CN114780393A - Marine unmanned cluster intelligent algorithm test training system - Google Patents
Marine unmanned cluster intelligent algorithm test training system Download PDFInfo
- Publication number
- CN114780393A CN114780393A CN202210369011.1A CN202210369011A CN114780393A CN 114780393 A CN114780393 A CN 114780393A CN 202210369011 A CN202210369011 A CN 202210369011A CN 114780393 A CN114780393 A CN 114780393A
- Authority
- CN
- China
- Prior art keywords
- algorithm
- test
- scene
- simulation
- management module
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 161
- 238000012360 testing method Methods 0.000 title claims abstract description 91
- 238000012549 training Methods 0.000 title claims abstract description 60
- 238000007726 management method Methods 0.000 claims abstract description 78
- 230000006870 function Effects 0.000 claims abstract description 21
- 230000008447 perception Effects 0.000 claims abstract description 18
- 238000004088 simulation Methods 0.000 claims description 48
- 230000033001 locomotion Effects 0.000 claims description 16
- 230000007246 mechanism Effects 0.000 claims description 16
- 238000000034 method Methods 0.000 claims description 16
- 230000004888 barrier function Effects 0.000 claims description 9
- 238000011156 evaluation Methods 0.000 claims description 8
- 230000015572 biosynthetic process Effects 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 7
- 230000001133 acceleration Effects 0.000 claims description 6
- 238000001514 detection method Methods 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 4
- 208000034699 Vitreous floaters Diseases 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000012217 deletion Methods 0.000 claims description 3
- 230000037430 deletion Effects 0.000 claims description 3
- 238000003384 imaging method Methods 0.000 claims description 3
- 238000001228 spectrum Methods 0.000 claims description 3
- 230000003068 static effect Effects 0.000 claims description 3
- 230000001131 transforming effect Effects 0.000 claims description 3
- 238000013519 translation Methods 0.000 claims description 3
- 238000011161 development Methods 0.000 claims description 2
- 238000009877 rendering Methods 0.000 claims description 2
- 230000010354 integration Effects 0.000 claims 1
- 238000009434 installation Methods 0.000 abstract description 6
- 230000019771 cognition Effects 0.000 abstract description 5
- 230000008569 process Effects 0.000 description 7
- 238000010801 machine learning Methods 0.000 description 4
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/36—Preventing errors by testing or debugging software
- G06F11/3664—Environments for testing or debugging software
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/36—Preventing errors by testing or debugging software
- G06F11/3668—Software testing
- G06F11/3672—Test management
- G06F11/3688—Test management for test execution, e.g. scheduling of test suites
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/36—Preventing errors by testing or debugging software
- G06F11/3668—Software testing
- G06F11/3696—Methods or tools to render software testable
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Hardware Design (AREA)
- Quality & Reliability (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a marine unmanned cluster intelligent algorithm test training system, which comprises an algorithm test training functional module and a basic support functional module; the basic support function module comprises a scene management module, a task management module, a history management module, an algorithm management module and a model management module. The system constructs a system capable of intelligently testing and training the unmanned cluster perception and cognition algorithm, the mission planning and the cooperative control algorithm, so that algorithm parameters can be fully optimized before the algorithm is accessed to the actual installation, and the actual installation test risk is reduced.
Description
Technical Field
The invention belongs to the field of artificial intelligence, and particularly relates to a marine unmanned cluster intelligent algorithm test training system.
Background
The unmanned cluster perception cognition, task planning and cooperative control algorithm is a core technology and a key algorithm of an unmanned cluster, but a mature and general cluster algorithm test training environment does not exist at present, the unmanned cluster task planning and cooperative control algorithm is connected to the actual assembly operation after limited simulation and interface test are completed, the pressure test under extreme conditions and an exhaustive scene is not performed on the algorithm, and accidents are easy to occur during the actual assembly operation, so that a large risk exists.
Disclosure of Invention
Aiming at the lack of the current cluster intelligent algorithm test training platform, the invention provides a cluster perception, planning and control algorithm training test system to fully test and train the capability of a cluster intelligent algorithm, thereby achieving the purposes of optimizing algorithm parameters and reducing the risk of real-installation test.
The invention provides a marine unmanned cluster intelligent algorithm test training system, which comprises an algorithm test training functional module and a basic support functional module; the basic support function module comprises a scene management module, a task management module, a history management module, an algorithm management module and a model management module;
the scene management module is used for environment editing and scene setting; the environment is that 3-dimensional simulation software is used for simulating marine environment, weather environment, terrain environment and obstacles, scene management is used for managing scene files stored in a simulator, newly building, modifying and deleting scenes, and customizing specific task scenes aiming at different test tasks, wherein the scene contents comprise ocean, weather, terrain, static/dynamic obstacles, a target task area, a restricted navigation area and tasks to be tested;
the task management module is used for setting the type of a test algorithm, the number of times of algorithm test cycles or the termination condition of the test and the algorithm evaluation, and managing the test and training tasks through a task file, wherein the task file is used for setting the algorithm to be tested and the number of times of the cycle test or the termination condition of the test and the algorithm on the basis of a scene file to realize the cycle iterative test of the algorithm; the algorithm evaluation comprises the evaluation of the cluster task planning time and the re-planning time of the cluster task planning algorithm, the formation transformation time of the cooperative control algorithm, the formation holding precision and the cluster collision and obstacle avoidance capability;
the history management module is used for history recording and history playback;
the algorithm management module is used for managing the algorithm version, starting and stopping the algorithm and uploading the algorithm after the algorithm test is successful;
the model management module is used for managing a motion mechanism model and a detection perception model; the motion mechanism model management is that a packaged hydrodynamic calculation module is communicated with a simulator kernel, current pose, rudder angle and speed information sent by the simulator kernel is received, and pose information at the next moment is output to a simulator after calculation; the stored motion mechanism models comprise an Abkowitz model and an MMG model, and the motion mechanisms of various ships and boats are simulated by adjusting parameters of different models;
the detection perception model management comprises photoelectric perception simulation, inertial navigation simulation and GNSS simulation;
the photoelectric sensing simulation is based on a 3-dimensional software development platform, rendered image data are obtained, image data including various boats are generated, a camera coordinate system is obtained through translation and rotation of a world coordinate system, the camera coordinate system is converted into an image coordinate system based on a small hole imaging model and a similar triangle principle, and the image coordinate system is converted into a pixel coordinate system through a coordinate origin and pixel conversion to obtain a photoelectric sensing simulation image;
the GNSS simulation is to convert the Cartesian coordinates of the development platform into longitude and latitude;
the inertial navigation simulation obtains the approximate angular acceleration through the three-axis rotation angle, the quotient of the angular difference value and the time difference of two close moments, and obtains the approximate acceleration through reading the position absolute coordinate and transforming the position absolute coordinate into a coordinate system of a cost body.
Further, the marine environment simulation generates real-time marine conditions by using fast Fourier transform according to set wave spectrum data of the sea, and the scene file records a spectrogram in a current set scene.
Further, the terrain environment simulation uses a terrain height map to complete the simulation of the terrain shape, uses a terrain texture map to complete the simulation of the terrain type, and the scene file records the terrain height map and texture map codes.
Further, the weather environment is simulated by using dynamic illumination and volume cloud to generate different weather conditions; the weather conditions comprise weather conditions in various time periods of sunny days, cloudy days, rain and snow and fog.
Further, the dynamic barrier simulation renders a moving body according to the real-time pose information of the dynamic barrier, wherein the moving body comprises a ship and a general floater, and a preset track graph is read in real time for rendering the moving body with a preset track; for the dynamic barrier, the scene file records the initial position, the initial azimuth angle, whether the current moving body is the own boat, whether the current moving body is the controlled boat, the control port and whether the current moving body is in a multicast mode in the scene.
Further, the target task area is set for setting the position, shape and size of the target task area and is recorded in the scene file;
the no-navigation area is set for setting the position, shape and size of the no-navigation area and is recorded in a scene file;
the task to be tested is set for setting the task type, task attribute and port information of sending and receiving of each data of algorithm test training and is recorded in a scene file.
Further, the history records the moving object data by using a history file: the method comprises the steps of obtaining time sequence pose information, time sequence instruction data, an operation scene name, a test algorithm name and a test score; the time sequence position and posture information comprises a three-axis position and a three-axis inclination angle, and the time sequence instruction data comprises inertial navigation simulation data and control instruction data.
Furthermore, the history playback is to analyze the history file, perform history playback work according to the position and the posture of the movable object in the scene file, use the history data as a key frame according to the initial scene data and the moving body time sequence pose data, draw the time sequence pose of the moving body into a three-dimensional scene by using linear interpolation, and realize the variable-speed playing of the scene by setting the interval of the interpolation.
Further, algorithm version management comprises the setting of addition, update and deletion of algorithms, and various algorithms are integrated;
the algorithm start-stop is used for controlling an intelligent algorithm to run when the simulator sends a start instruction, and controlling the algorithm to stop testing or training when a training test stop condition is met;
and the algorithm uploading function is used for uploading the algorithm to the storage server through the FTP after the algorithm test is successful, and the system automatically compiles the measured and calculated method name and version according to the test training time.
Compared with the prior art, the invention has the following advantages and beneficial effects:
the system constructs a system capable of intelligently testing and training the unmanned cluster perception and cognition algorithm, the mission planning and the cooperative control algorithm, so that algorithm parameters can be fully optimized before the algorithm is accessed to the actual installation, and the actual installation test risk is reduced.
Drawings
FIG. 1 is a frame diagram of a marine unmanned-cluster intelligent-algorithm test training system according to an embodiment of the invention;
FIG. 2 is a flow chart of intelligent algorithm test training of the marine unmanned cluster in an embodiment of the invention;
FIG. 3 is a 2-dimensional historical playback perspective view of an embodiment of the present invention;
fig. 4 is a 3-dimensional historical playback perspective view of an embodiment of the invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the invention. In addition, the technical features involved in the respective embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
The invention belongs to the field of artificial intelligence, and particularly relates to a test training system applicable to intelligent algorithms of unmanned ships at sea, unmanned ship groups, unmanned submersible ships and cross-domain unmanned clusters.
The offshore unmanned-cluster intelligent algorithm test training system provided by the embodiment of the invention mainly comprises an algorithm test training functional module and a basic support functional module, and the system composition is shown in figure 1. The basic support function module is composed of 5 modules, namely a scene management module, a task management module, a history management module, an algorithm management module and a model management module, and the technical key points of the modules are as follows:
(1) a scene management module:
the scene management module comprises the functions of environment editing and scene setting. The environment simulation function is to utilize 3-dimensional simulation software to complete the simulation of marine environment, weather environment, terrain environment and obstacles (ships). The scene management function is mainly used for managing scene files stored in the simulator, new construction, modification and deletion can be carried out on the scene, the scene contents mainly comprise oceans, weather, terrain, static/dynamic obstacles, target task areas, no-navigation areas, tasks to be tested and the like, and specific task scenes can be defined by users aiming at different testing tasks.
The marine environment simulation uses fast Fourier transform to generate real-time marine conditions according to set marine wave spectrum data, and the scene file records a spectrogram in a current set scene.
The terrain environment simulation uses a terrain height map to complete the simulation of terrain shape, uses a terrain texture map to complete the simulation of terrain type, and a scene file records the terrain height map and texture map codes.
The weather environment simulation uses dynamic illumination and volume cloud to generate different weather conditions, including weather conditions in various time periods such as sunny days, cloudy days, rain and snow, fog and the like.
The method comprises the steps that barrier (ship) simulation is carried out, moving bodies (including ships and general floaters) are rendered according to dynamic barrier real-time pose information, and for the moving bodies with preset tracks, a preset track graph is read in real time to be rendered; for the dynamic barrier, the scene file records the initial position, the initial azimuth angle, whether the current moving body is the ship of the own party, whether the current moving body is the controlled ship, the control port, whether the current moving body is in a multicast form and other information in the scene.
The target task area setting is mainly used for setting the position, shape, size and the like of the target task area and is recorded in the scene file.
The no-navigation area setting is mainly used for setting the position, shape, size and the like of the no-navigation area and recording the no-navigation area in a scene file.
The task setting to be tested is mainly used for setting information such as task types, task attributes and ports for sending and receiving data of algorithm test training, and is recorded in a scene file.
(2) The task management module:
the main functions of the task management module include: the method comprises the following steps of setting the type of a test algorithm, setting the number of algorithm test cycles or the termination condition of the test and evaluating the algorithm, mainly managing the test and training tasks through task files, and setting the algorithm to be tested and the number of the cycle test and training cycles or the termination condition of the training on the basis of scene files by the task files to realize the cycle iterative test of the algorithm. The algorithm evaluation function comprises evaluation of cluster task planning time and re-planning time of the cluster task planning algorithm, formation transformation time of the cooperative control algorithm, formation holding precision, cluster collision and obstacle avoidance capability and the like.
(3) A history management module:
the functions of the history management module include history recording and history playback functions.
Wherein, the history record mainly records the data of the moving body: the method comprises time sequence position and attitude information (three-axis position and three-axis inclination angle), time sequence instruction data (including inertial navigation simulation data and control instruction data), an operation scene name, a test algorithm name, a test score and the like.
The history playback function mainly comprises the steps of analyzing a history file, performing history playback work according to the position and the posture of a movable object in a scene file, using the history data as key frames according to initial scene data and moving body time sequence position and posture data, drawing the time sequence position and posture of a moving body into a three-dimensional scene by using linear interpolation, and realizing variable-speed playing of the scene by setting interpolation intervals.
(4) An algorithm management module:
the algorithm management module mainly comprises the functions of algorithm version management, algorithm start and stop, algorithm uploading and the like.
The algorithm version management function comprises the following steps: the method comprises the steps of adding, updating and deleting the algorithms, and integrating various algorithms.
An algorithm start-stop function: the simulator sends a starting instruction to control an intelligent algorithm to run, and the algorithm is controlled to stop testing or training when a training test stopping condition is reached.
The algorithm uploading function: after the algorithm test is successful, the user uploads the algorithm to the storage server through the FTP protocol, and the system automatically compiles the measured and calculated method name and version according to the test training time.
(5) A model management module:
the functions of the model management module comprise movement mechanism model management and detection perception model management.
The motion mechanism model management is that the packaged hydrodynamic calculation module is communicated with the simulator kernel, and receives the information of the current pose, rudder angle, speed and the like sent by the simulator kernel, and after calculation, the pose information at the next moment is output to the simulator. And the hydrodynamic calculation module calculates and generates the pose at the next moment according to the current pose, the rudder angle, the navigational speed and the motion mechanism model. Wherein the stored motion mechanism model comprises: the motion mechanism of various ships can be simulated by adjusting parameters of different models such as an Abkowitz model and an MMG model.
The detection perception model includes: photoelectric sensing simulation, inertial navigation simulation, GNSS simulation and the like.
The photoelectric sensing simulation is based on a 3-dimensional software development platform, rendered image data are obtained, image data including various boats can be generated, a camera coordinate system is obtained through translation and rotation of a world coordinate system, the camera coordinate system is converted into an image coordinate system based on a small hole imaging model and a similar triangle principle, and the image coordinate system is converted into a pixel coordinate system through a coordinate origin and pixel conversion to obtain a photoelectric sensing simulation image.
The GNSS simulation is mainly converted into latitude and longitude in WGS-84 format by reading Cartesian coordinates of a 3-dimensional software development platform.
The inertial navigation simulation obtains the approximate angular acceleration through the three-axis rotation angle, the quotient of the angular difference value and the time difference of two close moments, and obtains the approximate acceleration through reading the position absolute coordinate and transforming the position absolute coordinate into a coordinate system of a cost body.
According to the system, a system capable of intelligently testing and training the unmanned cluster perception cognition algorithm, the task planning and cooperative control algorithm is constructed through a scene management module, a task management module, a history management module, an algorithm management module and a model management module 5, so that algorithm parameters can be fully optimized before the algorithm is connected to actual assembly, and the actual assembly test risk is reduced.
Example 1:
as shown in fig. 3, the objective of a certain cluster decision-making planning algorithm is to complete capture of an unmanned ship from outside under a level 3 sea state, and prevent the unmanned ship from approaching within 100m of the center of a circle within 3 hours.
In order to complete the training test of the cluster decision planning algorithm, the test training process is shown in fig. 2, the goal of completing the algorithm training is manually set in the task management module, the training is stopped when the success rate of the task is more than 90%, the entering mode of an external boat is random, and the sea state is 3 grades.
The number of the external boats is manually set to be 1 in the environment management module, and the number of the existing unmanned boats is set to be 8.
The task management module calls an unmanned ship motion mechanism model and a perception model in the model management module according to environment setting, a tested algorithm is called through the algorithm management module, interface intercommunication self-checking is completed, and a training test process is started. For the algorithm based on machine learning, according to a set entering mode of the external boat, the external boat in each direction is randomly generated, and the task management module records the distance of the external boat close to the circle center in the operation process of the algorithm. And if the distance is close to the circle center within 100m within 3 hours, the task is failed to be fed back to the algorithm, and the task times and the task success rate are recorded.
The history recording module records the position and environment data of the unmanned ship at each moment, and the playback module calls the data to form history playback.
And adjusting algorithm parameters according to the feedback of the task management module by the algorithm based on machine learning, and randomly generating external boats when the task success rate is lower than 90% until the task success rate of the algorithm is higher than 90%.
When the task success rate of the algorithm is higher than 90%, the algorithm management module stores the algorithm and marks a time version.
And the algorithm management module uploads the algorithm to the unmanned boat end through a transmission protocol according to manual needs.
Fig. 3 shows a 2-dimensional view of a training process replay of the algorithm.
Example 2:
as shown in fig. 4, the objective of an unmanned swarm cooperative control algorithm is to make the swarm smoothly pass through a narrow water course through formation transformation under the sea state of level 2, and the success rate is higher than 98%.
In order to complete the training test of the cluster cooperation algorithm, according to the test training flow of fig. 2, the goal of completing the algorithm training is manually set in the task management module to stop training when the success rate of the random water channel task is greater than 98%, the generation mode of the narrow water channel is set to be random, and the sea state is set to be level 2.
The number of the overbending of the narrow water channel is manually set to be 5 in the environment management module, and the number of the unmanned boats is set to be 4.
The task management module calls an unmanned ship motion mechanism model and a perception model in the model management module according to the environment setting, calls a tested algorithm through the algorithm management module, completes interface intercommunication self-checking and starts a training test process. For an algorithm based on machine learning, various narrow water channels with 5 curved channels are randomly generated by adopting terrains or restricted areas according to a set narrow water channel generation mode, a task management module records the condition that clusters collide with the water channels in the operation process of the algorithm, the narrow water channels pass through smoothly, and the task is recorded as successful. And if the cluster collides with the water channel, the task is failed to be fed back to the algorithm, and the task times and the task success rate are recorded.
The historical recording module records the position and the environment data of the unmanned ship at each moment, and the playback module calls the data to form historical playback.
And adjusting algorithm parameters according to the feedback of the task management module by the algorithm based on machine learning, and when the task success rate is lower than 98%, continuing to randomly generate a narrow water channel until the task success rate of the algorithm is higher than 98%.
When the task success rate of the algorithm is higher than 98%, the algorithm management module stores the algorithm and marks a time version.
And the algorithm management module uploads the algorithm to the unmanned boat end through a transmission protocol according to manual needs.
Fig. 4 shows a 3-dimensional view of a training process replay of the algorithm.
Aiming at improvement of research and development requirements of various unmanned equipment, the invention provides an algorithm universal test training platform with marine environment simulation, weather simulation, cluster movement mechanism simulation and obstacle simulation, and fully tests and trains a cluster perception cognition algorithm, a task planning algorithm and a cooperative control algorithm by an intelligent test training means, so that algorithm parameters are optimized, and the risk of a real-installation test is reduced.
It will be understood by those skilled in the art that the foregoing is only an exemplary embodiment of the present invention, and is not intended to limit the invention to the particular forms disclosed, since various modifications, substitutions and improvements within the spirit and scope of the invention are possible and within the scope of the appended claims.
Claims (9)
1. An intelligent algorithm test training system for an offshore unmanned cluster is characterized by comprising an algorithm test training function module and a basic support function module; the basic support function module comprises a scene management module, a task management module, a history management module, an algorithm management module and a model management module;
the scene management module is used for environment editing and scene setting; the environment is that 3-dimensional simulation software is used for simulating marine environment, weather environment, terrain environment and obstacles, scene management is used for managing scene files stored in a simulator, newly building, modifying and deleting scenes, and customizing specific task scenes aiming at different test tasks, wherein the scene contents comprise ocean, weather, terrain, static/dynamic obstacles, a target task area, a restricted navigation area and tasks to be tested;
the task management module is used for setting the type of a test algorithm, the number of algorithm test cycles or the termination condition of the test and the algorithm evaluation, and managing the test and training tasks through a task file, wherein the task file is used for setting the algorithm to be tested and the number of the cycle test and the training cycle or the termination condition of the test and the algorithm on the basis of a scene file to realize the cycle iterative test of the algorithm; the algorithm evaluation comprises the evaluation of the cluster task planning time and the re-planning time of the cluster task planning algorithm, the formation transformation time of the cooperative control algorithm, the formation holding precision and the cluster collision and obstacle avoidance capability;
the history management module is used for history recording and history playback;
the algorithm management module is used for algorithm version management, algorithm start and stop and algorithm uploading after algorithm test is successful;
the model management module is used for managing a motion mechanism model and a detection perception model; the motion mechanism model management is to communicate the packaged hydrodynamic calculation module with a simulator kernel, receive current pose, rudder angle and speed information sent by the simulator kernel, and output pose information at the next moment to the simulator after calculation; the stored motion mechanism models comprise an Abkowitz model and an MMG model, and the motion mechanisms of various ships and boats are simulated by adjusting parameters of different models;
the detection perception model management comprises photoelectric perception simulation, inertial navigation simulation and GNSS simulation;
the photoelectric perception simulation is based on a 3-dimensional software development platform, rendered image data are obtained, image data including various boats are generated, a camera coordinate system is obtained through translation and rotation of a world coordinate system, the camera coordinate system is converted into an image coordinate system based on a small hole imaging model and a similar triangle principle, and the image coordinate system is converted into a pixel coordinate system through a coordinate origin and pixel conversion to obtain a photoelectric perception simulation image;
the GNSS simulation is to convert a Cartesian coordinate of a development platform into latitude and longitude;
the inertial navigation simulation obtains approximate angular acceleration through a triaxial rotation angle, a quotient of an angular difference value and a time difference of two close moments, and obtains the approximate acceleration through reading a position absolute coordinate and transforming a cost coordinate system.
2. The offshore unmanned-cluster intelligent algorithm test training system as claimed in claim 1, wherein the marine environment simulation uses fast fourier transform to generate real-time marine conditions according to set marine wave spectrum data, and the scene file records a spectrogram in a currently set scene.
3. The marine unmanned-swarm intelligent algorithm test training system of claim 1, wherein the terrain environment simulation uses a terrain height map to complete the terrain shape simulation, uses a terrain texture map to complete the terrain type simulation, and the scene file records the terrain height map and texture map coding.
4. The offshore unmanned-cluster intelligent algorithm test training system of claim 1, wherein the weather environment simulation uses dynamic lighting, volume clouds to generate different weather conditions; the weather conditions comprise weather conditions in various time periods of sunny days, cloudy days, rain and snow and fog.
5. The offshore unmanned cluster intelligent algorithm test training system of claim 1, wherein dynamic barrier simulation renders moving bodies according to real-time pose information of dynamic barriers, wherein the moving bodies comprise ships and general floaters, and for the moving bodies with preset tracks, a preset track graph is read in real time for rendering; for the dynamic barrier, the scene file records the initial position, the initial azimuth, whether the current moving body is the ship of our party, whether the current moving body is the controlled ship, the control port and whether the current moving body is in a multicast form.
6. The offshore unmanned-cluster intelligent algorithm test training system of claim 1, wherein the target task area is configured to set a position, a shape, and a size of the target task area, and is recorded in a scene file;
the no-navigation area is set for setting the position, shape and size of the no-navigation area and is recorded in a scene file;
the task to be tested is set for setting the task type, task attribute and port information of sending and receiving of each data of algorithm test training and is recorded in a scene file.
7. The offshore unmanned swarm intelligence algorithm test training system of claim 1, wherein the history record is a record of moving object data using a history file: the method comprises the steps of obtaining time sequence pose information, time sequence instruction data, an operation scene name, a test algorithm name and a test score; the time sequence pose information comprises three-axis positions and three-axis inclination angles, and the time sequence instruction data comprises inertial navigation simulation data and control instruction data.
8. The offshore unmanned swarm intelligent algorithm test training system of claim 7, wherein the history playback is performed by parsing a history file, performing history playback by the position and posture of a movable object in a scene file, using the history data as a key frame according to the initial scene data and the moving object time sequence pose data, drawing the time sequence pose of the moving object into a three-dimensional scene by using linear interpolation, and realizing variable-speed playing of the scene by setting an interval of the interpolation.
9. The offshore unmanned-cluster intelligent algorithm test training system of claim 1, wherein algorithm version management comprises setting of addition, update and deletion of algorithms, and integration of multiple algorithms is performed;
the algorithm start-stop is used for controlling an intelligent algorithm to run when the simulator sends a start instruction, and controlling the algorithm to stop testing or training when a training test stop condition is reached;
and the algorithm uploading function is used for uploading the algorithm to the storage server through the FTP after the algorithm test is successful, and the system automatically compiles the measured and calculated method name and version according to the test training time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210369011.1A CN114780393B (en) | 2022-04-08 | 2022-04-08 | Marine unmanned cluster intelligent algorithm test training system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210369011.1A CN114780393B (en) | 2022-04-08 | 2022-04-08 | Marine unmanned cluster intelligent algorithm test training system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114780393A true CN114780393A (en) | 2022-07-22 |
| CN114780393B CN114780393B (en) | 2024-04-09 |
Family
ID=82428750
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210369011.1A Active CN114780393B (en) | 2022-04-08 | 2022-04-08 | Marine unmanned cluster intelligent algorithm test training system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114780393B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117130987A (en) * | 2023-10-27 | 2023-11-28 | 北京卓翼智能科技有限公司 | Flight control management method for large-scale unmanned aerial vehicle cluster |
| WO2024119471A1 (en) * | 2022-12-09 | 2024-06-13 | 京东方科技集团股份有限公司 | Algorithm testing method, algorithm testing platform, and computer readable storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112182977A (en) * | 2020-10-12 | 2021-01-05 | 中国人民解放军国防科技大学 | Control method and system for cooperative game confrontation of unmanned cluster |
| CN112507622A (en) * | 2020-12-16 | 2021-03-16 | 中国人民解放军国防科技大学 | Anti-unmanned aerial vehicle task allocation method based on reinforcement learning |
| CN113312172A (en) * | 2021-02-23 | 2021-08-27 | 中国人民解放军国防科技大学 | Multi-unmanned aerial vehicle cluster dynamic task scheduling model based on adaptive network |
| US20210294356A1 (en) * | 2020-03-19 | 2021-09-23 | Beihang University | Dynamic recovery method and system for uavs and storage medium |
-
2022
- 2022-04-08 CN CN202210369011.1A patent/CN114780393B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210294356A1 (en) * | 2020-03-19 | 2021-09-23 | Beihang University | Dynamic recovery method and system for uavs and storage medium |
| CN112182977A (en) * | 2020-10-12 | 2021-01-05 | 中国人民解放军国防科技大学 | Control method and system for cooperative game confrontation of unmanned cluster |
| CN112507622A (en) * | 2020-12-16 | 2021-03-16 | 中国人民解放军国防科技大学 | Anti-unmanned aerial vehicle task allocation method based on reinforcement learning |
| CN113312172A (en) * | 2021-02-23 | 2021-08-27 | 中国人民解放军国防科技大学 | Multi-unmanned aerial vehicle cluster dynamic task scheduling model based on adaptive network |
Non-Patent Citations (1)
| Title |
|---|
| 刘立辉;赵彦杰;赵小虎;李志飞;李岩;: "一种无人集群系统仿真平台设计", 中国电子科学研究院学报, no. 05, 20 October 2017 (2017-10-20) * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2024119471A1 (en) * | 2022-12-09 | 2024-06-13 | 京东方科技集团股份有限公司 | Algorithm testing method, algorithm testing platform, and computer readable storage medium |
| CN117130987A (en) * | 2023-10-27 | 2023-11-28 | 北京卓翼智能科技有限公司 | Flight control management method for large-scale unmanned aerial vehicle cluster |
| CN117130987B (en) * | 2023-10-27 | 2024-02-23 | 北京卓翼智能科技有限公司 | Flight control management method for large-scale unmanned aerial vehicle cluster |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114780393B (en) | 2024-04-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110221546B (en) | Virtual-real integrated ship intelligent control system test platform | |
| CN114780393B (en) | Marine unmanned cluster intelligent algorithm test training system | |
| Corke et al. | Experiments with underwater robot localization and tracking | |
| US20210191400A1 (en) | Autonomous vessel simulation system and operating method thereof | |
| AU2018436279B2 (en) | System and method of operation for remotely operated vehicles for simultaneous localization and mapping | |
| CN105241457A (en) | Establishing method of three-dimensional aided navigation system for ship handling | |
| CN109714567A (en) | A kind of real-time construction method of three-dimensional virtual scene based on infrared viewing device and device | |
| CN106292656B (en) | A kind of environmental modeling method and device | |
| CN113781650B (en) | Dynamic object visual three-dimensional reconstruction method and system based on data fusion | |
| CN111292419A (en) | Intelligent ship navigation digital sand table system based on electronic chart | |
| CN114692520B (en) | Multi-scene-oriented unmanned ship virtual simulation test platform and test method | |
| CN117214904A (en) | Intelligent fish identification monitoring method and system based on multi-sensor data | |
| Hsu et al. | Stereo-vision-based AUV navigation system for resetting the inertial navigation system error | |
| Lončar et al. | MARUS-a marine robotics simulator | |
| CN112033640B (en) | Simulation experience system for ship hydrodynamic performance | |
| CN113593043A (en) | Point cloud three-dimensional reconstruction method and system based on generation countermeasure network | |
| CN111413691A (en) | Semantic positioning and mapping method adopting distributed structure | |
| CN117911886A (en) | Space target intelligent detection system and method based on digital twin technology | |
| Ribas Romagós | Underwater slam for estructured environments using and imaging sonar | |
| Xanthidis et al. | Towards multi-robot shipwreck mapping | |
| CN115687453A (en) | Dynamic modeling method for ship running environment based on situation awareness data drive | |
| Allotta et al. | Thesaurus project: Design of new autonomous underwater vehicles for documentation and protection of underwater archaeological sites | |
| CN113989350A (en) | Monitoring system for unmanned ship autonomous exploration and unknown environment three-dimensional reconstruction | |
| Xue-min et al. | Research and Practice of Key Technologies of Inland Intelligent Ships | |
| CN113010958B (en) | Simulation system of self-propelled ship and operation method thereof |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |