A dynamic obstacle avoidance and capture control system for intelligent multiple underwater robots

Abstract

The invention discloses an intelligent multi-underwater robot dynamic obstacle avoidance and encirclement control system, which relates to the field of dynamic encirclement technology. An object recognition module is set, and a CNN deep neural network model is used to train a model that can identify encirclement targets. After the underwater robot finds the encirclement target, an encirclement construction module is set, and the position of entering the encirclement is calculated in advance for each underwater robot through the encirclement range pre-constructed around the encirclement target. Each underwater robot directly goes to the position of entering the encirclement from the current position to complete the encirclement. When a certain underwater robot encounters an obstacle that is difficult to bypass, other underwater robots dynamically construct a new encirclement. After the encirclement is completed, the encirclement target is captured by reducing the encirclement. The problems of dynamic encirclement and dynamic obstacle avoidance are solved.

Description

Technical Field

[0001] The invention belongs to the field of underwater robots and relates to a dynamic capture technology, in particular to an intelligent multi-underwater robot dynamic obstacle avoidance and capture control system. Background Art

[0002] As the application field of mobile robots continues to expand, the requirements for robots are also getting higher and higher. When faced with some complex tasks that need to be completed in parallel, a single robot is not competent, and multiple robots need to coordinate and cooperate to complete the task together. Among the tasks of multi-robot coordination and cooperation, underwater capture tasks are difficult because the capture target is in motion and the underwater environment is deeply affected by the environment;

[0003] In the existing underwater capture technology, static capture targets are often considered, and the problem of avoiding obstacles is not considered during the capture process;

[0004] Therefore, a dynamic obstacle avoidance and capture control system for intelligent multiple underwater robots was proposed. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes a dynamic obstacle avoidance and capture control system for multiple intelligent underwater robots, which sets an object recognition module and uses a CNN deep neural network model to train a model that can identify the capture target; after the underwater robot finds the capture target, an encirclement construction module is set, and the position of entering the encirclement is calculated in advance for each underwater robot by constructing an encirclement range around the capture target in advance; each underwater robot can complete the encirclement by going directly from the current position to the position of entering the encirclement; and when a certain underwater robot encounters an obstacle that is difficult to bypass, other underwater robots dynamically construct a new encirclement; and after the encirclement is completed, the capture target is captured by shrinking the encirclement; the problem of dynamic capture and dynamic obstacle avoidance is solved.

[0006] To achieve the above-mentioned purpose, an embodiment of the first aspect of the present invention provides an intelligent multi-underwater robot dynamic obstacle avoidance and encirclement control system, comprising an object recognition module, an encirclement construction module, a dynamic obstacle avoidance module and an encirclement module;

[0007] The object recognition module is used to distinguish between capture targets and non-capture targets;

[0008] The object recognition module includes an object recognition program installed on each underwater robot; the object recognition program distinguishes between a capture target and a non-capture target and includes the following steps:

[0009] Step S1: the object recognition module manually collects a number of images of encircled targets and non-encircled targets in advance; and marks each image as an encircled target and a non-encircled target; the non-encircled target image can be an image of any object other than the encircled target;

[0010] Step S2: input the collected images as training sets into the CNN neural network model, take the capture target and non-capture target as target values, judge whether each image in the training set is a capture target; and take the judgment accuracy as the training target; train the CNN neural network model; until the accuracy reaches more than 95%; mark the trained CNN neural network model as M;

[0011] During the movement, the underwater robot acquires the environment in the field of view in real time through the smart camera it carries; the smart camera sends the environment in the field of view to the object recognition module, and the object recognition module uses the CNN neural network model M to determine in real time whether there are capture targets and non-capture targets in the field of view;

[0012] Each underwater robot carries a positioning and communication device; the positioning device is used to determine its own position; the communication device is used to exchange information with other underwater robots in real time; each underwater robot broadcasts its own position information to other underwater robots via wireless signals in real time;

[0013] When an underwater robot pre-discovers a capture target through an object recognition module, it sends the capture target position, moving direction and speed to other underwater robots; wherein the moving direction and speed are determined according to the position change of the capture target after a number of video frames;

[0014] The encirclement construction module is mainly used for the underwater robot to construct an encirclement to surround the encirclement target after discovering the encirclement target;

[0015] The encirclement construction module constructs an encirclement comprising the following steps:

[0016] Step P1: Each underwater robot determines its own position in the encirclement according to the positions of other underwater robots and the position, moving direction and speed of the encirclement target;

[0017] Each underwater robot confirms its position in the encirclement by the following steps:

[0018] Step Q1: mark the number of underwater robots as N; number the underwater robot closest to the capture target as No. 1; in the horizontal direction, take No. 1 underwater robot as the starting point, rotate clockwise, and number the remaining underwater robots from No. 2 to No. N in sequence;

[0019] Step Q2: The underwater robot No. 1 calculates the range of the encirclement in the horizontal direction, with the encirclement target as the center and R as the radius; R is the encirclement radius set according to actual experience; the robot No. 1 calculates the position of the point closest to the encirclement; and marks the position as P1;

[0020] Step Q3: Machine No. 1 broadcasts position P1 to the other underwater robots; the other underwater robots calculate their own positions entering the encirclement respectively; the calculation method is to start from point P1 and rotate clockwise along a circle with the encirclement target as the center and R as the radius, and every rotation distance The positions of the underwater robots numbered 2, 3, ...N entering the encirclement are marked as Pi, where i = 1, 2, ...N;

[0021] Step P2: It is understandable that the position of the capture target changes in real time; therefore, underwater robot No. 1 obtains the position, moving speed and direction of the capture target in real time; and broadcasts the position, moving speed and direction of the capture target to other underwater robots;

[0022] Each underwater robot i synchronously updates its position Pi of entering the encirclement in real time, and moves to position Pi at the maximum speed along the straight line between the underwater robot and the encirclement target in real time; until the distance to the encirclement target is less than R+r; where r is the control distance set according to actual experience;

[0023] Step P3: When the distance to the capture target is less than R+r, the underwater robot i moves toward the position Pi at a speed greater than the amplitude α of the capture target's moving speed; the amplitude α is set according to actual experience; until each underwater robot i reaches the corresponding position Pi; the speed of the underwater robot is adjusted to be the same as the moving speed of the capture target;

[0024] The dynamic obstacle avoidance module is mainly used for the underwater robot to actively avoid non-encirclement targets.

[0025] In a preferred embodiment, the dynamic obstacle avoidance module is a decision-making program installed on each underwater robot; during the movement of the underwater robot i, the object recognition module analyzes objects appearing in the forward route in real time; when the detected object is a non-hunting target, the dynamic obstacle avoidance module uses image analysis technology to calculate the volume of the non-hunting target;

[0026] It is understandable that a long time to avoid obstacles will cause the underwater robot i's forward route plan to be delayed, which may cause the encircled target to escape from the direction of the underwater robot i; therefore, if the target volume is greater than a preset volume threshold V, the underwater robot i sends an encirclement reconstruction signal to other underwater robots; the volume threshold V is set according to actual experience; after receiving the encirclement reconstruction signal, the other underwater robots reconfirm their own positions in the encirclement; at this time, the encirclement position does not include the position of the underwater robot i; until the underwater robot i bypasses the non-encircled target, each underwater robot reconfirms its own position in the encirclement;

[0027] The capture module is mainly used to capture the capture target in the encirclement;

[0028] In a preferred embodiment, the capture module is a capture program on each underwater robot; when the underwater robots form an encirclement around a capture target, each underwater robot moves toward the capture target to shorten the distance to the capture target; when the distance between a certain underwater robot and the capture target is less than the capture distance, the capture target is captured.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] The present invention sets an object recognition module and uses a CNN deep neural network model to train a model that can identify the encirclement target; after the underwater robot finds the encirclement target, an encirclement construction module is set, and the position of entering the encirclement is calculated in advance for each underwater robot through the encirclement range constructed in advance around the encirclement target; each underwater robot directly goes to the position of entering the encirclement from the current position to complete the encirclement; and when a certain underwater robot encounters an obstacle that is difficult to bypass, other underwater robots dynamically construct a new encirclement; and after the encirclement is completed, the encirclement target is captured by reducing the encirclement, thereby solving the problems of dynamic encirclement and dynamic obstacle avoidance. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is a schematic diagram of the present invention. DETAILED DESCRIPTION

[0032] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0033] like Figure 1As shown, an intelligent multi-underwater robot dynamic obstacle avoidance and encirclement control system includes an object recognition module, an encirclement construction module, a dynamic obstacle avoidance module and an encirclement module;

[0034] The object recognition module is mainly used to distinguish between the captured targets and the non-captured targets;

[0035] In a preferred real-time example, the object recognition module includes an object recognition program installed on each underwater robot; the object recognition program distinguishes between the capture target and the non-capture target and includes the following steps:

[0036] Step S1: the object recognition module manually collects a number of images of encircled targets and non-encircled targets in advance; and marks each image as an encircled target and a non-encircled target; the non-encircled target image can be an image of any object other than the encircled target;

[0037] Step S2: input the collected images as training sets into the CNN neural network model, take the capture target and non-capture target as target values, judge whether each image in the training set is a capture target; and take the judgment accuracy as the training target; train the CNN neural network model; until the accuracy reaches more than 95%; mark the trained CNN neural network model as M;

[0038] During the movement, the underwater robot acquires the environment in the field of view in real time through the smart camera it carries; the smart camera sends the environment in the field of view to the object recognition module, and the object recognition module uses the CNN neural network model M to determine in real time whether there are capture targets and non-capture targets in the field of view;

[0039] It is understandable that each underwater robot carries a positioning and communication device; the positioning device is used to determine its own position; the communication device is used to exchange information with other underwater robots in real time; each underwater robot broadcasts its own position information to other underwater robots via wireless signals in real time;

[0040] Furthermore, when an underwater robot pre-discovers a capture target through an object recognition module, it sends the capture target position, moving direction and speed to other underwater robots; wherein the moving direction and speed are determined according to the position change of the capture target after a number of video frames;

[0041] The encirclement construction module is mainly used for the underwater robot to construct an encirclement to surround the target after finding the target;

[0042] It is understandable that the position of the target is changing dynamically, and in order to ensure that the target has no escape direction, the best choice is to use the encirclement method;

[0043] In a preferred embodiment, the encirclement construction module constructs the encirclement including the following steps:

[0044] Step P1: Each underwater robot determines its own position in the encirclement according to the positions of other underwater robots and the position, moving direction and speed of the encirclement target;

[0045] Each underwater robot confirms its position in the encirclement by the following steps:

[0046] Step Q1: mark the number of underwater robots as N; number the underwater robot closest to the capture target as No. 1; in the horizontal direction, take No. 1 underwater robot as the starting point, rotate clockwise, and number the remaining underwater robots from No. 2 to No. N in sequence;

[0047] Step Q2: The underwater robot No. 1 calculates the range of the encirclement in the horizontal direction, with the encirclement target as the center and R as the radius; R is the encirclement radius set according to actual experience; the robot No. 1 calculates the position of the point closest to the encirclement; and marks the position as P1;

[0048] Step Q3: Machine No. 1 broadcasts position P1 to the other underwater robots; the other underwater robots calculate their own positions entering the encirclement respectively; the calculation method is to start from point P1 and rotate clockwise along a circle with the encirclement target as the center and R as the radius, and every rotation distance The positions of the underwater robots numbered 2, 3, ...N entering the encirclement are marked as Pi, where i = 1, 2, ...N;

[0049] Step P2: It is understandable that the position of the capture target changes in real time; therefore, underwater robot No. 1 obtains the position, moving speed and direction of the capture target in real time; and broadcasts the position, moving speed and direction of the capture target to other underwater robots;

[0050] Each underwater robot i synchronously updates its position Pi of entering the encirclement in real time, and moves to the position Pi at the maximum speed along the straight line between the underwater robot and the encircled target in real time; until the distance to the encircled target is less than R+r; where r is the control distance set according to actual experience; it can be understood that the maximum speed of the underwater robot should be greater than the moving speed of the encircled target;

[0051] Step P3: When the distance to the capture target is less than R+r, the underwater robot i moves toward the position Pi at a speed greater than the amplitude α of the capture target's moving speed; the amplitude α is set according to actual experience; until each underwater robot i reaches the corresponding position Pi; the speed of the underwater robot is adjusted to be the same as the moving speed of the capture target;

[0052] It can be understood that the use of a circular encirclement to encircle the encircled target in the above steps is only a preferred embodiment of the present invention; other encirclement shapes are within the protection scope of the present invention;

[0053] The dynamic obstacle avoidance module is mainly used for the underwater robot to actively avoid non-encirclement targets.

[0054] Understandably, the underwater environment is more complex; there are often obstacles in the underwater robot's path, which require the underwater robot to avoid in advance;

[0055] In a preferred embodiment, the dynamic obstacle avoidance module is a decision-making program installed on each underwater robot; during the movement of the underwater robot i, the object recognition module analyzes objects appearing in the forward route in real time; when the detected object is a non-hunting target, the dynamic obstacle avoidance module uses image analysis technology to calculate the volume of the non-hunting target;

[0056] It is understandable that a long time to avoid obstacles will cause the underwater robot i's forward route plan to be delayed, which may cause the encircled target to escape from the direction of the underwater robot i; therefore, if the target volume is greater than a preset volume threshold V, the underwater robot i sends an encirclement reconstruction signal to other underwater robots; the volume threshold V is set according to actual experience; after receiving the encirclement reconstruction signal, the other underwater robots reconfirm their own positions in the encirclement; at this time, the encirclement position does not include the position of the underwater robot i; until the underwater robot i bypasses the non-encircled target, each underwater robot reconfirms its own position in the encirclement;

[0057] The capture module is mainly used to capture the capture target in the encirclement;

[0058] It can be understood that when the encirclement is formed, the distance between each underwater robot and the encirclement target is maintained within the range of radius R; therefore, the encirclement target can be captured by only reducing the range of radius R;

[0059] In a preferred embodiment, the capture module is a capture program on each underwater robot; when the underwater robots form an encirclement around a capture target, each underwater robot moves toward the capture target to shorten the distance to the capture target; when the distance between a certain underwater robot and the capture target is less than the capture distance, the capture target is captured.

[0060] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An intelligent multi-underwater robot dynamic obstacle avoidance and capture control system, characterized in that: It includes object recognition module, encirclement construction module, dynamic obstacle avoidance module and capture module; The object recognition module is used to distinguish between capture targets and non-capture targets; During the movement, the underwater robot acquires the environment in the field of view in real time through the smart camera it carries. The smart camera sends the environment in the field of view to the object recognition module to identify whether the field of view contains the capture target. When an underwater robot pre-discovers the capture target through the object recognition module, it sends the capture target location, moving direction and speed to other underwater robots. The encirclement construction module is used for the underwater robot to calculate the corresponding encirclement position for each underwater robot after the underwater robot finds the encirclement target, so as to construct an encirclement surrounding the encirclement target; The dynamic obstacle avoidance module is used for the underwater robot to actively avoid non-encirclement targets when encountering them, and to make a decision on whether to reconstruct the encirclement. The encirclement module is used to capture the encirclement target when the encirclement is formed; The encirclement construction module constructs an encirclement comprising the following steps: Step P1: Each underwater robot confirms its own position in the encirclement according to the positions of other underwater robots and the position, moving direction and speed of the encirclement target; Step P2: The No. 1 underwater robot obtains the position, moving speed and direction of the capture target in real time; and broadcasts the capture target position, moving speed and direction to other underwater robots; Each underwater robot i synchronously updates its position Pi of entering the encirclement in real time, and moves to position Pi at the maximum speed along the straight line between the underwater robot and the encirclement target in real time; until the distance to the encirclement target is less than R+r; where r is the control distance set according to actual experience; Step P3: When the distance to the capture target is less than R+r, the underwater robot i moves toward the position Pi at a speed greater than the amplitude α of the capture target's moving speed; the amplitude α is set according to actual experience; until each underwater robot i reaches the corresponding position Pi; the speed of the underwater robot is adjusted to be the same as the moving speed of the capture target; Each underwater robot confirms its position in the encirclement by the following steps: Step Q1: mark the number of underwater robots as N; number the underwater robot closest to the capture target as No. 1; in the horizontal direction, take No. 1 underwater robot as the starting point, rotate clockwise, and number the remaining underwater robots from No. 2 to No. N in sequence; Step Q2: The underwater robot No. 1 calculates the range of the encirclement in the horizontal direction, with the encirclement target as the center and R as the radius; R is the encirclement radius set according to actual experience; the robot No. 1 calculates the position of the point closest to the encirclement; and marks the position as P1; Step Q3: Machine No. 1 broadcasts position P1 to the other underwater robots; the other underwater robots calculate their own positions entering the encirclement respectively; the calculation method is to start from point P1 and rotate clockwise along a circle with the encirclement target as the center and R as the radius, and every rotation distance The positions of the underwater robots numbered 2, 3, ...N entering the encirclement are marked as Pi, where i = 1, 2, ...N.

2. According to claim 1, the intelligent multi-underwater robot dynamic obstacle avoidance and capture control system is characterized in that: The object recognition module distinguishes between the captured target and the non-captured target, comprising the following steps: Step S1: the object recognition module manually collects a number of images of encircled targets and non-encircled targets in advance; and marks each image as an encircled target or a non-encircled target; Step S2: Input the collected images as a training set into the CNN neural network model, take the capture target and non-capture target as the target value, judge whether each image in the training set is a capture target; and take the judgment accuracy as the training target; train the CNN neural network model; until the accuracy reaches more than 95%; mark the trained CNN neural network model as M.

3. The intelligent multi-underwater robot dynamic obstacle avoidance and capture control system according to claim 1, characterized in that: The moving direction and speed are determined according to the position change of the captured target after a number of video frames.

4. The intelligent multi-underwater robot dynamic obstacle avoidance and capture control system according to claim 1, characterized in that: Each underwater robot carries a positioning and communication device; the positioning device is used to determine its own position; the communication device is used to exchange information with other underwater robots in real time; each underwater robot broadcasts its own position information to other underwater robots via wireless signals in real time.

5. The intelligent multi-underwater robot dynamic obstacle avoidance and capture control system according to claim 1, characterized in that: During the movement of the underwater robot i, the object recognition module analyzes objects appearing in the forward route in real time; when the detected object is a non-hunting target, the dynamic obstacle avoidance module uses image analysis technology to calculate the volume of the non-hunting target; If the target volume is larger than a preset volume threshold V, the underwater robot i sends an encirclement reconstruction signal to other underwater robots; the volume threshold V is set according to actual experience; after receiving the encirclement reconstruction signal, the other underwater robots reconfirm their positions in the encirclement; at this time, the position of the encirclement does not include the position of the underwater robot i; until the underwater robot i bypasses the non-encirclement target, each underwater robot reconfirms its position in the encirclement.

6. The intelligent multi-underwater robot dynamic obstacle avoidance and capture control system according to claim 1, characterized in that: When the underwater robots form an encirclement around the encirclement target, each underwater robot moves toward the encirclement target to shorten the distance to the encirclement target; when the distance between a certain underwater robot and the encirclement target is less than the capture distance, the encirclement target is captured.

Images