System for semi-naturalistic modeling of trajectory of autonomous underwater vehicle

Abstract

FIELD: semi-naturalistic modeling; testing of autonomous underwater vehicle systems.SUBSTANCE: system for semi-naturalistic modeling of the trajectory of the movement of an autonomous underwater vehicle which includes a trajectory generator, a trajectory parameter estimation module and a trajectory selector, connected in series, a reward processing processor, the output of which is connected to the input of a model training processor connected to the corresponding input of the trajectory parameter estimation module, a mission planning module, the outputs of which are connected to the inputs of the trajectory generator and the state encoding module, wherein the outputs of the state encoding module are connected to the corresponding inputs of the model training processor and the trajectory parameter estimation module, an autonomous underwater vehicle, including a control system, the inputs of which are connected to the output of the trajectory selector and the output of a television camera, and the output is connected to an electromechanical system, further comprises a trajectory visualization module of the autonomous underwater vehicle, the output of which is connected to the corresponding input of the state encoding module, a functioning subsystem of the autonomous underwater vehicle, including a series-connected simulator of a propeller of an autonomous underwater vehicle and a module of spatial dynamics of an autonomous underwater vehicle apparatus connected to an environment simulator, wherein the input of the propeller simulator of the autonomous underwater vehicle is connected to the output of the electromechanical system of the autonomous underwater vehicle, the output of the spatial dynamics module of the autonomous underwater vehicle is connected to the corresponding inputs of the reward processing processor, the trajectory visualization module of the autonomous underwater vehicle and the control system of the autonomous underwater vehicle.EFFECT: expansion of the functional capabilities of the system for semi-naturalistic modeling of the trajectory of movement of an autonomous underwater vehicle.1 cl, 3 dwg

Description

[0001] The invention relates to the field of hardware-software and semi-naturalistic modeling of autonomous underwater vehicle systems. The claimed invention can be used to evaluate the accuracy of autonomous navigation and the trajectory capabilities of underwater movement, which determine the scope of application of autonomous underwater vehicles and the categories of tasks they can solve.

[0002] A method for developing a platform for functional modeling of a remotely operated underwater vehicle with six degrees of freedom is known [US 12019960 B2, IPC G06F 30 / 23, G05B 17 / 02, G06F 30 / 15, published 06 / 25 / 2024]. One of the implementations of the proposed platform for functional modeling of a remotely operated underwater vehicle with six degrees of freedom includes:

[0003] - a system for simulating open sea conditions, which allows you to calculate the speed and direction of wind, the profile and direction of wave propagation, the speed and direction of the current;

[0004] - a simulation model of the underwater communication cable, including a dynamic model of the underwater communication cable and a hydrodynamic model of the control system of the underwater communication cable;

[0005] - model of the dynamics and seaworthiness of the carrier vessel in the time domain;

[0006] - nonlinear dynamics model of a controlled underwater vehicle with six degrees of freedom;

[0007] - data display system.

[0008] The disadvantages of this 6DOF ROV functional simulation platform are that it does not provide:

[0009] - construction and optimization of the trajectory of a remotely controlled underwater vehicle;

[0010] - calculation and evaluation of the quality indicators of the solution of a navigation problem by means of navigation equipment of a remotely controlled underwater vehicle and a carrier vessel.

[0011] A method of model-based reinforcement learning for planning the behavior of autonomous vehicles of various types is known [US 2022 / 0138568 A1, IPC G06N 3 / 08, G06N 3 / 04, published 05.05.2025]. The implementation of the proposed method of model-based learning, as applied to the problem of planning the behavior of autonomous ground vehicles, includes:

[0012] - autonomous vehicle route planner;

[0013] - autonomous vehicle distribution planner on the roadway;

[0014] - autonomous vehicle behavior planning manager, including vehicle speed profile generator, oncoming vehicle flow generator, vehicle speed limiter, vehicle trajectory estimation module, vehicle trajectory optimization module;

[0015] - autonomous vehicle behavior model using reinforcement learning technology;

[0016] - autonomous vehicle behavior model training module.

[0017] The disadvantage of the proposed implementation of the model-based reinforcement learning method for planning the behavior of autonomous vehicles is that the scope of application of this method is limited to the tasks of planning the trajectory and behavior of autonomous ground vehicles, while the implementation of this method does not provide for the simulation modeling of the operating conditions of other types of autonomous vehicles (unmanned aerial vehicles, autonomous underwater vehicles).

[0018] The closest in technical essence to the claimed invention are the method for predicting and planning the trajectory of movement and the system that implements it [US 11835958 B2, IPC G01C 22 / 00, G05D 1 / 00, published 05.12.2023].

[0019] The object of trajectory forecasting and planning is an autonomous underwater vehicle, which includes:

[0020] - control system;

[0021] - an electromechanical system including a cruise electric motor of the underwater vehicle and a cruise electric motor control unit;

[0022] - a sensor system that may include the following components: a navigation radar station, a bottom relief laser scanning module, a television camera, a satellite navigation module and a set of built-in underwater vehicle telemetry sensors.

[0023] The control system, electromechanical system and sensor system of the autonomous underwater vehicle are connected in series.

[0024] The implementation of the presented system for forecasting and planning the movement trajectory includes:

[0025] - mission planning module;

[0026] - module for planning the behavior of an autonomous underwater vehicle;

[0027] - An autonomous underwater vehicle trajectory planner, which consists of a trajectory generator, a state encoding module, a reward processor, a model training processor, a trajectory parameter estimation module, and a trajectory selector.

[0028] The trajectory generator, trajectory parameter estimation module, and trajectory selector are connected in series, with the output of the trajectory selector connected to the input of the autonomous underwater vehicle's control system. The output of the state encoding module is connected to the corresponding inputs of the trajectory parameter estimation module and the model training processor, which is connected to the reward processor.

[0029] The disadvantages of this system are that it does not provide:

[0030] - simulation of operating conditions of an autonomous underwater vehicle;

[0031] - visualization of the trajectory of the autonomous underwater vehicle, taking into account possible obstacles.

[0032] The technical result of the claimed invention is the expansion of the functional capabilities of the system for semi-naturalistic modeling of the trajectory of movement of an autonomous underwater vehicle by providing imitation of the hydrodynamic conditions of the functioning of the autonomous underwater vehicle and visualization of the trajectory of movement of the autonomous underwater vehicle taking into account possible obstacles.

[0033] The specified technical result is achieved in that a system for semi-naturalistic modeling of the trajectory of movement of an autonomous underwater vehicle, including a trajectory generator, a trajectory parameter evaluation module and a trajectory selector connected in series, a reward processing processor, the output of which is connected to the input of a model training processor connected to the corresponding input of the trajectory parameter evaluation module, a mission planning module, the outputs of which are connected to the inputs of the trajectory generator and the state encoding module, wherein the outputs of the state encoding module are connected to the corresponding inputs of the model training processor and the trajectory parameter evaluation module, an autonomous underwater vehicle, including a control system, the inputs of which are connected to the output of the trajectory selector and the output of the television camera, and the output is connected to an electromechanical system, distinguished those, Whatfurther comprises a module for visualising the trajectory of an autonomous underwater vehicle, the output of which is connected to the corresponding input of the state encoding module, a subsystem for operating an autonomous underwater vehicle, including a series-connected simulator of a propeller of an autonomous underwater vehicle and a spatial dynamics module of an autonomous underwater vehicle connected to an environment simulator, wherein the input of the simulator of the propeller of an autonomous underwater vehicle is connected to the output of the electromechanical system of an autonomous underwater vehicle, the output of the spatial dynamics module of an autonomous underwater vehicle is connected to the corresponding inputs of a reward processing processor, a module for visualising the trajectory of an autonomous underwater vehicle and a control system of an autonomous underwater vehicle.

[0034] The essence of the proposed technical solution is illustrated by graphic materials, which show:

[0035] Fig. 1 - structural diagram of the claimed system for semi-naturalistic modeling of the trajectory of movement of an autonomous underwater vehicle;

[0036] Fig. 2 - an example of visualization of the construction of the trajectory of movement of an autonomous underwater vehicle;

[0037] Fig. 3 - block diagram of the method for semi-naturalistic modeling of the trajectory of movement of an autonomous underwater vehicle, implemented by the claimed system.

[0038] The proposed system for semi-naturalistic modeling of the trajectory of the autonomous underwater vehicle includes a mission planning module (1), a trajectory planner (2), an autonomous underwater vehicle (3), and a functioning subsystem of the autonomous underwater vehicle (4).

[0039] The object of the semi-naturalistic simulation is an autonomous underwater vehicle (3), including a control system (12) connected to an environment simulator (16), a television camera (14), and an electromechanical system (13) of the autonomous underwater vehicle. Input 1 of the control system (12) is connected to the output of the spatial dynamics module of the autonomous underwater vehicle (17), input 2 is connected to the output of the trajectory selector (11), and input 3 is connected to the output of the television camera (14). The output of the electromechanical system (13) is connected to the input of the propeller simulator of the autonomous underwater vehicle (15). Hereinafter, the abbreviation "AUV" is used to denote the autonomous underwater vehicle.

[0040] The trajectory planner (2) contains the following components: a trajectory generator (5), a state encoding module (6), an AUV trajectory visualization module (7), a reward processor (8), a model training processor (9), a trajectory parameter estimation module (10), and a trajectory selector (11). The AUV operation subsystem (4) contains the following components: an AUV propeller simulator (15), an environment simulator (16), and an AUV spatial dynamics module (17).

[0041] Output 1 of the mission planning module (1) is connected to input 1 of the state encoding module (6), output 1 of which is connected to input 1 of the model training processor (9), and output 2 is connected to input 2 of the trajectory parameter estimation module (10). Output 2 of the mission planning module (1) is connected to the input of the trajectory generator (5). Input 2 of the state encoding module (6) is connected to the output of the trajectory visualization module of the autonomous underwater vehicle (7), the input of which is connected to the output of the spatial dynamics module of the autonomous underwater vehicle (17). Input 2 of the model training processor (9) is connected to the output of the reward processing processor (8), the input of which is connected to the output of the spatial dynamics module of the autonomous underwater vehicle (17). Input 1 of the trajectory parameter estimation module (10) is connected to the output of the model training processor (9), input 3 is connected to the output of the trajectory generator (5), and the output of the trajectory parameter estimation module (10) is connected to the input of the trajectory selector (11).Input 1 of the spatial dynamics module (17) is connected to the output of the environment simulator (16), input 2 is connected to the output of the propeller simulator of the autonomous underwater vehicle (15).

[0042] Semi-naturalistic modeling of the trajectory of the autonomous underwater vehicle and display of the current position of the autonomous underwater vehicle (3) involves the placement and visualization of the following elements to ensure controlled movement of the AUV along a given route (Fig. 2).

[0043] - nodal points for control of the AUV movement trajectory (18.1-18.7);

[0044] - reference points for constructing the trajectory of the AUV movement (19.1-19.6);

[0045] - static (20.1, 20.2) and dynamic (20.3) obstacles.

[0046] Here, static (20.1, 20.2) obstacles are defined as natural or artificial obstacles with fixed spatial coordinates, including elements of the bottom relief of the simulated water area, structural elements of hydraulic structures, pipeline transport, and various types of artificial barriers. A set of pyramid-shaped reference points placed on the bottom surface of the simulated water area can be considered as reference points for constructing the AUV's trajectory (19.1-19.6).

[0047] The planned AUV trajectory (21.1) is indicated by dashed line segments between the AUV trajectory control nodal points (18.1-18.7). The actual AUV trajectory (21.2) is indicated by solid line segments between the AUV trajectory control nodal points (18.1-18.7).

[0048] The AUV's propulsion control elements are located in a two-dimensional, fixed x-0-y coordinate system, referenced to Earth. The following assumptions are made:

[0049] - the assessment of the bottom relief of the simulated water area is not taken into account;

[0050] - the AUV immersion depth is assumed to be constant for all sections of the AUV trajectory and is not taken into account further.

[0051] The placement of elements to ensure controlled movement of the AUV along a given trajectory is determined by:

[0052] - nodal points of control of the AUV movement trajectory (18.1-18.7) - a pair of coordinates x,y coordinate systems x-0-y;

[0053] - reference points for constructing the trajectory of the AUV movement (19.1-19.6) - a pair of coordinates x,y of the coordinate system x-0-y;

[0054] - static obstacles (20.1, 20.2) - pairs of coordinates x,y of the coordinate system x-0-y of the nodal points of geometric primitives, limiting the zones of placement of static obstacles.

[0055] Dynamic Obstacle Placement (20.3) is set:

[0056] - pairs of coordinates x,y of the x-0-y coordinate system of nodal points of geometric primitives limiting the area of ​​placement of a dynamic obstacle;

[0057] - coordinates x,y of the x-0-y coordinate system of the initial A1 and final A2 points of the displacement vector of the geometric center of the dynamic obstacle placement zone in a given coordinate system. The initial position of the dynamic obstacle (20.3) is indicated in Fig. 2 by a solid contour, the final position of the dynamic obstacle is indicated by a dotted contour.

[0058] The implementation of the components of an autonomous underwater vehicle (3) is possible using the following solutions known from the level of science and technology:

[0059] - control system (12) of an autonomous underwater vehicle - a microcontroller based on the ARM Cortex microprocessor core, includes a set of control and stabilization circuits for linear, angular velocities and spatial orientation angles of an autonomous underwater vehicle (3);

[0060] - television camera (14) - a gyrostabilized three-axis gimbal with a television camera of the visible spectral range of electromagnetic radiation placed on it in a waterproof design, ensures fixation of the contour of the reference points for constructing the trajectory of the AUV movement (19.1-19.6);

[0061] - electromechanical system (13) of an autonomous underwater vehicle - includes a cruise brushless DC electric motor and a cruise electric motor control unit.

[0062] The shaft of the main electric motor of the electromechanical system (13) is directly connected to:

[0063] - a hysteresis dynamometer, at the output of which the measured values ​​of torque and resistance moment of the main electric motor are generated;

[0064] - a shaft rotation speed sensor based on the Hall effect, the output of which generates a measured value of the shaft rotation speed of the main electric motor.

[0065] The measured values ​​of the torque, the drag torque, and the rotational speed of the propulsion motor shaft are sent through the intermediate data acquisition and analysis device to the output of the electromechanical system (13) and transmitted to the input of the propeller simulator of the autonomous underwater vehicle (15). The components of the electromechanical system (13) are not considered in the presented technical solution, are not shown in Fig. 1, and are introduced solely to clarify the operating principle of the proposed system for semi-naturalistic modeling of the trajectory of the autonomous underwater vehicle.

[0066] The module for assessing the trajectory parameters (10) of the autonomous underwater vehicle movement provides a comparative calculation of the following indicators of the quality of the solution of the navigation problem for:

[0067] - predicted trajectory of the AUV movement - is generated by the model training processor (9), the parameters of the predicted trajectory of the AUV movement are received from the output of the model training processor (9) to the input 1 of the trajectory parameter estimation module (10);

[0068] - actual trajectory of the AUV movement (21.2) - is formed based on the results of modeling the spatial dynamics and operating environment of the autonomous underwater vehicle, the parameters of the actual trajectory of the AUV movement are received from output 2 of the state coding module (6) to input 2 of the trajectory parameter evaluation module (10);

[0069] - planned trajectory of the AUV movement (21.1) - is formed by the trajectory generator (5), the parameters of the planned trajectory of the AUV movement are fed to the input 3 of the trajectory parameter estimation module (10).

[0070] At the output of the trajectory parameter evaluation module (10), the values ​​of the following indicators of the quality of the solution of the navigation problem for the planned, predicted and actual trajectories of the autonomous underwater vehicle are generated:

[0071] - root mean square positioning error of the AUV, m;

[0072] - average absolute positioning error of the AUV, m;

[0073] - average minimum distance to obstacles, with active obstacle avoidance by the AUV, m;

[0074] - relative error of AUV positioning by coordinates, %;

[0075] - average reaction time of the AUV to an obstacle, from the moment the object is detected until the start of the evasive maneuver, sec.;

[0076] - the number of trajectory corrections per 100 meters of AUV movement.

[0077] Based on the results of the comparative calculation of the quality indicators of the solution of the navigation problem, the trajectory selector (11) determines the optimal set of values ​​of the trajectory parameters of the AUV movement (spatial coordinates, spatial orientation angles, linear and angular velocities of the AUV) and uploads them to the control system (12) of the autonomous underwater vehicle (3).

[0078] The trajectory generator (5) provides for the calculation of the parameters of the planned trajectory of the AUV movement (21.1) based on the list of nodal points for monitoring the trajectory of the AUV movement (18.1-18.7), for each of which the following are defined:

[0079] - spatial coordinates x,y of a given nodal point;

[0080] - angles of spatial orientation of the AUV at a given nodal point: roll, trim, course;

[0081] - linear and angular velocities of the AUV at a given nodal point;

[0082] - estimated model time to reach the specified nodal point of control of the AUV movement trajectory.

[0083] The reward processor (8) and the model training processor (9) jointly implement the autonomous navigation model of the AUV, which includes the following elements:

[0084] - convolutional layers for selecting reference points for constructing the trajectory of the AUV movement (19.1-19.6), static (20.1, 20.2) and dynamic (20.3) obstacles;

[0085] - Deep reinforcement learning neural network layers for predicting and optimizing the actual trajectory of the AUV (21.2)

[0086] - fully connected output layer for generating parameters of the predicted trajectory of the AUV movement.

[0087] In accordance with the principles of reinforcement learning known from the state of the art, the model training processor (9) is considered as an intelligent agent that interacts with the external operating environment, which is considered to be the components (15-17) of the operating subsystem of the autonomous underwater vehicle (4). Feedback from the external operating environment to the intelligent agent is implemented via the reward processing processor (8). At the output of the reward processing processor (8), positive or negative assessments of the results of the formation of the parameters of the predicted trajectory of the AUV are formed, which are fed to the inputs of the layers of the deep learning neural network with reinforcement of the AUV autonomous navigation model.

[0088] Based on the results of the operation of the autonomous navigation model of the AUV, the following parameters of the predicted trajectory of the autonomous underwater vehicle are generated for a given estimated model time:

[0089] - spatial coordinates x,y,z of the autonomous underwater vehicle;

[0090] - values ​​of spatial orientation angles (roll, pitch and course) of an autonomous underwater vehicle;

[0091] - values ​​of linear and angular velocities of an autonomous underwater vehicle.

[0092] Autonomous underwater vehicle trajectory visualization module (7) provides:

[0093] - visualization of the trajectory of the autonomous underwater vehicle (3);

[0094] - display of current values ​​of spatial coordinates, spatial orientation angles, linear and angular velocities of an autonomous underwater vehicle in an easy-to-read format;

[0095] - visualization of elements for ensuring controlled movement of the UAV along a given route in accordance with Fig. 2.

[0096] Status coding module (6) provides:

[0097] - formation of a data array for training the autonomous navigation model of the AUV, implemented by the model training processor (9);

[0098] - determination of the values ​​of the parameters of the actual trajectory of the AUV movement (spatial coordinates, spatial orientation angles, linear and angular velocities of the AUV) and their uploading to the trajectory parameter estimation module (10).

[0099] The data array for training the autonomous navigation model of the AUV, generated by the state coding module (6), includes:

[0100] - spatial coordinates and contours of reference points for constructing the trajectory of the AUV movement (19.1-19.6) - received from the mission planning module (1);

[0101] - current spatial coordinates of the AUV, values ​​of linear, angular velocities and accelerations of the AUV - received from the AUV trajectory visualization module (7).

[0102] The mission planning module (1) is designed to input initial data and scenario parameters for the semi-naturalistic simulation of the autonomous underwater vehicle's trajectory. Next, for the selected scenario, the trajectory generator (5) calculates the elements of the desired trajectory of the autonomous underwater vehicle (21.2). Three example scenarios for the semi-naturalistic simulation of the trajectory of the autonomous underwater vehicle are provided below.

[0103] 1. Scenario 1. Providing AUV navigation in conditions of limited visibility:

[0104] - The AUV must navigate a 50-meter long corridor with turns that simulates an underwater tunnel with limited visibility.

[0105] - Obstacles to be considered: random objects (rocks or individual underwater structures) located in the path of the AUV.

[0106] 2. Scenario 2. Crossing the AUV water area while avoiding dynamic obstacles:

[0107] - The UAV must pass through a test area with underwater dynamic obstacles that can change their location.

[0108] - Obstacles to be considered: moving objects (other underwater vehicles).

[0109] 3. Scenario 3. Active maneuvering of the AUV with bypass of dynamic obstacles:

[0110] - The AUV must follow a given trajectory that includes several sharp turns and movement through an area with dynamic random obstacles.

[0111] - Obstacles to be considered: moving objects (other underwater vehicles).

[0112] For the selected scenario of semi-naturalistic modeling of the trajectory of the autonomous underwater vehicle, the mission planning module (1) determines:

[0113] - reference values ​​of route completion criteria: root mean square positioning error of the AUV, average absolute positioning error of the AUV, relative positioning error of the AUV by coordinates;

[0114] - scenarios for the appearance of dynamic obstacles - values ​​of the x,y coordinates of the initial A1 and final A2 points of the displacement vector of the geometric center of the placement zone for each dynamic obstacle at specified moments of the model time.

[0115] The mission planning module (1) also includes a database of reference points for constructing the AUV's trajectory (19.1-19.6). For each reference point, the following are specified:

[0116] - examples of graphical visualization of the contour of the reference point taking into account the weather conditions of the AUV application, the visibility of the reference point and the direction of the AUV approach to the reference point;

[0117] - spatial coordinates x,y,zof the reference point;

[0118] - current spatial coordinates x,y,z of the APA, in which the reference point was fixed;

[0119] - values ​​of linear, angular velocities and spatial orientation angles of the AUV at the moment of fixing the reference point.

[0120] The propeller simulator of the autonomous underwater vehicle (15) implements a model of the hydrodynamic forces and moments generated by the rudder-propeller group of the autonomous underwater vehicle with a propeller of a fixed geometric pitch. The environmental simulator (16) implements a model of hydrodynamic forces and moments of a viscous nature during the movement of the AUV in the horizontal and vertical planes. In addition, the environmental simulator (16) ensures dynamic adaptation of environmental conditions based on the current parameters of the AUV's motion in space. In this case, a dynamic adaptation circuit for conducting autonomous tests of the AUV in real time is formed, which includes: the environmental simulator (16), the spatial dynamics module of the AUV (17), and the control system (12). The spatial dynamics module of the AUV (17) implements the following systems of dynamic equations:

[0121] - linear motion of the center of mass of an autonomous underwater vehicle in a fixed coordinate system associated with the Earth;

[0122] - angular motion of the autonomous underwater vehicle in the coordinate system associated with the center of mass of the AUV.

[0123] At the output of the autonomous underwater vehicle spatial dynamics module (17), the values ​​of the following parameters of the actual trajectory of the autonomous underwater vehicle for a given calculated model time are generated, taking into account external environmental factors and the interaction of the AUV propeller with the aquatic environment:

[0124] - spatial coordinates x,y,z of the autonomous underwater vehicle;

[0125] - values ​​of spatial orientation angles (roll, pitch and course) of an autonomous underwater vehicle;

[0126] - values ​​of linear and angular velocities of the autonomous underwater vehicle.

[0127] The following are the procedures of the method of semi-naturalistic modeling of the trajectory of the autonomous underwater vehicle, implemented by the claimed system.

[0128] 1. Formation of a scenario for a semi-naturalistic simulation of the AUV motion trajectory (Fig. 3, pos. 301) by the mission planning module (1).

[0129] 2. Calculation of the planned trajectory of the AUV movement (Fig. 3, pos. 302) by the trajectory generator (5).

[0130] 3. Formation of a data array for training the autonomous navigation model of the AUV (Fig. 3, pos. 303) by the state coding module (6).

[0131] 4. If the estimates of the predicted trajectory (Fig. 3, pos. 304) are not available, then the model training processor (9) performs the initial training of the autonomous navigation model of the AUV (Fig. 3, pos. 305). In this case, the convolutional layers for extracting the reference points for constructing the AUV movement trajectory and the fully connected output layer for generating the parameters of the predicted AUV movement trajectory are subject to training.

[0132] 5. The trajectory parameter evaluation module (10) evaluates the parameters of the route trajectory (Fig. 3, pos. 306), based on the results of which it is determined whether the trajectory passing criteria (Fig. 3, pos. 307) have been achieved in accordance with the scenario of the semi-naturalistic modeling of the AUV movement trajectory (Fig. 3, pos. 301) or not.

[0133] 6. If the criteria for passing the trajectory (Fig. 3, pos. 307) are not achieved, then the trajectory selector (11) determines the parameters of the AUV movement trajectory (Fig. 3, pos. 308) with subsequent uploading to the AUV control system (Fig. 3, pos. 309).

[0134] 7. The AUV spatial dynamics module (17) initiates the launch of a semi-naturalistic simulation of the AUV motion trajectory (Fig. 3, pos. 310).

[0135] 8. The environment simulator (16) implements scenarios of the appearance of dynamic obstacles (Fig. 3, pos. 311) in accordance with the scenario of semi-naturalistic modeling of the trajectory of the AUV movement (Fig. 3, pos. 301).

[0136] 9. Collection of telemetry data from the AUV (Fig. 3, pos. 312) by the AUV spatial dynamics module (17) with subsequent transmission to the AUV control system (12).

[0137] 10. Visualization of the trajectory of the AUV movement (Fig. 3, pos. 313) by the AUV trajectory visualization module (7).

[0138] 11. Updating the environmental parameters (Fig. 3, pos. 314) of conducting a semi-naturalistic simulation of the trajectory of the autonomous underwater vehicle by the environmental simulator (16).

[0139] 12. If the end point of the trajectory is reached (Fig. 3, pos. 315), then the reward processing processor (8) downloads the assessments of the passage of the trajectory of the AUV movement (Fig. 3, pos. 316).

[0140] 13. If the end point of the trajectory has not yet been reached (Fig. 3, pos. 315), then the telemetry data from the AUV is collected again (Fig. 3, pos. 312), followed by visualization of the AUV trajectory (Fig. 3, pos. 313) and updating of the environmental parameters (Fig. 3, pos. 314).

[0141] 14. If the predicted trajectory estimates (Fig. 3, pos. 304) from the reward processing processor (8) are available, the model training processor (9) downloads the AUV motion trajectory estimates (Fig. 3, pos. 316) and performs secondary training of the AUV autonomous navigation model (Fig. 3, pos. 317). In this case, the layers of the deep learning neural network with reinforcement are subject to training for predicting and optimizing the actual AUV motion trajectory. Then the procedures (Fig. 3, pos. 306-315) are repeated.

[0142] 15. If the criteria for completing the trajectory specified in the semi-naturalistic modeling scenario of the AUV movement trajectory (Fig. 3, pos. 301) are achieved (Fig. 3, pos. 307), then retraining the AUV autonomous navigation model taking into account the results of the semi-naturalistic modeling of the AUV movement trajectory is not required. The semi-naturalistic modeling process is then completed (Fig. 3, pos. 318).

[0143] Thus, the declared system of semi-naturalistic modeling of the trajectory of the autonomous underwater vehicle provides imitation of the operating conditions of the autonomous underwater vehicle, visualization of the trajectory of the autonomous underwater vehicle and assessment of the criteria for passing the trajectory taking into account the operating conditions.

Claims

A system for semi-naturalistic modeling of the trajectory of movement of an autonomous underwater vehicle, including a trajectory generator, a trajectory parameter estimation module and a trajectory selector connected in series, a reward processing processor, the output of which is connected to the input of a model training processor connected to the corresponding input of the trajectory parameter estimation module, a mission planning module, the outputs of which are connected to the inputs of the trajectory generator and the state encoding module, wherein the outputs of the state encoding module are connected to the corresponding inputs of the model training processor and the trajectory parameter estimation module, an autonomous underwater vehicle, including a control system, the inputs of which are connected to the output of the trajectory selector and the output of a television camera, and the output is connected to an electromechanical system, characterized in that it additionally contains a module for visualizing the trajectory of the autonomous underwater vehicle,the output of which is connected to the corresponding input of the state encoding module, a subsystem for the functioning of an autonomous underwater vehicle, including a series-connected simulator of the propeller of the autonomous underwater vehicle and a spatial dynamics module of the autonomous underwater vehicle connected to the simulator of the environment, wherein the input of the simulator of the propeller of the autonomous underwater vehicle is connected to the output of the electromechanical system of the autonomous underwater vehicle, the output of the spatial dynamics module of the autonomous underwater vehicle is connected to the corresponding inputs of the reward processing processor, the trajectory visualization module of the autonomous underwater vehicle and the control system of the autonomous underwater vehicle.

Images