CN116625386A

CN116625386A - Map construction method and multi-robot path planning method based on image filtering

Info

Publication number: CN116625386A
Application number: CN202310430842.XA
Authority: CN
Inventors: 赵威威; 林睿
Original assignee: Suzhou University
Current assignee: Suzhou University
Priority date: 2023-04-21
Filing date: 2023-04-21
Publication date: 2023-08-22

Abstract

The application relates to the field of intelligent control of mobile robots, and discloses a map construction method and a multi-robot path planning method based on image filtering, which comprise the following steps: constructing an initial global grid map by using an instant positioning and map construction technology, preprocessing, generating global topological nodes on the preprocessed global grid map by using image filtering, and carrying out sparsification processing on the global topological nodes; establishing a topological relation of the global topological nodes after the sparsification processing by comparing the position and state information of the nodes, and generating a final environment topological map; setting a starting point set and a target point set of the multiple robots, and carrying out global path planning of the multiple robots by combining the traffic cost values on the basis of an A-algorithm. The application can realize discrete topology of the grid map, meet the autonomous navigation requirement when the complex dynamic scene operation of the multi-mobile robot is performed, and relieve the problems of global path superposition and conflict under the multi-robot.

Description

Map construction method and multi-robot path planning method based on image filtering

Technical Field

The application relates to the field of intelligent control of mobile robots, in particular to a map construction method and a multi-robot path planning method based on image filtering.

Background

With the rapid development of social economy and science and technology, more and more mobile robots are moving into the field of vision of people. From various house sweeping robots to various service and meal delivery robots in service places, and to logistics robots in logistics and industrial fields. It can be said that mobile robots with various forms and functions have more and more contributions in improving the living standard of people, improving the production efficiency and increasing the social benefit. The application and popularization level of the intelligent mobile robot can be one of important expression forms of the social science and technology level, and it is expected that the industry of the mobile robot direction will continue to develop rapidly in the near future.

Each robot can be regarded as an independent intelligent body, and besides personalized functions of sweeping, guiding, distributing and the like, each robot has basic functions of positioning and path planning, and can independently and independently realize positioning and navigation services. However, more and more robots are required to face large-scale and multi-task working requirements, so as to meet the working requirements, multiple robots are introduced into the same working environment, and each independent robot can cause problems of path conflict, blockage and collision among the robots in some special environments (such as long and narrow channels, intersections and the like) due to lack of integral coordinated control and communication.

The mobile robot is used for various industries of social development, such as factories, hospitals, families, hotels, exhibition halls, restaurants and the like, and when the mobile robot performs work tasks such as logistics, transportation, distribution and the like, the mobile robot needs to navigate autonomously in a work scene, so that the mobile robot can accurately position and safely travel. The operation scene is complex, such as a plurality of dynamic objects, a driving area is narrow, and the like, and the movement of the mobile robot needs to follow the traffic rules similar to motor vehicles, such as a forbidden area, a right-hand area, a forbidden area, forbidden lines, single lines, and the like, so that the orderly passage of the mobile robot in the complex operation scene can be ensured, and the problems of traffic jam and even congestion possibly occurring between the mobile robot and the dynamic objects and between the mobile robots are solved. In order to solve the problem, the mobile robot generally builds a global map of surrounding working scenes in advance in the working scenes, and safe traveling from a starting point to a target point is realized according to the requirements of the working tasks. However, the global path planning algorithm of the common mobile robot only searches for the nearest or the shortest short-cut path, but does not follow the traffic rules of the travelable path, which results in poor smoothness of traveling in scenes such as a relatively large number of dynamic objects and a relatively narrow travelable area.

In order to ensure the traffic efficiency and safety of the mobile robot, a certain traffic rule setting needs to be performed on a pre-constructed scene map of the mobile robot, and the traffic rule is similar to the constraint that the traffic rule of a human motor vehicle identifies which areas can walk, which areas cannot walk, which areas can only walk unidirectionally and the like. The mobile robot is based on the traffic rule constraint conditions, and improves an A-based global path searching algorithm by defining traffic values for the global grid map, so that global path planning based on the traffic rule constraint can be realized. At the same time, grid maps remain the primary choice for autonomous navigation of a single robot within a certain range. However, in recent years, autonomous mobile robots have been widely used in mobile environments in large man-machine interaction, which means that the working scenario of mobile robots has been changed from single robot actions to cooperation of multiple robots in a large environment, placing higher demands on global path planning efficiency of robots and collaborative scheduling between robots. However, the prior art does not describe an environmental method, and it is not possible to solve path conflicts of multiple robots while improving the efficiency of the global path plane.

Disclosure of Invention

Therefore, the technical problem to be solved by the application is to overcome the defects in the prior art, and provide a map construction method and a multi-robot path planning method based on image filtering, which can realize discrete topology of a grid map, meet the autonomous navigation requirement when the multi-mobile robot works in a complex dynamic scene, and alleviate the problems of global path superposition and conflict under the multi-robot.

In order to solve the technical problems, the application provides a map construction method based on image filtering, which comprises the following steps:

constructing an initial global grid map by using an instant positioning and map construction technology, preprocessing, generating global topological nodes on the preprocessed global grid map by using image filtering, and carrying out sparsification processing on the global topological nodes;

and establishing a topological relation of the global topological nodes after the sparsification processing by comparing the position and state information of the nodes, and generating a final environment topological map.

In one embodiment of the application, the preprocessing includes binarizing the initial global grid map and edge-glitch optimization.

In one embodiment of the application, the edge spike optimization process is using median filtering.

In one embodiment of the present application, the image filtering is used to generate global topology nodes on the preprocessed global grid map, specifically:

acquiring the size of a robot moving in a scene, and generating feasible points in a grid map according to the size of the robot;

and generating global topological nodes by taking the feasible points as centers and taking the size of the robot as a radius.

In one embodiment of the present application, the feasible points are generated in the grid map according to the size of the robot, specifically:

the filter template tau is constructed according to the size of the robot:

wherein ψ is the size of the robot, res is the resolution of the grid map, and N is a natural integer;

calculating a feasible point index gamma (x, y) according to the filtering template tau as follows:

wherein a= (tau-1)/2,K (i, j) is a filtering template element, f (x+i, y+j) is an original image element, i, j are relative coordinates of the element in two directions of a plane in a filtering calculation process;

when the feasible point index Γ (x, y) is smaller than the preset threshold, all points are taken as the feasible points.

In one embodiment of the present application, the thinning process is performed on the global topology node, specifically:

step 1: acquiring central nodes of all global topological nodes as a node list L1;

step 2: calculating the horizontal coordinate difference value of the No. 1 node and the No. 2 node in the node list L1, and when the horizontal coordinate difference value is smaller than a preset threshold dist _min Deleting the node No. 2; reading the next node in L1 as a node No. 2, recalculating the difference value of the horizontal coordinates of the node No. 1 and the node No. 2, and when the difference value of the horizontal coordinates is larger than a preset threshold dist _min When the node 2 is assigned as the node 1, reading the next node in the L1 as a new node 2;

step 3: when the row coordinates of the No. 1 node and the No. 2 node change, the No. 2 node is assigned as the No. 1 node, and the next node in the L1 is read as a new No. 2 node;

step 4: and (3) repeatedly executing the step (2-3) until all the nodes in the L1 are traversed, and taking the global topological node corresponding to the node sequence obtained at the moment as the global topological node after the sparsification processing.

In one embodiment of the present application, the establishing a topological relation of the global topological node after the sparsification processing by comparing the position and the state information of the node specifically includes:

calculating the distance between every two global topological nodes after the sparsification treatment, and considering that the edge building condition is met when the distance between the two global topological nodes is smaller than or equal to the minimum reachable distance H; and establishing a topological relation between every two global topological nodes under the condition of side establishment, and generating a final environment topological map.

The application also provides a multi-robot path planning method, which comprises the following steps:

surrounding environment information in a dynamic operation scene of the multiple robots is perceived in real time through the laser sensor, and an environment topological map is constructed by using the map construction method based on image filtering;

setting a starting point set and a target point set of the multiple robots, and carrying out global path planning of the multiple robots by combining traffic cost values on the basis of an A-algorithm to obtain a global path node sequence of each robot.

In one embodiment of the present application, the global path planning of multiple robots is performed by combining traffic cost values based on an algorithm a, specifically:

when the algorithm A is used for carrying out global path planning of multiple robots, the calculation formula of the cost function of each point is as follows:

f(p)＝g(p)+h(p)+w(p)，

wherein f (p) is the cost function of the p-th point, and p represents the next hop node in path planning; g (P) represents a cost value from the starting point to the P node, and h (P) represents a heuristic value from the P node to the target point; w (p) represents the node traffic value from the current node to the p node.

The application also provides a map construction system based on image filtering, which comprises:

the initial global grid map construction module is used for constructing an initial global grid map by using an instant positioning and map construction technology and preprocessing the initial global grid map;

the global topological node processing module is used for generating global topological nodes on the preprocessed global grid map by using image filtering and carrying out sparsification processing on the global topological nodes;

the environment topology map construction module is used for establishing the topological relation of the global topology nodes after the sparsification processing by comparing the position and state information of the nodes, and generating a final environment topology map.

Compared with the prior art, the technical scheme of the application has the following advantages:

according to the application, the construction of the environment grid map is realized through SLAM technology, discrete topology of the grid map is realized through image filtering on the basis, intelligent identification and screening are carried out on feasible areas in the grid map based on the size information of the robot, free topology nodes are generated, and the environment topology map is constructed. Under the environment topological map, the autonomous navigation requirement of the multi-mobile robot in complex dynamic scene operation can be met, the safety and the order are good, the real-time performance is high, the problems of global path superposition and collision under the multi-robot are relieved, and the method can be popularized and applied to path planning of the multi-mobile robot in different complex dynamic scenes.

Drawings

In order that the application may be more readily understood, a more particular description of the application will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings, in which:

fig. 1 is a flowchart of a map construction method based on image filtering in the present application.

Fig. 2 is a specific step diagram of a map construction method based on image filtering in the present application.

Fig. 3 is a flow chart for establishing a topology in the present application.

Fig. 4 is a specific step diagram of the multi-robot path planning method in the present application.

Fig. 5 is a diagram of the result of global path planning using a conventional a algorithm in an embodiment of the present application.

Fig. 6 is a diagram of the result of global path planning using the multi-robot path planning method of the present application in an embodiment of the present application.

Fig. 7 is a graph of the results of comparing the global path with the conventional a-algorithm and the multi-robot path planning method in terms of the path overlap point in an embodiment of the present application.

Detailed Description

The present application will be further described with reference to the accompanying drawings and specific examples, which are not intended to be limiting, so that those skilled in the art will better understand the application and practice it.

Example 1

Referring to fig. 1 and 2, the application discloses a map construction method based on image filtering, which comprises the following steps:

s1: an initial global grid map is constructed and pre-processed using an instant localization and mapping (Simultaneous Localization and Mapping, SLAM) technique.

The preprocessing comprises the steps of carrying out binarization processing and edge burr optimization processing on the initial global grid map. The edge burr optimization process is to use median filtering.

In order to eliminate tiny obstacle grid, image noise and burrs caused by sensor noise and robot positioning errors in the map construction process. And (3) optimizing the image through binarization and edge burr optimization, and eliminating the influence of image noise on image filtering calculation, wherein the binarization makes the edge characteristics more prominent. In the edge burr optimization process, edges are flattened by median filtering.

S2: and generating global topological nodes on the preprocessed global grid map by using image filtering.

S2-1: the size of a robot moving in a scene is obtained.

S2-2: feasible points are generated in the grid map according to the size of the robot.

S2-2-1: the filter template tau is constructed according to the size of the robot:

s2-2-2: calculating a feasible point index gamma (x, y) according to the filtering template tau as follows:

wherein a= (tau-1)/2,K (i, j) is a filtering template element, f (x+i, y+j) is an original image element, i, j are relative coordinates of the element in two directions of a plane in a filtering calculation process, namely, relative coordinates on an x axis and a y axis;

s2-2-3: when all points with the feasible point index Γ (x, y) smaller than the preset threshold are used as the feasible points, the preset threshold is set to be 0 for the map image which has undergone binarization processing.

S2-3: and generating global topological nodes by taking the feasible points as centers and taking the size of the robot as a radius.

S3: and carrying out sparsification treatment on the global topological node. Because of the template matching based pixel processing, the global topology nodes generated by node filtering are rich. In the later multi-robot running process, the latest passing distance among robots is also required to be considered so as to achieve smooth passing among the robots. Therefore, the application is based on the latest passing distance dist _min The sparse processing of the topological nodes is carried out so as to meet the passing requirements among robots.

S3-1: and acquiring the central nodes of all the global topological nodes as a node list L1, namely a feasible point node list L1.

S3-2: calculating the horizontal coordinate difference value of the No. 1 node and the No. 2 node in the node list L1When the difference value of the horizontal coordinates is smaller than a preset threshold dist _min Node number 2 is deleted, dist in this embodiment _min The value is 15 pixel units; reading the next node in L1 as a node No. 2, recalculating the difference value of the horizontal coordinates of the node No. 1 and the node No. 2, and when the difference value of the horizontal coordinates is larger than a preset threshold dist _min When the node 2 is assigned as the node 1, reading the next node in the L1 as a new node 2;

s3-3: when the row coordinates of nodes 1 and 2 change, i.e.And when the node 2 is assigned as the node 1, reading the next node in the L1 as a new node 2.

S3-4: repeatedly executing S3-2-S3-3 until all nodes in the L1 are traversed, and taking the global topological node corresponding to the node sequence obtained at the moment as the global topological node after the sparsification processing.

S4: the topological relation between topological nodes represents the passing relation and passing cost between nodes, namely the reachable relation and distance between two areas in the environment. As shown in fig. 3, the position and state information of the nodes are compared to establish the topological relation of the global topological nodes after the sparsification processing, and a final environment topological map is generated.

Every two global topological nodes (X _N1 ,Y _N1 ) And (X) _N2 ,Y _N2 ) Distance betweenWherein (X) _N1 ,Y _N1 ) And (X) _N2 ,Y _N2 ) The coordinates of the node 1 and the node 2 are respectively; when the distance between two global topological nodes is smaller than or equal to the minimum reachable distance H, the edge building condition is considered to be met, and in the embodiment, the value of H is 20 pixel units; and establishing a topological relation between every two global topological nodes under the condition of side establishment, and generating a final environment topological map. Topological relationships include one-way, two-way traffic relationships.

Example two

As shown in fig. 4, the present application further discloses a multi-robot path planning method, where, under the condition that starting points of tasks of each robot are known, global path point planning is performed by using an improved a-x algorithm based on an environment topology map, including:

s1: surrounding environment information in a dynamic operation scene of the multiple robots is perceived in real time through laser sensors, and an environment topological map is constructed by using the map construction method based on image filtering described in the first embodiment.

S2: setting a starting point set and a target point set of the multi-robot.

S3: and carrying out global path planning of multiple robots by combining the traffic cost values on the basis of the A-algorithm to obtain a global path node sequence of each robot.

f(p)＝g(p)+h(p)+w(p)，

wherein f (p) is the cost function of the p-th point, and p represents the next hop node in path planning; g (P) represents a cost value from the starting point to the P node, and h (P) represents a heuristic value from the P node to the target point; w (p) represents the node traffic value from the current node to the p node. The application adds the passing cost value w (p) of each node on the basis of the traditional A-algorithm, the value is equal to the weight of the edges between the nodes, the occupation condition of the node can be measured, for example, when the node is occupied, the edge entering weight of the node can be properly improved, so that the change point is selected again when the later path planning is avoided. The node is planned on the global path through updating the node in-edge and out-edge weights, so that the problems of path coincidence and conflict of multiple robots are relieved. The dynamic a-multipath planning algorithm flow.

Example III

The application also discloses a map construction system based on image filtering, which comprises an initial global grid map construction module, a global topological node processing module and an environment topological map construction module.

And the initial global grid map construction module is used for constructing an initial global grid map by using an instant positioning and map construction technology and preprocessing the initial global grid map. And the global topological node processing module is used for generating global topological nodes on the preprocessed global grid map by using image filtering and carrying out sparsification processing on the global topological nodes. The environment topology map construction module is used for establishing the topological relation of the global topology nodes after the sparsification processing by comparing the position and state information of the nodes, and generating a final environment topology map.

Compared with the prior art, the application has the advantages that:

1. according to the application, the construction of the environment grid map is realized through SLAM technology, discrete topology of the grid map is realized through image filtering on the basis, intelligent identification and screening are carried out on feasible areas in the grid map based on the size information of the robot, free topology nodes are generated, and the environment topology map is constructed. Under the environment topological map, the autonomous navigation requirement of the multi-mobile robot in complex dynamic scene operation can be met, the safety and the order are good, the real-time performance is high, and the method can be popularized and applied to the path planning problem of the multi-mobile robot in different complex dynamic scenes.

2. The constructed environment topological map with high discretization can provide a foundation for autonomous navigation work of the multi-mobile robot in the same space during complex dynamic scene operation, and improves the efficiency of global path planning.

3. On the basis of the constructed environment topological map, the global path planning of the multiple robots is carried out by combining the A-type algorithm and the current cost value, the automatic planning of a plurality of global paths is carried out by inputting the starting point and the target point set, the entering side weight of the occupied node is dynamically updated in the planning process, and the problems of global path superposition and collision under the multiple robots can be relieved.

In order to further illustrate the beneficial effects of the present application, in the present embodiment, in the same global topological map, a conventional a-algorithm and a modified a-algorithm multi-robot path planning method (hereinafter referred to as ModifiedA algorithm) in the present application are used to perform multi-robot global path planning, and the results are shown in fig. 5 and 6. Global paths are planned for 5 robots in two ways, respectively, wherein the coincidence node between the paths can be defined as a worthless coincidence point of the paths, as shown by the dashed line parts in fig. 5 and 6. The number of the path non-value overlapping points reflects the overlapping condition among the global path planning to a certain extent, and the effect of the path non-value overlapping points on the multi-robot global path planning can be well measured. As can be seen from a comparison of fig. 5 and fig. 6, the ModifiedA algorithm updates the ingress weights of the already occupied nodes with redundant topology nodes, so that the paths planned thereafter avoid generating coincident nodes as much as possible. As can be seen from a comparison of fig. 5 and 6, the number of coincident nodes between path D and path E is significantly reduced by eliminating the coincident nodes of path a and path B.

As shown in fig. 7, comparing the conventional a-algorithm with the modified a-algorithm of the present application on the same topological map, it can be found that as the number of robots in the environment increases, the number of times that the global paths overlap with each other is significantly increased, because the number of topological nodes in the environment is constant, and when the number of robots increases, the global paths also increase, so that under the condition that the number of integral nodes is unchanged, the number of idle nodes gradually decreases, and thus, overlap nodes are inevitably generated between the paths. However, as can be seen by comparing fig. 7, although the worthless coincident points between paths cannot be completely eliminated, the present application has fewer path coincident points than the conventional a-algorithm, and can alleviate the problem of path collision of the later robot.

It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

It is apparent that the above examples are given by way of illustration only and are not limiting of the embodiments. Other variations and modifications of the present application will be apparent to those of ordinary skill in the art in light of the foregoing description. It is not necessary here nor is it exhaustive of all embodiments. And obvious variations or modifications thereof are contemplated as falling within the scope of the present application.

Claims

1. The map construction method based on image filtering is characterized by comprising the following steps of:

2. The image filtering-based map construction method according to claim 1, wherein: the preprocessing comprises binarization processing and edge burr optimization processing on the initial global grid map.

3. The image filtering-based map construction method according to claim 2, wherein: the edge burr optimization process is to use median filtering.

4. The image filtering-based map construction method according to claim 1, wherein: the global topological node is generated on the preprocessed global grid map by using image filtering, and specifically comprises the following steps:

5. The image filtering-based map construction method according to claim 4, wherein: the feasible points are generated in the grid map according to the size of the robot, specifically:

the filter template tau is constructed according to the size of the robot:

6. The image filtering-based map construction method according to claim 1, wherein: the global topological node is subjected to sparsification treatment, specifically:

7. The image filtering-based map construction method according to any one of claims 1 to 6, wherein: the method comprises the steps of establishing a topological relation of the global topological node after sparsification processing by comparing the position and state information of the nodes, and specifically comprises the following steps:

8. A multi-robot path planning method, comprising:

sensing surrounding environment information in a dynamic operation scene of the multiple robots in real time through laser sensors, and constructing an environment topological map by using the image filtering-based map construction method according to any one of claims 1 to 7;

9. The multi-robot path planning method of claim 8, wherein: the method for planning the global path of the multiple robots by combining the traffic cost values on the basis of the A-algorithm specifically comprises the following steps:

f(p)＝g(p)+h(p)+w(p)，

10. A map construction system based on image filtering, comprising: