at least one alternative target, the characteristics comprising at least historical vessel density data associated with the at least one alternative target; and a processor configured to determine the alternative target for the vessel, the processor comprising: a prediction module for using a trained machine learning model to predict a vessel density at the at least one alternative target at an estimated vessel arrival time of the vessel at the at least one alternative target based on the characteristics; ​ ​ ​ determining module to determine a desired alternative target for the vessel based on a predicted vessel density at the at least one alternative target; updating module to update the route such that the route includes a plurality of edges from the current location of the vessel to the desired alternative target; and outputting module to output the route including the desired alternative target to the control system.

7. The system for determining an alternative target for a vessel of claim 6, wherein the storage device is further configured to store a route graph including a plurality of nodes representing real-world locations and a plurality of edges between the plurality of nodes, and wherein the route is based on the route graph.

8. The system for determining an alternative target for a vessel of claim 6 or claim 7, wherein the processor is a machine learning processor configured to execute the trained machine learning model.

9. The system for determining an alternative target for a vessel of any one of claims 6 to 8, further comprising a target selection module to determine whether to select the desired alternative target based on a characteristic of the desired alternative target.

10. A computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to determine an alternative target for a vessel, the instructions comprising: obtaining at least a route for the vessel, the route including a plurality of edges from a current location of the vessel to at least a desired target; obtaining a characteristic of at least one alternative target from a remote server, the characteristic including at least historical vessel density data associated with the at least one alternative target; predicting, using a trained machine learning model, a vessel density at the at least one alternative target at an estimated vessel arrival time of the vessel at the at least one alternative target based on the characteristic; determining a desired alternative target for the vessel based on at least the predicted vessel density at the at least one alternative target; updating the route such that the route includes the plurality of edges from the current location of the vessel to the desired alternative target; and outputting the route to a control system associated with the vessel.

Images