An unmanned ship autonomous exploration method based on fusion of RRT and image edge detection

Abstract

The application provides an unmanned ship autonomous exploration method combining RRT and image edge detection, which has the advantages of flexibility, low cost, high efficiency and safety without human intervention in the autonomous exploration of the unmanned ship in unknown waters. The application carries out an engineering application of the lightweight autonomous exploration of the unmanned ship in the unknown area by combining the dynamic pruning fast random tree and the image edge detection for detailed exploration of the local map. In the process of real-time positioning and map construction of the unmanned ship, the dynamic pruning fast random tree and the Canny edge detection algorithm are applied to obtain the exploration front point, so that the lightweight random tree is ensured while the comprehensiveness of the boundary point detection is ensured. Compared with the previous exploration method, the exploration target point evaluation function with turning constraint and the path planning algorithm combining Dijkstra and Dubins are proposed in view of the inherent motion characteristics of the unmanned ship, that is, the unmanned ship cannot self-turn during sailing and it is not easy to turn at a large angle in a narrow sea area, so as to reduce the generation of inefficient exploration caused by the motion defects.

Description

Technical Field

[0001] This invention relates to the field of autonomous exploration and decision-making technology for unmanned vessels, and more particularly to an autonomous exploration method for unmanned vessels that integrates RRT and image edge detection. Background Technology

[0002] With the increasing demand for work in areas such as water exploration and water rescue, especially in harsh aquatic environments, unknown underwater environments, and extreme weather conditions, humans often cannot complete their work and may even suffer health damage. Therefore, it is necessary to utilize advanced machines to explore complex environments under conditions where humans cannot work normally. Among these, unmanned surface vessels (USVs) have become a key research direction in the field of water exploration in recent years due to their high maneuverability, low risk, and model flexibility. During autonomous exploration, USVs must perform simultaneous localization and mapping (SLAM) in real time, select observation points based on this, and finally reach the target observation point through path planning. These three aspects are closely intertwined, making the process of autonomous environmental exploration by USVs extremely complex.

[0003] Currently, numerous excellent algorithms have emerged for various process modules of autonomous exploration, with breakthroughs focusing on the discovery and decision-making aspects of observation points. Algorithms such as Rapid Expanding Random Trees (RRT) and image edge detection are relatively mature and classic algorithms with high practical value. Deep learning, which has become very popular in the research community in recent years, has also been used in autonomous exploration, reducing exploration time and improving efficiency to some extent. However, its robustness in practical applications is not strong enough, and further research is needed.

[0004] Because unmanned ships autonomously explore on outdoor water surfaces, the uncertainties such as wind, waves, and currents are far greater than those on land-based machines like unmanned vehicles. This requires the relevant algorithms to be more robust and to be adjusted to accommodate the motion limitations of unmanned ships. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides an autonomous exploration method for unmanned surface vessels (USVs) that integrates Resonant Radiation Detection (RRT) and image edge detection. By fusing these two methods, this invention prioritizes selecting boundary points closer to the current direction of motion and pre-locating boundary points during the boundary point exploration phase. This results in a faster exploration speed and greater suitability for autonomous exploration of USV motion models.

[0006] The technical means employed in this invention are as follows:

[0007] A lightweight method for autonomous exploration of unknown regions by unmanned surface vessels (USVs) that integrates rapid extended random trees and image edge detection includes:

[0008] S1. Determine the bounded space to be explored. and local planning scope The upper half of the local map is defined using the geometric center of the unmanned vessel as the dividing line. and the lower half

[0009] S2, with the unmanned vessel currently at position P Boat Point A is the root node of the RRT. An extended tree is constructed in the upper half of the local map, and the nodes generated by crossing the boundary are selected as the local boundary points of the upper half of the map.

[0010] S3. In the lower half of the local map, the Canny edge detection algorithm is used to extract boundary information, and the center of each boundary is selected as the local boundary point of the lower half of the region.

[0011] S4. Use Mean-Shift clustering to integrate all local boundary points X. L The scores of local boundary points after integration and clustering are calculated using a custom evaluation function, and the local boundary point with the highest score is selected as the navigation target boundary point.

[0012] S5. Use Dijkstra's algorithm to plan the global path and Dubins' algorithm to achieve dynamic obstacle avoidance and trajectory smoothing in the local area;

[0013] S6. When the unmanned surface vessel travels to the local navigation target, dynamically prune the rapidly expanding random tree and include boundary points outside the local map into the global boundary point set X. G ;

[0014] S7. When the local boundary point set is empty, execute the effective backtracking strategy, and continuously repeat steps S3, S4, S5 and S6 until the exploration is completed.

[0015] Furthermore, the coverage area of ​​the local planning range L determined in step S1 is greater than the effective field of view (FOV) of the sensor. l ).

[0016] Further, step S2 specifically includes:

[0017] S21. Divide the bounded space V into three regions, including the known free region V. free Unknown region V unknown The occupied obstacle area V occu V is defined unknown =V free ∪V occu ;

[0018] S22, with the position of the unmanned vessel as the root node X of the RRT. initIn the overlapping area L of the upper half of the local map and the free area T ∩V free In the middle, generate random point X rand And find the nearest node X in the existing RRT tree nodes. nearest If the distance between two points is greater than the set step size λ, then the distance X on the straight line between the two points is taken. rand The point at a distance of λ is X. new If it is less than the set step size λ, take X. nearest As X new ;

[0019] S23, Determine X new With X nearest Are there any obstacles between them? If there are obstacles, discard X. new If no obstacle exists, return to step S22 to regenerate a random point. If no obstacle exists, then set X... new It is included in the RRT tree as the next node;

[0020] S24, If X new and X nearest If there is an unknown region, then X will be... new Add to the boundary point set.

[0021] Furthermore, step S3 also includes:

[0022] Local boundary points of the lower half region Local boundary points of the upper region Integrate into a local boundary point set X L ,

[0023] Furthermore, the evaluation function in step S4 is specifically as follows:

[0024]

[0025]

[0026]

[0027] The weight λ is used to make the gain T·I similar in magnitude to the navigation cost N; The information gain at the target point is represented by the value of the sensor and is expressed as the area covered. Indicates the current position P of the unmanned vessel. Boat To the target point The navigation cost, excluding feasible paths caused by obstacle obstruction, is represented by Euclidean distance; T gain λ is the steering gain coefficient, whose magnitude is related to the steering angle ψ.i This represents the steering weight coefficient, used for the order of magnitude of the balance calculation. λ1 > 10λ2 is selected to ensure the priority of small-angle selection.

[0028] Furthermore, in step S6, when the unmanned vessel is heading towards the local navigation target, the rapidly expanding random tree is dynamically pruned, specifically including:

[0029] Update the physical structure of the fast random tree by changing the root from A to B before the iteration;

[0030] Delete all branches and nodes that are obscured by obstacles or located outside the current upper half of the local area;

[0031] Connect the single node that has lost its parent node to the current root B, that is, make the root B the parent node.

[0032] Furthermore, the global boundary point set X in step S6 G It includes boundary points for image edge detection outside the current local map region. Boundary points explored by RRT It reduces the amount of data required for the algorithm.

[0033] Furthermore, the backtracking strategy in step S7 specifically includes:

[0034] The score of the global boundary point is calculated using an evaluation function, and the point with the highest score is taken as the next navigation target point of the unmanned vessel.

[0035] Delete the existing local RRT tree, and regenerate the local RRT tree after reaching the global target point.

[0036] Compared with the prior art, the present invention has the following advantages:

[0037] 1. The unmanned surface vessel autonomous exploration method fused with RRT and image edge detection provided by this invention is designed to leverage the advantages of unmanned surface vessels in autonomous exploration of unknown sea areas, such as flexibility, low cost, high efficiency, no need for human intervention, and strong safety. It is applied to the engineering application of a lightweight autonomous exploration method for unknown areas of unmanned surface vessels that integrates dynamic pruning of fast random trees and image edge detection for detailed exploration of local maps.

[0038] 2. The autonomous exploration method for unmanned vessels that integrates RRT and image edge detection provided by this invention applies dynamically pruned fast expanding random trees and the Canny edge detection algorithm to obtain exploration frontier points during the real-time localization and map building process of the unmanned vessel. This lightweight random tree method ensures the comprehensiveness of boundary point detection.

[0039] 3. The autonomous exploration method for unmanned vessels that integrates RRT and image edge detection provided by this invention, compared with previous exploration methods, addresses the inherent motion characteristics of unmanned vessels, such as the inability to rotate during navigation and the difficulty of making large-angle turns in narrow sea areas. It proposes an exploration target point evaluation function with turning constraints and a path planning algorithm that integrates Dijkstra and Dubins, thereby reducing the occurrence of inefficient exploration due to motion defects.

[0040] Based on the above reasons, this invention can be widely applied in fields such as autonomous exploration and decision-making by unmanned vessels. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is a flowchart of the method of the present invention.

[0043] Figure 2 This is a schematic diagram illustrating the optimization of the exploration strategy of the present invention. Detailed Implementation

[0044] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0045] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the present invention or its application or use. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0047] Unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of the invention. It should also be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale. Techniques, methods, and devices known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and devices should be considered part of the specification. In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values. It should be noted that similar reference numerals and letters in the following figures denote similar items; therefore, once an item is defined in one figure, it need not be further discussed in subsequent figures.

[0048] In the description of this invention, it should be understood that the orientation or positional relationship indicated by directional terms such as "front, back, up, down, left, right", "horizontal, vertical, horizontal" and "top, bottom" is generally based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing this invention and simplifying the description. Unless otherwise stated, these directional terms do not indicate or imply that the device or element referred to must have a specific orientation or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on the scope of protection of this invention. The directional terms "inner" and "outer" refer to the inner and outer contours relative to the outline of each component itself.

[0049] For ease of description, spatial relative terms such as "above," "over," "on the upper surface of," "above," etc., are used herein to describe the spatial positional relationship of a device or feature as shown in the figures to other devices or features. It should be understood that spatial relative terms are intended to encompass different orientations in use or operation besides the orientation of the device as described in the figures. For example, if the device in the figures is inverted, a device described as "above" or "above" other devices or structures would subsequently be positioned as "below" or "under" other devices or structures. Thus, the exemplary term "above" can include both "above" and "below." The device may also be positioned in other different ways (rotated 90 degrees or in other orientations), and the spatial relative descriptions used herein will be interpreted accordingly.

[0050] Furthermore, it should be noted that the use of terms such as "first" and "second" to define components is merely for the purpose of distinguishing the corresponding components. Unless otherwise stated, the above terms have no special meaning and therefore should not be construed as limiting the scope of protection of this invention.

[0051] like Figure 1As shown, this invention provides an autonomous exploration method for unmanned surface vessels that fuses RRT and image edge detection, including:

[0052] S1. Determine the bounded space to be explored. and local planning scope The upper half of the local map is defined using the geometric center of the unmanned vessel as the dividing line. and the lower half

[0053] S2, with the unmanned vessel currently at position P Boat Point A is the root node of the RRT. An extended tree is constructed in the upper half of the local map, and the nodes generated by crossing the boundary are selected as the local boundary points of the upper half of the map.

[0054] S3. In the lower half of the local map, the Canny edge detection algorithm is used to extract boundary information, and the center of each boundary is selected as the local boundary point of the lower half of the region.

[0055] S4. Use Mean-Shift clustering to integrate all local boundary points X. L The scores of local boundary points after integration and clustering are calculated using a custom evaluation function, and the local boundary point with the highest score is selected as the navigation target boundary point.

[0056] S5. Use Dijkstra's algorithm to plan the global path and Dubins' algorithm to achieve dynamic obstacle avoidance and trajectory smoothing in the local area;

[0057] S6. When the unmanned surface vessel travels to the local navigation target, dynamically prune the rapidly expanding random tree and include boundary points outside the local map into the global boundary point set X. G ;

[0058] S7. When the local boundary point set is empty, execute the effective backtracking strategy, and continuously repeat steps S3, S4, S5 and S6 until the exploration is completed.

[0059] In this embodiment, the local area range L is first determined, and then the position P of the unmanned vessel is used as the reference. Boat Point A is used as the root node of a fast random tree to construct the tree. Then, the fast random tree is used to explore local boundary points in the upper half of the local region, and its set is used... This indicates that the Canny edge detection algorithm is used to extract the boundary in the lower half of the region, and the center point of the boundary is selected as the boundary point. The set of these points is then used... This indicates that the two sets mentioned above are merged into a single set of local boundary points. Then, Mean-Shift is used to cluster and integrate all local boundary points. If local boundary points... Using the evaluation function designed in this paper Calculate the scores of local boundary points after clustering, and select the point with the highest score as the exploration target point. Otherwise, if local boundary points are empty, determine whether a global boundary point X exists. G If a global boundary point exists, then the evaluation function is used. Calculate the scores of global boundary points and select the point with the highest score as the exploration target point. After determining the target point, use Dijkstra's algorithm to plan the global path and DWA to plan the local path. During the movement of the unmanned vessel, continuously update the physical structure of the fast random tree, transforming the root from A to B before the iteration, deleting all branches and nodes obscured by obstacles or outside the current local upper half, and adding boundary points outside the local range to the global boundary point set.

[0060] In a specific implementation, as a preferred embodiment of the present invention, the coverage area of ​​the local planning range L determined in step S1 is greater than the effective field of view (FOV) of the sensor. l ).

[0061] In a specific implementation, as a preferred embodiment of the present invention, step S2 specifically includes:

[0062] S21. Divide the bounded space V into three regions, including the known free region V. free Unknown region V unknown The occupied obstacle area V occu V is defined unknown =V free ∪V occu ;

[0063] S22, with the position of the unmanned vessel as the root node X of the RRT. init In the overlapping area L of the upper half of the local map and the free area T ∩V free In the middle, generate random point X rand And find the nearest node X in the existing RRT tree nodes. nearest If the distance between two points is greater than the set step size λ, then the distance X on the straight line between the two points is taken. rand The point at a distance of λ is X. new If it is less than the set step size λ, take X. nearest As X new ;

[0064] S23, Determine X new With X nearest Are there any obstacles between them? If there are obstacles, discard X. newIf no obstacle exists, return to step S22 to regenerate a random point. If no obstacle exists, then set X... new It is included in the RRT tree as the next node;

[0065] S24, If X new and X nearest If there is an unknown region, then X will be... new Add to the boundary point set.

[0066] In a specific implementation, as a preferred embodiment of the present invention, step S3 further includes:

[0067] Local boundary points of the lower half region Local boundary points of the upper region Integrate into a local boundary point set X L ,

[0068] In a specific implementation, as a preferred embodiment of the present invention, the evaluation function in step S4 is specifically as follows:

[0069]

[0070]

[0071]

[0072] The weight λ is used to make the gain T·I similar in magnitude to the navigation cost N; The information gain at the target point is represented by the value of the sensor and is expressed as the area covered. Indicates the current position P of the unmanned vessel. Boat To the target point The navigation cost, excluding feasible paths caused by obstacle obstruction, is represented by Euclidean distance; T gain λ is the steering gain coefficient, whose magnitude is related to the steering angle ψ. i The value represents the turning weight coefficient, used for the order of magnitude in the balancing calculation. λ1 > 10λ2 is chosen to ensure priority for small-angle selection. This evaluation function addresses the inherent constraints of unmanned surface vessel (USV) motion, aiming to avoid large turns and improve exploration efficiency and map quality while ensuring USV navigation safety.

[0073] In a specific implementation, as a preferred embodiment of the present invention, step S6, which involves dynamically pruning the rapidly expanding random tree when the unmanned vessel heads towards the local navigation target, specifically includes:

[0074] Update the physical structure of the fast random tree by changing the root from A to B before the iteration;

[0075] Delete all branches and nodes that are obscured by obstacles or located outside the current upper half of the local area;

[0076] Connect the single node that has lost its parent node to the current root B, that is, make the root B the parent node.

[0077] In this embodiment, Figure 2 This demonstrates how the invention prunes the RRT tree and handles unexplored valid boundary points as the unmanned vessel moves forward during the exploration phase. Figure (a) shows the exploration phase, displaying the coverage of the local layer L at the current moment and the observation point X obtained through the fusion exploration method. L In terms of graphical representation, we follow the traditional method, with the white area representing the free region V. free The black area represents the area with obstacles. occu The gray area represents the unknown region V. unknown Exploration will be conducted in the expansion of V free The target is represented by gray dots, which indicate two local boundaries, and the unmanned vessel will use [the following]. As navigation target points during the exploration phase. Figure (b) shows the pruning phase, with the random tree updated in one iteration and the local observation point X. L In L B The dark gray dots and arrows inside represent pruned, random tree nodes and branches. T The light gray dots inside represent empty nodes that have lost their parent nodes due to pruning. These empty nodes are connected to the current root node B. The gray dot in the lower right corner indicates the node after one iteration from X. L Transform into X G The observation points will be explored during the effective backtracking phase.

[0078] In a specific implementation, as a preferred embodiment of the present invention, the global boundary point set X in step S6... G It includes boundary points for image edge detection outside the current local map region. Boundary points explored by RRT It reduces the amount of data required for the algorithm.

[0079] In a specific implementation, as a preferred embodiment of the present invention, the backtracking strategy in step S7 specifically includes:

[0080] The score of the global boundary point is calculated using an evaluation function, and the point with the highest score is taken as the next navigation target point for the unmanned vessel. The existing local RRT tree is deleted, and the local RRT tree is regrown after reaching the global target point.

[0081] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for autonomous exploration of unmanned surface vessels by fusing RRT and image edge detection, characterized in that, include: S1. Determine the bounded space to be explored. and local planning scope The upper half of the local map is defined by using the geometric center of the unmanned vessel as the dividing line. and the lower half ; S2, based on the current position of the unmanned vessel Point A is the root node of the RRT. An extended tree is constructed in the upper half of the local map, and the nodes generated by crossing the boundary are selected as the local boundary points of the upper half of the map. ; S3. In the lower half of the local map, the Canny edge detection algorithm is used to extract boundary information, and the center of each boundary is selected as the local boundary point of the lower half of the region. ; S4. Use Mean-Shift clustering to integrate all local boundary points. The scores of local boundary points after integration and clustering are calculated using a custom evaluation function, and the local boundary point with the highest score is selected as the navigation target boundary point. S5. Use Dijkstra's algorithm to plan the global path and Dubins' algorithm to achieve dynamic obstacle avoidance and trajectory smoothing in the local area; S6. When the unmanned vessel travels to the local navigation target, dynamically prune the rapidly expanding random tree and include boundary points outside the local map into the global boundary point set. ; S7. When the local boundary point set is empty, execute the effective backtracking strategy, and continuously repeat steps S3, S4, S5 and S6 until the exploration is completed.

2. The unmanned surface vessel autonomous exploration method based on the fusion of RRT and image edge detection according to claim 1, characterized in that, The local planning scope determined in step S1 The coverage area is greater than the sensor's effective field of view. .

3. The unmanned surface vessel autonomous exploration method based on the fusion of RRT and image edge detection according to claim 1, characterized in that, Step S2 specifically includes: S21, Bounded Space Divided into three regions: known free regions Unknown area Occupied obstacle area , where defined ; S22, Using the position of the unmanned vessel as the root node of the RRT. In the overlapping area of ​​the upper half of the local map and the free area In the process of generating random points And find the nearest node in the existing RRT tree nodes. If the distance between two points is greater than the set step size Then take the distance on the straight line between the two points. Distance is The point is If it is less than the set step size ,Pick As ; S23, Judgment and Are there any obstacles in between? If there are obstacles, discard the item. If no obstacle exists, then return to step S22 to regenerate random points. It is included in the RRT tree as the next node; S24, if and If there are unknown areas, then... Add to the boundary point set.

4. The unmanned surface vessel autonomous exploration method based on the fusion of RRT and image edge detection according to claim 1, characterized in that, Step S3 also includes: Local boundary points of the lower half region Local boundary points of the upper region Integrate into a local boundary point set , .

5. The unmanned surface vessel autonomous exploration method based on the fusion of RRT and image edge detection according to claim 1, characterized in that, The evaluation function in step S4 is specifically as follows: Among them, weight Used to increase gain Navigation costs They have similar orders of magnitude; The information gain at the target point is represented by the value of the sensor and is expressed as the area covered. Indicates the current location of the unmanned vessel. To the target point The navigation cost, which does not consider feasible paths caused by obstacle obstruction, is represented by Euclidean distance. This is the steering gain coefficient, the magnitude of which is related to the steering angle. Related, This represents the steering weight coefficient, used for the order of magnitude in balance calculations, and is selected accordingly. This ensures that the selection of small angles is prioritized.

6. The unmanned surface vessel autonomous exploration method based on the fusion of RRT and image edge detection according to claim 1, characterized in that, In step S6, when the unmanned vessel is heading towards the local navigation target, the rapidly expanding random tree is dynamically pruned, specifically including: Update the physical structure of the fast random tree by changing the root from A to B before the iteration; Delete all branches and nodes that are obscured by obstacles or located outside the current upper half of the local area; Connect the single node that has lost its parent node to the current root B, that is, make the root B the parent node.

7. The unmanned surface vessel autonomous exploration method based on the fusion of RRT and image edge detection according to claim 1, characterized in that, The global boundary point set in step S6 It includes boundary points for image edge detection outside the current local map region. Boundary points explored by RRT .

8. The autonomous exploration method for unmanned vessels fused with RRT and image edge detection according to claim 1, characterized in that, The backtracking strategy in step S7 specifically includes: The score of the global boundary point is calculated using an evaluation function, and the point with the highest score is taken as the next navigation target point of the unmanned vessel. Delete the existing local RRT tree, and regenerate the local RRT tree after reaching the global target point.

Images