Method for collision avoidance system of host marine ship related to updating ship classification and navigation status of marine ship

Abstract

To provide a method for a collision avoidance system of a host marine ship, and to update a ship classification and a navigation status of a marine ship.SOLUTION: The present method is started by obtaining sensory data from a sensor, which allows the present system to detect and classify marine targets based on their type and navigation status. Incorporating current route planning and navigation marine chart data allows the present system to assess potential obstructions and adjust a navigation route accordingly.SELECTED DRAWING: Figure 5

Description

[Technical Field]

[0001] The present invention relates generally to a method for a collision avoidance system of a host marine vessel and updating the vessel classification and navigation status of the marine vessel. [Background technology]

[0002] One central aspect in autonomous ship navigation and control is the Collision Detection and Collision Avoidance (CDCA) system, which essentially mimics the risk assessment and decision-making procedures of a human navigator to enable safe and efficient operation of the vessel while taking action to avoid running aground or colliding with static and dynamic obstacles.

[0003] Decisions regarding possible evasive maneuvers to avoid a collision depend on the interpretation of the situation. If the CDCA's interpretation of the situation is incorrect, the human operator will need to take manual control of the vessel to safely execute the situation.

[0004] There is room for such improvement in autonomous ship operation. Summary of the Invention

[0005] In view of the above and other shortcomings of the prior art, it is an object of the present invention to provide a method for at least partially alleviating the shortcomings of the prior art.

[0006] According to a first aspect of the present invention, there is provided a computer-implemented method for a collision avoidance system for a host marine vessel, the method comprising: obtaining sensory data from at least one sensor configured to sense a surrounding of the host marine vessel; detecting at least one marine target in the sensory data; classifying the at least one marine target to be one of a number of classes related to vessel type and navigation status; obtaining a current route plan comprising a set of planned positions for the host marine vessel; obtaining navigation chart data including locations of obstacles in at least an immediate surrounding of the host marine vessel; and classifying the at least one classified marine target to determine a second route plan for navigating past the obstacles and the classified at least one marine target. , evaluating the current route plan and navigation chart data; providing a second route plan output including an indication of the navigation status of the marine target and / or the classification of the marine target; receiving at least one of an updated classification of the marine target or an updated navigation status of the marine target from the human-machine interface; evaluating at least one of the updated classification of the marine target or the updated navigation status of the marine target, along with the second route plan and navigation chart data, to determine a third route plan for navigating past the obstacle and the updated at least one marine target; and providing a third route plan output including an indication of the updated navigation status of the marine target or the updated classification of the marine target.

[0007] The present invention is based, at least in part, on the realization that a human operator can adjust the interpretation of a situation by manually adjusting the classification of obstacles. Classification may include (but is not limited to) an overall classification into "ship" or "special vessel," navigation status (e.g., "limited maneuverability," "normal," etc.), and target interpretation with respect to COLREG (e.g., stand-on, give-way, head-on, crossing, overtaking, passing from left, etc.). Additionally, the human operator may adjust target-specific parameters for the CDCA algorithm (which may include minimum passing distance in different directions, maximum passing speed, etc.).

[0008] The method begins by obtaining sensory data from sensors, which allows the system to detect and classify marine targets based on their type and navigation status. Incorporating current route planning and navigation chart data allows the system to assess potential obstacles and adjust navigation path accordingly.

[0009] By classifying maritime targets and considering their navigation status, the system tailors its response to different maritime scenarios, enhancing compliance with various navigation rules and conditions. The output of the new route plan not only informs the vessel's navigation system, but also serves as a transparent indication of how the vessel should navigate relative to other identified maritime targets.

[0010] The integration of the human-machine interface allows for manual updates to the classification and status of marine targets. This feature is crucial for situations where sensor data may be ambiguous or insufficient, ensuring that human expertise can supplement and correct machine-based interpretations. By reevaluating the updated classification in conjunction with the existing navigation plan, the system can formulate a more optimized route plan. This iterative process ensures that the navigation system is responsive and adaptive to real-time human input and environmental changes, resulting in safer and more reliable vessel operations.

[0011] The third route can be considered a deviation from the second route, i.e., starting from the second route, if the classification and status are updated, the third route is determined therefrom as a deviation from the second route.

[0012] In an embodiment, the method may comprise determining target-specific collision avoidance parameters for at least one classified marine target and obstacle in the navigation chart data, determining a second route plan further based on the target-specific collision avoidance parameters, receiving updated target-specific collision avoidance parameters for at least one of the classified at least one marine target or obstacle from a human-machine interface, and determining a third route plan further based on the updated target-specific collision avoidance parameters. Technical advantages may include the ability to more precisely tailor collision avoidance actions by incorporating target-specific parameters such as minimum passing distance or maximum allowable speed. This may result in safer and more efficient navigation strategies tailored to the specific circumstances of each encountered marine target and obstacle.

[0013] In an embodiment, the method may comprise storing in a memory at least an updated classification of the marine target, or an updated navigation status of the marine target, or updated target-specific collision avoidance parameters for at least one classified marine target and obstacle in the navigation chart data. Technical advantages may include the ability for iterative improvement and customization of the collision avoidance system. By storing updated classifications and navigation statuses, the system may refine its understanding and predictions over time, preferably increasing the accuracy and reliability of navigation decisions made by the autonomous collision avoidance system.

[0014] In embodiments, the method may include using the updated collision avoidance parameters and the updated classification to train a machine learning algorithm configured to generate at least the collision avoidance parameters, the classification, and the navigation status. Technical benefits may include leveraging machine learning algorithms to continuously improve the system's decision-making capabilities. As the system accumulates more data regarding successful navigation strategies and updated parameters, the system may learn to better predict optimal navigation paths, resulting in increased safety and efficiency in operations. The updated parameters may be recorded along with route plans, weather, etc. In this case, after collecting such data for some time, or periodically, the model may be trained to learn user preferences for situations so that it can provide solutions that are closer to what a human operator desires, thereby requiring less input and interaction with the system by a human operator in future situations. That is, the method may include learning, for example, by reinforcement learning, to predict target-specific and general parameters based on typical changes a user makes in different situations.

[0015] In an embodiment, the second route plan and the third route plan outputs are provided on a user interface. Technical benefits may include providing the crew with clear, actionable output regarding route plan adjustments. This visibility may assist a human operator in understanding the system's decisions and making informed choices, especially in complex navigation scenarios. The user interface may be a display.

[0016] In embodiments, the method may comprise prompting a user interface for additional input to classify the at least one marine target when classifying the at least one marine target results in an uninterpretable vessel type, an uninterpretable object, or an uninterpretable navigation status. The object may include a navigation hazard, a buoy, a log, a fishing net, debris, a marine mammal, a rock, etc. Technical benefits may include increasing the system's responsiveness to non-standard situations by actively prompting for human input when automatic classification fails. This ensures that the navigation system remains functional and effective even when encountering ambiguous or unclear data.

[0017] In an embodiment, the steps of the method are repeated continuously as the host marine vessel travels. Technical benefits may include the ability to dynamically adapt to changing conditions and new information as the vessel moves. The continuous operation of the method ensures that the vessel's navigation systems can quickly adjust to the evolving maritime environment, thereby maintaining safety and operational efficiency.

[0018] In an embodiment, the classification of a marine target includes at least its ability to comply with COLREG (Convention on International Regulations for Preventing Collisions at Sea). Technical benefits may include ensuring compliance with the International Collision Regulations at Sea (COLREG) by classifying marine targets based on their ability to adhere to these regulations. This may significantly reduce the risk of misunderstandings and collisions in international waters. Collision avoidance algorithms need to consider the status of each target relative to the host vessel, given the planned action. COLREG also defines actions for situations where the other vessel is a special case, such as a vessel with limited maneuverability, a sailing vessel (under sail), etc.

[0019] In embodiments, multiple maritime targets may be considered. Technical benefits may include the ability to manage multiple targets simultaneously, enhancing the utility of the system in high-traffic or congested maritime domains. This allows the system to handle complex scenarios with multiple interacting vessels, each with its own navigation path and intentions.

[0020] In an embodiment, the host marine vessel may be a semi-autonomous or fully autonomous marine vessel.

[0021] In an embodiment, the method may comprise executing a third route plan. Executing the third route plan based on the refined inputs and updates ensures the vessel follows an up-to-date and optimized route for safety and efficiency.

[0022] In an embodiment, the method may comprise sending a request to a human-machine interface prompting a user to confirm execution of the third route plan, and executing the third route plan in response to confirmation received on the human-machine interface.

[0023] There is also provided a control unit configured to carry out the method.

[0024] There is also provided a marine vessel comprising a control unit.

[0025] According to a second aspect of the present invention, there is provided a computer program product comprising program code for a collision avoidance system for a host marine vessel, the computer program product comprising: code for obtaining sensory data from at least one sensor configured to sense a surrounding of the host marine vessel; code for detecting at least one marine target in the sensory data; code for classifying the at least one marine target to be one of a number of classes related to vessel type and navigation status; code for obtaining a current route plan comprising a set of planned positions for the host marine vessel; code for obtaining navigational chart data including positions of obstacles in at least an immediate surrounding of the host marine vessel; and code for determining a second route plan for navigating past the obstacles and the classified at least one marine target. code for evaluating the classified marine target, the current route plan, and the navigation chart data; code for providing a second route plan output including an indication of the navigation status of the marine target and / or the classification of the marine target; code for receiving at least one of the updated classification of the marine target or the updated navigation status of the marine target from the human-machine interface; code for evaluating at least one of the updated classification of the marine target or the updated navigation status of the marine target, along with the second route plan and the navigation chart data, to determine a third route plan for navigating past the obstacle and the updated at least one marine target; and code for providing a third route plan output including an indication of the updated navigation status of the marine target or the updated classification of the marine target.

[0026] Further advantages and features of the second aspect of the invention are largely similar to those described above in relation to the first aspect of the invention.

[0027] Further features and advantages of the present invention will become apparent upon review of the appended claims and the following description. Those skilled in the art will recognize that different features of the present invention can be combined to create embodiments other than those described below without departing from the scope of the invention.

[0028] These and other aspects of the present invention will now be described in more detail with reference to the accompanying drawings, in which exemplary embodiments of the invention are shown. [Brief explanation of the drawings]

[0029] [Figure 1] FIG. 1 illustrates a schematic diagram of a host marine vessel according to one embodiment of the present invention. [Figure 2-1] FIG. 2-1 is a flowchart of method steps according to an embodiment of the present invention. [Figure 2-2] FIG. 2-2 is a flowchart of method steps according to an embodiment of the present invention. [Figure 3] FIG. 3 is a flowchart of method steps according to an embodiment of the present invention. [Figure 4] FIG. 4 is a flowchart of method steps according to an embodiment of the present invention. [Figure 5] FIG. 5 is a flowchart of method steps according to an embodiment of the present invention. [Figure 6] FIG. 6 is an example situation as interpreted by a control system of a host marine vessel according to one embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0030] In the detailed description of the present invention, various embodiments of the present invention are described herein with reference to specific implementations. In describing the embodiments, specific terminology is used for clarity. However, the present invention is not intended to be limited to the specific terminology so selected. While specific exemplary embodiments are described, it should be understood that this is done for illustrative purposes only. Those skilled in the art will recognize that other components and configurations can be used without departing from the scope of the present invention.

[0031] FIG. 1 schematically depicts a top view of a marine vessel 10. The marine vessel 10 may be, for example, a ship, a boat, a barge, or a floating production, storage, and offloading (FPSO) unit. The vessel 10 is equipped with a plurality of thrusters, shown in FIG. 1 as a first thruster 12a, a second thruster 12b, a third thruster 12c, and a fourth thruster 12d. One, some, or all of the thrusters 12a-12d may also be referred to by the reference numeral "12."

[0032] Each thruster 12 may be rotatable in a horizontal plane (parallel to the plane of the drawing in FIG. 1 ) to different thrust directions, or may be mounted in a fixed orientation. Each thruster 12 may include a propeller and an engine that drives the propeller. By increasing the engine speed of the engine, the thrust of the thruster 12 may be increased, and vice versa. The thrusters 12 are exemplified herein as azimuth thrusters.

[0033] The vessel may also include propulsion units of the type trochoidal, cycloidal, water jet, etc. In the case of a fixed axis and rudder, two different devices are typically involved: a propeller and a rudder.

[0034] The vessel 10 further comprises a control system 14. The control system 14 in this example comprises a control unit 16 and a memory 18. The memory 18 may have a computer program 20 stored thereon for controlling the vessel, or at least the vessel's collision avoidance system.

[0035] The vessel 10 further comprises one or more position sensors 22. The position sensors 22 are arranged to provide position data 24 to the control system 14. The position data 24 indicates the position of the vessel 10 in a horizontal plane. The position sensors 22 may be, for example, Global Navigation Satellite System (GNSS) devices and further provide the position data 24 to the control system 14. The GNSS devices may typically provide a heading for the vessel 10. In other examples, the heading is provided by a heading sensor. Some vessels may include a gyrocompass, a satellite compass, or a magnetic compass.

[0036] The vessel 10 further comprises one or more sensors 26 configured to sense the surroundings of the vessel 10 and provide surrounding sensory data 27 to the control system 14 .

[0037] The marine vessel 10 may further include sensors 29 for measuring environmental conditions, such as wind speed and direction, local ocean current conditions, or wave characteristics. Environmental data 30 indicative of environmental conditions is provided to the control system 14 from environmental sensors 29 onboard the marine vessel or from remote sensors, such as weather stations, via wireless transmission devices 32. The environmental data may further include weather forecasts, such as information on wind strength and direction, wave and current strength and direction, etc.

[0038] The control system 14 may be in communication with a control center 28 using a radio transmitter 32. A human operator may provide input to the control system 14 from a remote control center 28. It is also envisioned that the control center may be located on the bridge of the vessel, without remote control from a location away from the vessel.

[0039] The host marine vessel 10 may be a semi-autonomous or fully autonomous marine vessel. In other embodiments, the vessel is a conventional vessel in which a collision avoidance system is used for advisory information.

[0040] FIG. 2 is a flowchart of method steps according to an embodiment of the present invention.

[0041] The method steps are for a collision avoidance system of a host marine vessel and involve the vessel's navigation status and classification.

[0042] In step S102, sensory data is obtained from at least one sensor 29 configured to detect the surroundings of the host marine vessel. Such a sensor 29 may be, for example, radar data, an AIS (or equivalent transponder system), a camera, an infrared camera (SWIR, LWIR, etc.), a laser-based sensor such as LiDAR, etc.

[0043] In step S104, at least one marine target is detected in the sensory data 30. Preferably, multiple marine targets can be detected and considered.

[0044] In step S106, at least one marine target is classified as being one of several classes related to vessel type and navigation status. The classification refers to the status of each target relative to the host marine vessel. For example, the marine target may be a conventional vessel or a so-called special vessel. A "conventional vessel" is typically a power-driven vessel and is not restricted by anything related to its ability to comply with COLREG (Convention on International Regulations for Preventing Collisions at Sea). A conventional vessel follows rules regarding the main encounter situations of meeting, crossing, and overtaking regarding the actions of retaining vessels and giving way vessels. A "special vessel" does not necessarily follow the conventional rules for various reasons, such as operational tasks (e.g., fishing, pipe laying, towing, etc.), limited maneuverability (e.g., anchored, powered down, etc.), or other limitations (e.g., draft restrictions, sailing vessels).

[0045] Typically, object detection and tracking is performed. Tracking can be considered as a post-processing of the detection data to isolate objects from the data and track their movement over time. This makes it possible to estimate and predict the future motion of the objects and provide this as input for a collision avoidance system. It is envisaged to use so-called track-before-detection.

[0046] When several sources of tracking and detection are available, a fusion of the data is advantageously performed. This can be done as "early fusion" at the detection level, or as "late fusion" at the tracking level. It is preferable to use tracking fusion (late fusion).

[0047] Detection, tracking and integration based on point cloud data (e.g., radar, lidar, etc.) or visual imaging (camera, thermal camera, etc.) are standard methods known to those skilled in the art and will not be described in detail herein.

[0048] In optional step S109, when classifying the at least one marine target results in an uninterpretable vessel type, an uninterpretable object, or an uninterpretable navigation status, the user interface prompts for additional input to classify the at least one marine target. A human operator may add an undetected user-specified target, add a user-specified no-go area, or add a user-specified navigation hazard.

[0049] In step S108, a current route plan is obtained comprising a set of planned positions for the host marine vessel. The planned positions may be, for example, GPS coordinates that describe the current route plan. The current route plan may be obtained from the vessel's memory or from a remote center.

[0050] In step S110, navigational chart data is obtained that includes the locations of obstacles in at least the immediate surroundings of the host marine vessel 10. The navigational chart data, e.g., a "nautical chart," may be retrieved from memory or a remote center.

[0051] In step S112, the at least one classified marine target, the current route plan, and the navigation chart data are evaluated to determine a second route plan for navigating past the obstacle and the at least one classified marine target. In other words, based on the classification and the navigation chart data, the current route is updated with a second route plan that passes the obstacle.

[0052] In step S114, a second route plan output is provided that includes an indication of the navigation status of the marine target and / or the classification of the marine target. The output may be provided on a user interface 33 at the remote operation center 28. The user interface may be a display 33.

[0053] In step S116, at least one of an updated classification of the marine target or an updated navigation status of the marine target is received from the human-machine interface 34. That is, the human operator may review and update the classification and / or navigation status if the classification is inaccurate.

[0054] In step S118, at least one of the updated classification of the marine target or the updated navigation status of the marine target is evaluated along with the second route plan and the navigation chart data to determine a third route plan for navigating past the obstacle and the updated at least one marine target. Thus, when the updated classification of the marine target or the updated navigation status of the marine target is received, the route is re-evaluated taking into account the human operator input, and a third route plan is determined.

[0055] In step S120, a third route plan output is provided, which includes an indication of the updated navigation status of the marine target or the updated classification of the marine target. The third route plan output may be shown on the display 33. Preferably, the third route plan is executed by the control system 14. In some implementations, a request is first sent to the human-machine interface 34 prompting the user to confirm execution of the third route plan. The third route plan is executed only after the third route plan is confirmed via the human-machine interface 34.

[0056] Several common collision avoidance parameters are considered when determining a route. A non-exhaustive list of common collision avoidance parameters is as follows: - Narrow channels or waterways. - Separate passage system. - Open sea. - Stay to the right of the lane. - Allows entry to the left side of the channel in narrow channels. - Maximum safe speed. - Minimum speed. - Minimum rate of turn. - Maximum turn rate. - Maximum acceleration. - Maximum deceleration. - Reliable perception range. - Minimum time to closest point of approach (TCPA). - Minimum Distance to Closest Point of Approach (DCPA). - Minimum Closest Point of Approach (CPA). - Minimum turning radius (or course change) for the action. - Minimum distance / time before the encounter to initiate the action. - Minimum passing distance.

[0057] FIG. 3 is a flowchart of additional method steps according to an embodiment of the present invention.

[0058] In step S202, memory 18 is stored with at least an updated classification of the marine target, or an updated navigation status of the marine target, or updated target-specific collision avoidance parameters for at least one classified marine target and obstacle in the navigation chart data. Storing the updated classification, navigation status, and parameters allows them to be used to train a machine learning algorithm configured to generate at least the collision avoidance parameters, classification, and navigation status in step S204. The machine learning algorithm may be a reinforcement learning-based algorithm.

[0059] FIG. 4 is a flowchart of additional method steps according to an embodiment of the present invention.

[0060] In step S302, target-specific collision avoidance parameters are determined for at least one classified marine target and obstacle in the navigation chart data.

[0061] Target-specific collision avoidance parameters include any parameters that affect the risk of collision. For example, a non-exhaustive list is as follows: - Safe passing speed. - Threshold for initiating action if the give-way vessel does not act. - Type of action to be taken if the giving vessel does not act. - Matched encounter point (currently predicted encounter point). - Matched passing side. - Minimum turning radius (or course change) for the action. - Minimum distance / time before the encounter to initiate the action. - Minimum passing distance. - Minimum time to closest point of approach (TCPA). - Minimum Distance to Closest Point of Approach (DCPA). - Minimum Closest Point of Approach (CPA). - Minimum turning radius (or course change) for the action. - Minimum distance / time before the encounter to initiate the action. - Predicted target route / speed / course (this is something the user may have information about on their wireless device, e.g. the user will know that the target is about to turn right).

[0062] The above-mentioned step S112 now further includes determining a second route plan further based on the target-specific collision avoidance parameters.

[0063] Step S304 includes receiving, from the human-machine interface, updated target-specific collision avoidance parameters for at least one of the classified at least one marine target or obstacle.

[0064] Step S120 mentioned above now further includes determining a third route plan further based on the updated target-specific collision avoidance parameters.

[0065] 5 is a schematic functional flowchart of method steps of an embodiment of the present invention. The method, in this embodiment, is based on a data-driven learning algorithm 500 based on data collected during the operation of the vessel and stored in a database 504. The data-driven modeling is adapted to learn to improve parameters based on how a human operator would adjust parameters, navigation status, etc. in different situations. However, the method can be implemented without data-driven modeling.

[0066] The control system receives parameters 506 from the data-driven learning algorithm 500, data 508 such as sensory sensor data 30, GPS data 24, environmental data 27, navigational chart data, etc.

[0067] The data 508 and parameters 506 may be integrated 510 to determine an interpretation 512 of the situation, which may include classification and navigation status of detected objects, and preferably a suggested route to avoid obstacles.

[0068] This interpretation is provided to a human machine interface 514 where a human operator can check 516 the interpretation for correctness and adjust the interpretation, if necessary, using the human machine interface.

[0069] Adjusting the interpretation of the situation involves adjusting for each detected target one or more of its classification as a "ship" or "special vessel", as an "avoid" or "hold" target, as a "meet", "cross" or "overtake" target, as a "pass from left" target, etc.

[0070] Adjusting the interpretation of the target's navigational status may include for parameters related to limited maneuverability, draft constraints, status under operational work, etc. The term "draft-constrained vessel" means a power-driven vessel that is severely limited in its ability to deviate from the course it is following because of the vessel's draft relative to the available depth and width of navigable waters.

[0071] The collision avoidance algorithm 518 receives the updated interpretation and determines parameters for each obstacle, such as how close the system should allow the obstacle to pass.

[0072] The output parameters of the collision avoidance algorithm 518 are provided to the human machine interface 514 so that a human operator can check 520 the results of the collision avoidance algorithm 518. The human operator can adjust 522 parameters, for example, target-specific collision avoidance parameters or general collision avoidance parameters, and feed the updated parameters back to the collision avoidance algorithm 518.

[0073] The collision avoidance algorithm 518 reevaluates and provides an updated maneuver plan for execution 524 .

[0074] Advantageously, at each iteration, parameters from the collision avoidance algorithm 518 may be stored in database 504, an interpretation of the situation 512 may be stored in database 504, data 508 may be stored in database 504, and an analysis and synthesis of data 510 may be stored in database 504 to train the data-driven learning algorithm 500.

[0075] In another example, assume that radar and AIS are used to detect the vessel's surroundings. A first route exists or is calculated before starting the method. The first route may simply be to maintain the vessel's course and speed.

[0076] A second route is calculated as an avoidance route to avoid collisions by clustering radar detections that are clustered to identify objects that belong together. Radar detections are tracked over time to estimate the movement of moving objects. Radar-based targets are then integrated with AIS-based targets. The host vessel's own route and the tracking of other vessels are simulated forward, also with respect to static objects and navigational charts.

[0077] Risk assessment is carried out based on the route and decisions (interpretations) on how to treat each vessel, static object and chart based on the information from them and also how they are encountered.

[0078] The second route is based on this interpretation, so that the host vessel follows the COLREG according to its interpretation of vessel type and voyage status.

[0079] A third route is calculated as an updated avoidance route based on user input. The user may check the interpretation from the user interface 34. The user may adjust the interpretation. A new route is calculated based on the user-modified parameters. The system may automatically execute the third route, which may be shown on the user interface as an advisory, or the system may require user confirmation before automatically executing the avoidance route.

[0080] Situations and user inputs are recorded as data, and based on data recorded over a longer period of time, machine learning can be trained and used to predict collision avoidance parameters, navigation status, etc. The next time an avoidance route calculation / redirection from the nominal route (first route) is required, the machine learning model 500 can be used to suggest new parameters that the user can choose to use.

[0081] 6 shows an example situation being interpreted by the control system 14 of the host marine vessel 10 on a planned route 601. The planned route may be determined before the vessel leaves the port of departure or may be updated during the voyage. The planned route 601 is determined based on what is known in advance before reaching a particular location, for example, based on weather forecasts, navigational charts, etc. A nominal route plan 601 is assumed to already exist or to be calculated before departure. In the absence of a pre-planned route plan, it is assumed that the current speed and course should be maintained as an initial primary route.

[0082] In Figure 6, the first trajectory 602 results from considering all targets 604-608 as "normal vessels." In that case, the normal COLREG encounter situation is in effect. For example, the host vessel 10 estimates that it is approaching vessel 604 for an encounter and therefore makes a right turn. The second vessel 603 may be predicted to move away before the host vessel 10 reaches it, causing the host vessel to pass vessel 603 astern and consider the passage safe. The action for vessels 605 and 608 may be that vessel 605 is not moving, so the host vessel 10 passes it with a slight right turn, and vessel 608 becomes a give-way vessel because the host vessel 10 is approaching from the right.

[0083] The relationship of different targets 604-608 to the planned trajectory will change completely if target 604 is considered a "special vessel." In that case, i.e., if the user provides input to change the classification of target 604 from a "normal vessel" to a "special vessel," a second trajectory 612 will be generated. That is, the user-provided input may allow the host marine vessel 10 to pass target 604 with a left turn, which will completely change the interpretation of the status of different targets 604-608 relative to the planned trajectory 602. Here, in addition to target 604, only targets 605 and 607 are relevant. Both are considered "crossing" targets, and here the host vessel 10 is potentially an give-way vessel. Target 605 is considered an "overtaking" target 605, which is different from the first case (trajectory 602).

[0084] Note the cascading effect and how a manual change in the interpretation of one target, e.g., target 604, can lead to a completely different end result, resulting in a change in the importance of different targets. This example can be further extended to illustrate the effect of a proposed solution, for example, where the user changes the distance the system considers when passing a fishing vessel (orange) from the stern. If the fishing vessel is a trawler, it typically needs to have a long enough distance to pass from the stern so as not to interfere with the trawl net. This, too, will change the first trajectory. This illustration also does not consider the effect of changing speed during maneuvering and how that affects the predicted encounters with different targets.

[0085] The control unit may include a microprocessor, a microcontroller, a programmable digital signal processor, or another programmable device. The control unit may also, or instead, include an application specific integrated circuit, a programmable gate array, or programmable array logic, a programmable logic device, or a digital signal processor. When the control unit includes a programmable device, such as the microprocessor, microcontroller, or programmable digital signal processor described above, the processor may further include computer-executable code for controlling the operation of the programmable device.

[0086] In one or more examples, the functions described may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software, the functions may be stored on or transmitted over as one or more instructions or code on a computer-readable medium and executed by a hardware-based processing unit. Computer-readable media may include computer-readable storage media, which correspond to tangible media such as data storage media, or communication media, including any medium that facilitates transfer of a computer program from one place to another, for example according to a communications protocol. As such, computer-readable media may generally correspond to (1) tangible computer-readable storage media that is non-transitory, or (2) a communication medium such as a signal or carrier wave. Data storage media may be any available medium that can be accessed by one or more computers or one or more processors to retrieve instructions, code, and / or data structures for implementing the techniques described in this disclosure. A computer program product may include a computer-readable medium.

[0087] By way of example, and not limitation, such computer-readable storage media may comprise RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, flash memory, or any other medium that can be used to store desired program code in the form of instructions or data structures and that can be accessed by a computer.

[0088] While the invention has been described with reference to specific exemplary embodiments thereof, it is evident that many different alternatives, modifications and similar changes will be apparent to those skilled in the art.

[0089] Additionally, variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite articles "a" or "an" do not exclude a plurality. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

Claims

1. 1. A computer-implemented method for a collision avoidance system for a host marine vessel, the method comprising: obtaining sensory data from at least one sensor configured to sense the surroundings of the host marine vessel (S102); Detecting at least one marine target in the sensory data (S104); Classifying (S106) at least one of the marine targets to be one of several classes related to vessel type and navigation status; obtaining a current route plan comprising a set of planned positions for the host marine vessel (S108); obtaining navigational chart data (S110) including locations of obstacles in at least the immediate surroundings of the host marine vessel; evaluating (S112) the at least one classified marine target, the current route plan, and the navigational chart data to determine a second route plan for navigating past the obstacle and the at least one classified marine target; providing (S114) the second route planning output including an indication of the navigation status of the marine target and / or the classification of the marine target; receiving at least one of an updated classification of the marine target or an updated navigation status of the marine target from a human-machine interface (S116); evaluating the at least one of the updated classification of the marine target or the updated navigation status of the marine target together with the second route plan and the navigation chart data to determine a third route plan for navigating past the obstacle and the updated at least one marine target (S118); providing (S120) the third route planning output including an indication of the updated navigation status of the marine target or the updated classification of the marine target; A method comprising:

2. determining target-specific collision avoidance parameters for the classified at least one marine target and the obstacle in the navigation chart data; determining the second route plan further based on the target-specific collision avoidance parameters; receiving, from a human-machine interface, updated target-specific collision avoidance parameters for at least one of the classified at least one marine target or the obstacle; determining the third route plan further based on the updated target-specific collision avoidance parameters; The method of claim 1 , comprising:

3. 3. The method of claim 1 or 2, comprising storing in a memory at least the updated classification of the marine target, or the updated navigation status of the marine target, or updated target-specific collision avoidance parameters for the classified at least one marine target and the obstacle in the navigation chart data.

4. 4. The method of claim 3, comprising using the updated collision avoidance parameters and the updated classification to train a machine learning algorithm configured to generate at least the collision avoidance parameters, the classification, and the navigation status.

5. The method of any one of claims 1 to 4, wherein the output of the second route plan and the third route plan is provided on a user interface.

6. The method of claim 5 , wherein the user interface is a display.

7. 7. The method of claim 1, wherein when the classification of at least one of the marine targets results in an uninterpretable vessel type, an uninterpretable object, or an uninterpretable navigation status, the user interface prompts for additional input to classify at least one of the marine targets.

8. A method according to any preceding claim, comprising repeating the steps of the method continuously as the host marine vessel travels.

9. The method according to any one of claims 1 to 8, wherein the classification of the marine target comprises at least its ability to comply with COLREG (Convention on the International Regulations for Preventing Collisions at Sea).

10. The method according to any one of claims 1 to 9, wherein a plurality of marine targets is considered.

11. The method of any one of claims 1 to 10, wherein the host marine vessel is a semi-autonomous or fully autonomous marine vessel.

12. sending a request to a human machine interface prompting a user to confirm execution of the third route plan; executing the third route plan in response to the confirmation received on the human machine interface; and The method of any one of claims 1 to 11, comprising:

13. A control unit configured to carry out the method according to any one of claims 1 to 12.

14. A marine vessel (10) comprising the control unit according to claim 13.

15. 1. A computer program product comprising program code for a collision avoidance system for a host marine vessel, the computer program product comprising: code for obtaining sensory data from at least one sensor configured to sense the surroundings of the host marine vessel; code for detecting at least one marine target in the sensory data; a code for classifying at least one of the marine targets as being in one of several classes related to vessel type and navigation status; code for obtaining a current route plan comprising a set of planned positions for the host marine vessel; code for obtaining navigational chart data including the location of obstacles in at least the immediate vicinity of the host marine vessel; code for evaluating the at least one classified marine target, the current route plan, and the navigational chart data to determine a second route plan for navigating past the obstacle and the at least one classified marine target; code for providing the second route planning output including an indication of the navigation status of the marine target and / or the classification of the marine target; code for receiving, from a human machine interface, at least one of an updated classification of the marine target or an updated navigation status of the marine target; code for evaluating the at least one of the updated classification of the marine target or the updated navigation status of the marine target, along with the second route plan and the navigational chart data, to determine a third route plan for navigating past the obstacle and the updated at least one marine target; code for providing the third route planning output including an indication of the updated navigation status of the marine target or the updated classification of the marine target; A computer program product comprising:

Images