CN110502006B - Full-coverage path planning method for mobile robot in abandoned land of mining area - Google Patents

Full-coverage path planning method for mobile robot in abandoned land of mining area Download PDF

Info

Publication number
CN110502006B
CN110502006B CN201910661538.XA CN201910661538A CN110502006B CN 110502006 B CN110502006 B CN 110502006B CN 201910661538 A CN201910661538 A CN 201910661538A CN 110502006 B CN110502006 B CN 110502006B
Authority
CN
China
Prior art keywords
path
sub
vertex
map
neuron
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201910661538.XA
Other languages
Chinese (zh)
Other versions
CN110502006A (en
Inventor
周林娜
汪芸
张鑫
刘金浩
沈乐萍
陈黎明
赵建国
王众
杨春雨
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China University of Mining and Technology CUMT
Original Assignee
China University of Mining and Technology CUMT
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by China University of Mining and Technology CUMT filed Critical China University of Mining and Technology CUMT
Priority to CN201910661538.XA priority Critical patent/CN110502006B/en
Publication of CN110502006A publication Critical patent/CN110502006A/en
Application granted granted Critical
Publication of CN110502006B publication Critical patent/CN110502006B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/60—Intended control result
    • G05D1/648—Performing a task within a working area or space, e.g. cleaning
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/02—Control of position or course in two dimensions
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/20—Control system inputs
    • G05D1/24—Arrangements for determining position or orientation
    • G05D1/246—Arrangements for determining position or orientation using environment maps, e.g. simultaneous localisation and mapping [SLAM]
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/40—Control within particular dimensions
    • G05D1/43—Control of position or course in two dimensions
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D2105/00—Specific applications of the controlled vehicles
    • G05D2105/80—Specific applications of the controlled vehicles for information gathering, e.g. for academic research
    • G05D2105/87—Specific applications of the controlled vehicles for information gathering, e.g. for academic research for exploration, e.g. mapping of an area
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D2107/00—Specific environments of the controlled vehicles
    • G05D2107/70—Industrial sites, e.g. warehouses or factories
    • G05D2107/73—Mining

Landscapes

  • Engineering & Computer Science (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Feedback Control In General (AREA)
  • Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)

Abstract

The invention discloses a method for planning a full-coverage path of a mobile robot in a mining area waste land, which comprises the steps of obtaining a satellite map of the mining area ground in a remote sensing satellite system, wherein the satellite map is used as an original map of the mining area ground; adopting a cattle-ploughing decomposition method to carry out regional decomposition on an original map, and decomposing the original map into a plurality of sub-regions without obstacles; planning the traversal sequence of each subregion by adopting a depth-first search algorithm; adopting a biological excitation neural network algorithm to complete internal traversal of each sub-area and path transfer between the sub-areas, thereby obtaining complete coverage of the whole environment; design path cost function, and A*The algorithm carries out a comparison experiment, and the performance of the biostimulation neural network algorithm in the path transfer between the sub-regions is evaluated from the aspects of time cost and path cost. The method has good environment adaptability and obstacle avoidance capability, can effectively reduce the difficulty of realizing the full-coverage path planning, and realizes the full-coverage path planning of the abandoned land in the mining area.

Description

Full-coverage path planning method for mobile robot in abandoned land of mining area
Technical Field
The invention relates to a full-coverage path planning method, in particular to a full-coverage path planning method for a mobile robot in a abandoned land of a mining area.
Background
Land resources are carriers for human survival and development and are basic material conditions for agricultural production and industrial construction, but the population of China is large, and the human mole is the most important contradiction faced by China. A number of policies and regulations indicate that: the land reclamation of the abandoned land of the mining area is an important way for relieving the agricultural land and improving the ecological environment of the mining area. The environment of the abandoned land of the mining area has complexity, so that great difficulty exists in the land reclamation technology. The mobile robot has the capabilities of environmental perception, behavior decision, motion control and the like, and is widely applied to the fields of intelligent cleaning, farmland operation, military detection and the like. Therefore, the mobile robot full-coverage path planning method has practical significance in researching the mobile robot full-coverage path planning of the abandoned land of the mining area with the aim of improving the accuracy and the real-time performance of soil reclamation.
The full coverage path planning means that the robot traverses all areas except the obstacles without collision in an environment with the obstacles according to a certain working mode. According to the mastery degree of the environment information, the full coverage path planning can be divided into global path planning with known global environment information and local path planning with unknown environment information. The complete traversal path planning algorithm under the known environment mainly comprises a template model method and a unit decomposition method; the completely traversed path planning method under the unknown environment mainly comprises a random traversal method, a sensor-based method and an algorithm based on a biological excitation neural network. The template model method mainly completes traversal by matching the acquired environment information with each template, but the method lacks overall planning for the environment and is difficult to adapt to the changing environment. The unit decomposition method mainly comprises a trapezoidal decomposition method, a cattle cultivation type decomposition method and a Morse decomposition method, can simplify the complexity of the whole environment into the complexity of the environment in a subregion, and is widely applied to large known environments. The random traversal method is a path planning method based on an environment-free model, and the method has the advantages of simple cleaning process, high repetition rate, low coverage rate, overlong cleaning time and the like. The sensor-based method, which requires the construction of environmental information by various sensors, is costly and difficult in map creation. Therefore, no path planning method which has good environment adaptability and obstacle avoidance capability and is low in implementation difficulty of full-coverage path planning exists in the abandoned mining area at present.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a full-coverage path planning method for a mobile robot in a mining area abandoned land, which has good environment adaptability and obstacle avoidance capability, can effectively reduce the difficulty in realizing the full-coverage path planning, and realizes the full-coverage path planning of the mining area abandoned land.
In order to achieve the purpose, the invention adopts the technical scheme that: a full-coverage path planning method for mobile robots in abandoned areas of mining areas comprises the following specific steps:
A. obtaining an original map of a mining area: obtaining a satellite map of the ground of the mining area in a remote sensing satellite system, wherein the satellite map is used as an original map of the ground of the mining area;
B. regional decomposition of the original map: adopting a cattle-ploughing decomposition method to carry out regional decomposition on an original map, and decomposing the original map into a plurality of sub-regions without obstacles;
C. determining the traversal order of each sub-region: planning the traversal sequence of each subregion by adopting a depth-first search algorithm;
D. each sub-region traverses: adopting a biological excitation neural network algorithm to complete internal traversal of each sub-area and path transfer between the sub-areas, thereby obtaining complete coverage of the whole environment;
E. verifying the performance of the planned path: design path cost function, and A*The algorithm carries out a comparison experiment, and the performance of the biostimulation neural network algorithm in the path transfer between the sub-regions is evaluated from the aspects of time cost and path cost.
Further, the cattle cultivation type decomposition method in the step B comprises the following steps:
(1) carrying out corrosion operation on the image of the original map;
(2) traversing array columns of the processed image, judging slice connectivity, and returning the number of connectivity and a connected region;
(3) if the connectivity of the slices is changed, judging whether the connected areas returned in the step (2) are separation events or combination events, and returning the result to the current sub-area for storage;
(4) and representing the divided sub-areas on the original map.
Further, the specific steps of using the depth-first search algorithm in step C are:
(1) determining a starting point v: constructing an adjacency graph, representing each sub-region by each vertex in the adjacency graph, and selecting any vertex of the sub-region as a starting point v;
(2) starting from the starting point v, any adjacent vertex around the starting point v is visited as a vertex w1；
(3) From the vertex w1Starting, accessing and vertex w1Adjacent to vertices w that have not yet been visited2And so on until the last vertex w is reached where all adjacent vertices have been visitednUntil the end;
(4) from the vertex wnPerforming backtracking operation to return to the previous vertex wn-1Checking whether other missed vertexes exist or not, if so, accessing the vertex, and if not, returning the previous vertex from the current vertex;
(5) and circulating the above operations until all the vertexes are accessed, and outputting the vertex sequence of the array so as to determine the traversal order of each sub-region.
Further, the specific steps of using the biostimulation neural network algorithm in the step D are as follows:
firstly, rasterizing each sub-area map, wherein grids in the map have two states: a free grid and an obstacle grid, the free grid being represented by "0", the obstacle grid being represented by "1", each grid representing a neuron, the entire map becoming a topological state space composed of a neural network, wherein an activity value of the neuron is represented by the following formula:
Figure BDA0002138723290000031
the above equation is a flow-splitting equation, xiIs the activity value of the ith neuron; A. b and D are non-negative constantsWherein A represents the decay rate, B represents the upper limit of the neuronal activity state, and D represents the lower limit of the neuronal activity state; k represents the number of neurons adjacent to the ith neuron; i isiIs an external input; wherein IiIs defined as:
Figure BDA0002138723290000032
e is a far greater normal number than B;
Figure BDA0002138723290000033
and [ Ii]-Respectively representing excitatory and inhibitory inputs, wij＝f(dij)，dijIs the Euclidean distance of the position of the ith neuron and the jth neuron in the state space, and f is any monotone decreasing function; f is defined as:
Figure BDA0002138723290000034
the neuron is in (0, r)0) Has local links in a range called the receptive field of neuron i, and the weight of the connection between neurons is symmetric, i.e., wij＝wji；
The specific path for point-to-point path planning by adopting the biostimulation neural network algorithm is as follows:
Figure BDA0002138723290000041
the specific path for carrying out the full coverage path planning is as follows:
Figure BDA0002138723290000042
wherein c is a normal number, yjIs compared with the position p of the last step of the robotpCurrent location pcNext, nextPosition p of stepjFunction of correlation, yjThe function is defined as:
Figure BDA0002138723290000043
where Δ θj∈[0,π]Which represents the angle between the current direction of movement and the next direction of movement.
Further, the path cost function in step E is composed of a path transfer distance and a path transfer time, and defines a series of transfer paths P composed of points:
P＝{p1,p2,…,pk}
wherein p is1,p2,…,pkThe cost of the path P consumption is C (P) when the path P passes through the grid coordinate points, so that the path cost is the transfer distance d (P) between gridsi,pi+1) And the time t consumed for completing this task, the transition cost of said path P being represented by the following equation:
Figure BDA0002138723290000044
d (p) isi,pi+1) Representing the Euclidean distance between two grids, and the value is represented by the following formula:
Figure BDA0002138723290000045
therefore, the objective of the inter-area path transfer algorithm is to find a path P, so that c (P) satisfies:
C(P)＝minC(Pm)
wherein, PmAre m paths from the end of the previous region to the start of the next region.
Compared with the prior art, the invention adopts a region decomposition method combined with a biostimulation neural network algorithm to complete the full-coverage path planning of the mobile robot on the abandoned land of the mining area, and can complete the reciprocating coverage of the robot on the inner part of the sub-area and the path transfer among the sub-areas. The cattle cultivation type decomposition method carries out regional decomposition on the environment, and can effectively reduce the difficulty of realizing the full-coverage path planning; the 'reciprocating' covering is completed by adopting a biological excitation neural network algorithm in the subareas, so that the sudden situation can be effectively coped with, and an obstacle avoidance strategy is timely carried out on the obstacle; the algorithm adopted in the region transfer part can effectively reduce the path redundancy of the robot among the sub-regions. In addition, the bio-excitation neural network algorithm does not need a learning process, has small calculation complexity, is suitable for processing dynamic unknown conditions, and is an algorithm with high execution efficiency. By adopting the biostimulation neural network algorithm, unknown obstacles in the environment can be processed in time, and the path planning of the whole area of the mobile robot can be realized. The method has good environment adaptability and obstacle avoidance capability, can effectively reduce the difficulty of realizing the full-coverage path planning, and realizes the full-coverage path planning of the abandoned land in the mining area.
Drawings
FIG. 1 is a flow chart of the zone decomposition method of the present invention.
Fig. 2 is a simulated environment of a mine site abandoned according to the present invention.
FIG. 3 is a drawing showing the result of the cultivation of cattle according to the present invention.
FIG. 4 is a diagram of the adjacency of sub-regions in accordance with the present invention.
FIG. 5 is a sequence diagram of sub-region planning according to the present invention.
FIG. 6 is a flow chart of the biostimulating neural network algorithm of the present invention.
FIG. 7 is a graph of the full coverage results of the present invention without path shifting.
FIG. 8 is a graph of the full coverage results of the present invention with path shifting.
FIG. 9 is a comparison of the path transfer algorithm of the present invention.
FIG. 10 is a graph showing the comparison result of the paths of the present invention.
FIG. 11 is a graph showing the time comparison results of the present invention.
Detailed Description
The present invention will be further explained below.
As shown in the figure, a method for planning a full-coverage path of a mobile robot in a abandoned land of a mining area comprises the following specific steps:
A. obtaining an original map of a mining area: obtaining a satellite map of the ground of the mining area in a remote sensing satellite system, wherein the satellite map is used as an original map of the ground of the mining area;
B. regional decomposition of the original map: adopting a cattle-ploughing decomposition method to carry out regional decomposition on an original map, and decomposing the original map into a plurality of sub-regions without obstacles; the cattle-farming decomposition method is a regional decomposition method that can effectively divide an environment including irregular obstacles, and comprises the steps of:
(1) carrying out corrosion operation on the image of the original map; the erosion operation is to clearly compare the obstacle region and the free region in the image;
(2) traversing array columns of the processed image, judging slice connectivity, and returning the number of connectivity and a connected region;
(3) if the connectivity of the slices is changed, judging whether the connected areas returned in the step (2) are separation events or combination events, and returning the result to the current sub-area for storage;
(4) and representing the divided sub-areas on the original map.
The method assumes that a vertical line (called a slice) sweeps a bounded environment filled with polygonal barriers from left to right, and is divided every time a critical point with a vertex capable of extending up and down is encountered, and finally the environment is divided into a plurality of sub-areas without the barriers; fig. 2 is an environment map of a mining area waste land, and fig. 3 is an entire environment map divided by a plowing decomposition method.
C. Determining the traversal order of each sub-region: planning the traversal sequence of each subregion by adopting a depth-first search algorithm; the adjacent relation between the regions can be expressed by constructing an adjacent map among the subregions, a depth-first search algorithm is a search algorithm in a graph theory, is often used for traversing all points in the map and finding a detailed coverage path for driving, and comprises the following specific steps:
(1) determining a starting point v: constructing an adjacency graph, representing each sub-region by each vertex in the adjacency graph, and selecting any vertex of the sub-region as a starting point v;
(2) starting from the starting point v, any adjacent vertex around the starting point v is visited as a vertex w1；
(3) From the vertex w1Starting, accessing and vertex w1Adjacent to vertices w that have not yet been visited2And so on until the last vertex w is reached where all adjacent vertices have been visitednUntil the end;
(4) from the vertex wnPerforming backtracking operation to return to the previous vertex w n-1, checking whether other missed vertexes exist or not, if so, accessing the vertex, and if not, returning the previous vertex from the current vertex;
(5) and circulating the above operations until all the vertexes are accessed, and outputting the vertex sequence of the array so as to determine the traversal order of each sub-region. FIG. 4 is a diagram of adjacency between sub-regions, and FIG. 5 is a sequence between sub-regions planned after a depth-first search algorithm is employed;
D. each sub-region traverses: adopting a biological excitation neural network algorithm to complete internal traversal of each sub-area and path transfer between the sub-areas, thereby obtaining complete coverage of the whole environment; the method comprises the following specific steps:
firstly, rasterizing each sub-area map, wherein grids in the map have two states: a free grid and an obstacle grid, the free grid being represented by "0", the obstacle grid being represented by "1", each grid representing a neuron, the entire map becoming a topological state space composed of a neural network, wherein an activity value of the neuron is represented by the following formula:
Figure BDA0002138723290000071
the above equation is a flow-splitting equation, xiIs the activity value of the ith neuron; A. b and D are non-negative constants, wherein A represents the decay rate, B represents the upper limit of the neuron activity state, and D represents the lower limit of the neuron activity state; k represents andthe number of adjacent neurons of the ith neuron; i isiIs an external input; wherein IiIs defined as:
Figure BDA0002138723290000072
e is a far greater normal number than B;
Figure BDA0002138723290000073
and [ Ii]-Respectively representing excitatory and inhibitory inputs, wij＝f(dij)，dijIs the Euclidean distance of the position of the ith neuron and the jth neuron in the state space, and f is any monotone decreasing function; f is defined as:
Figure BDA0002138723290000074
the neuron is in (0, r)0) Has local links in a range called the receptive field of neuron i, and the weight of the connection between neurons is symmetric, i.e., wij＝wji(ii) a The calculation process of the bio-excitation neural network algorithm is shown in fig. 6.
The specific path for point-to-point path planning by adopting the biostimulation neural network algorithm is as follows:
Figure BDA0002138723290000075
the path generation process is as follows: the robot selects a neuron grid with the maximum activity value in adjacent neurons as a next position at the current position, and the next position becomes a new current position after the next position is reached, and so on until a target point grid is reached; if the activity values of the peripheral 8 adjacent neurons are not larger than the neuron activity of the current grid, the robot stays still in place and falls into a dead zone.
The specific path for carrying out the full coverage path planning is as follows:
Figure BDA0002138723290000081
wherein c is a normal number, yjIs compared with the position p of the last step of the robotpCurrent location pcThe next step is at position pjFunction of correlation, yjThe function is defined as:
Figure BDA0002138723290000082
where Δ θj∈[0,π]Which represents the angle between the current direction of movement and the next direction of movement.
The full coverage path planning path generation process is similar to the point-to-point path planning until the robot achieves full coverage of the environment. Fig. 7 is a process that the bio-excitation neural network algorithm completes the full coverage in each sub-region, the starting position of the robot is (2,2), the parameters of the adopted shunt equation are E-100, a-20, B-1, D-1, u-1, and c-1, as can be seen from the figure, the total coverage area of the robot reaches 100%, and the feasibility of the algorithm is effectively proved.
After the robot completes the coverage of the inside of one sub-area, the robot needs to transfer from one sub-area to another sub-area, and at this time, a point-to-point path planning needs to be completed by adopting a biological excitation neural network, and fig. 8 shows the result that the robot simultaneously completes the coverage of the inside of the sub-area and the transfer of the paths between the sub-areas. In the figure, the starting point of each subregion is indicated by an asterisk, the node of each subregion is indicated by a five-pointed star, and the algorithm mainly completes the path transfer between the subregions in three parts. Finally, after debugging, when mu is 1 between G-H areas, mu is 3 between H-I areas, and mu is 1 between I-J areas, the three transfer paths are shortest. The transfer path between the sub-regions completed by the biostimulated neural network algorithm is indicated by a cross in the figure.
E. Verifying the performance of the planned path: the cost function of the path is designed,and A*The algorithm carries out a comparison experiment, and the performance of the biostimulation neural network algorithm in the path transfer between the sub-regions is evaluated from the aspects of time cost and path cost.
The path cost function is composed of path transfer distance and path transfer time, and defines a series of transfer paths P composed of points:
P＝{p1,p2,…,pk}
wherein p is1,p2,…,pkThe cost of the path P consumption is C (P) when the path P passes through the grid coordinate points, so that the path cost is the transfer distance d (P) between gridsi,pi+1) And the time t consumed for completing this task, the transition cost of said path P being represented by the following equation:
Figure BDA0002138723290000091
d (p) isi,pi+1) Representing the Euclidean distance between two grids, and the value is represented by the following formula:
Figure BDA0002138723290000092
therefore, the objective of the inter-area path transfer algorithm is to find a path P, so that c (P) satisfies:
C(P)＝minC(Pm)
wherein, PmAre m paths from the end of the previous region to the start of the next region.
And selecting a classic path planning algorithm A to carry out a comparison test. In general, the path transfer distance and the path transfer time are important indexes for measuring the coverage task. The path transfer distance is the sum of Euclidean distances of all grids in the process that the robot is transferred from a node of one sub-region to the starting point of the other sub-region; the path transfer time is the time consumed for completing the task, and the total cost of the shortest transfer path of the two algorithms is evaluated through the two performance indexes. The "circles" in fig. 9 represent the region transfer paths obtained by the a-algorithm, and fig. 9 better shows the results of the comparative experiment of the two algorithms.
Under the same external conditions, fig. 10 and 11 show the comparison result of the two algorithms in the path transfer distance and the path transfer time consumption cost. Fig. 10 is a distance comparison graph of the transfer portions of the three regions for the two algorithms, and fig. 11 is a time comparison graph for the two algorithms. From the simulation result graph and the comparison graph, it can be seen that the biostimulation neural network algorithm is superior to the a-algorithm in terms of both the path transfer time and the path transfer distance. Simulation experiment results prove that the biostimulation neural network algorithm can obtain lower path transfer cost and has good performance in path planning problems.

Claims (1)

1. A full-coverage path planning method for mobile robots in abandoned areas of mining areas is characterized by comprising the following specific steps:
A. obtaining an original map of a mining area: obtaining a satellite map of the ground of the mining area in a remote sensing satellite system, wherein the satellite map is used as an original map of the ground of the mining area;
B. regional decomposition of the original map: adopting a cattle-ploughing decomposition method to carry out regional decomposition on an original map, and decomposing the original map into a plurality of sub-regions without obstacles; the method comprises the following specific steps:
(1) carrying out corrosion operation on the image of the original map;
(2) traversing array columns of the processed image, judging slice connectivity, and returning the number of connectivity and a connected region;
(3) if the connectivity of the slices is changed, judging whether the connected areas returned in the step (2) are separation events or combination events, and returning the result to the current sub-area for storage;
(4) representing the divided sub-areas on an original map;
C. determining the traversal order of each sub-region: planning the traversal sequence of each subregion by adopting a depth-first search algorithm; the method comprises the following specific steps:
(1) determining a starting point v: constructing an adjacency graph, representing each sub-region by each vertex in the adjacency graph, and selecting any vertex of the sub-region as a starting point v;
(2) starting from the starting point v, any adjacent vertex around the starting point v is visited as a vertex w1；
(3) From the vertex w1Starting, accessing and vertex w1Adjacent to vertices w that have not yet been visited2And so on until the last vertex w is reached where all adjacent vertices have been visitednUntil the end;
(4) from the vertex wnPerforming backtracking operation to return to the previous vertex wn-1Checking whether other missed vertexes exist or not, if so, accessing the vertex, and if not, returning the previous vertex from the current vertex;
(5) the operation is circulated until all the vertexes are accessed, and the vertex sequence of the array is output, so that the traversal sequence of each sub-region is determined;
D. each sub-region traverses: adopting a biological excitation neural network algorithm to complete internal traversal of each sub-area and path transfer between the sub-areas, thereby obtaining complete coverage of the whole environment; the specific steps of the biostimulation neural network algorithm are as follows:
firstly, rasterizing each sub-area map, wherein grids in the map have two states: a free grid and an obstacle grid, the free grid being represented by "0", the obstacle grid being represented by "1", each grid representing a neuron, the entire map becoming a topological state space composed of a neural network, wherein an activity value of the neuron is represented by the following formula:
Figure FDA0003099015530000021
the above equation is a flow-splitting equation, xiIs the activity value of the ith neuron; A. b and D are non-negative constants, wherein A represents the decay rate, B represents the upper limit of the neuron activity state, and D represents the lower limit of the neuron activity state; k represents the number of neurons adjacent to the ith neuron; i isiIs an external input; wherein IiIs defined as:
Figure FDA0003099015530000022
e is a far greater normal number than B;
Figure FDA0003099015530000023
and [ Ii]-Respectively representing excitatory and inhibitory inputs, wij＝f(dij)，dijIs the Euclidean distance of the position of the ith neuron and the jth neuron in the state space, and f is any monotone decreasing function; f is defined as:
Figure FDA0003099015530000024
the neuron is in (0, r)0) Has local links in a range called the receptive field of neuron i, and the weight of the connection between neurons is symmetric, i.e., wij＝wji；
The specific path for point-to-point path planning by adopting the biostimulation neural network algorithm is as follows:
Figure FDA0003099015530000025
the specific path for carrying out the full coverage path planning is as follows:
Figure FDA0003099015530000026
wherein c is a normal number, yjIs compared with the position p of the last step of the robotpCurrent location pcThe next step is at position pjFunction of correlation, yjThe function is defined as:
Figure FDA0003099015530000031
where Δ θj∈[0,π]It represents the angle between the current moving direction and the next moving direction;
E. verifying the performance of the planned path: design path cost function, and A*The algorithm carries out a comparison experiment, and the performance of the biostimulation neural network algorithm in the path transfer between the sub-regions is evaluated from the aspects of time cost and path cost; the path cost function is composed of path transfer distance and path transfer time, and defines a series of transfer paths P composed of points:
P＝{p1,p2,…,pk}
wherein p is1,p2,…,pkThe cost of the path P consumption is C (P) when the path P passes through the grid coordinate points, so that the path cost is the transfer distance d (P) between gridsi,pi+1) And the time t consumed for completing this task, the transition cost of said path P being represented by the following equation:
Figure FDA0003099015530000032
d (p) isi,pi+1) Representing the Euclidean distance between two grids, and the value is represented by the following formula:
Figure FDA0003099015530000033
therefore, the objective of the inter-area path transfer algorithm is to find a path P, so that c (P) satisfies:
C(P)＝min C(Pm)
wherein, PmAre m paths from the end of the previous region to the start of the next region.
CN201910661538.XA 2019-07-22 2019-07-22 Full-coverage path planning method for mobile robot in abandoned land of mining area Active CN110502006B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910661538.XA CN110502006B (en) 2019-07-22 2019-07-22 Full-coverage path planning method for mobile robot in abandoned land of mining area

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910661538.XA CN110502006B (en) 2019-07-22 2019-07-22 Full-coverage path planning method for mobile robot in abandoned land of mining area

Publications (2)

Publication Number Publication Date
CN110502006A CN110502006A (en) 2019-11-26
CN110502006B true CN110502006B (en) 2021-12-24

Family

ID=68586666

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910661538.XA Active CN110502006B (en) 2019-07-22 2019-07-22 Full-coverage path planning method for mobile robot in abandoned land of mining area

Country Status (1)

Country Link
CN (1) CN110502006B (en)

Families Citing this family (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111202471A (en) * 2020-01-08 2020-05-29 上海高仙自动化科技发展有限公司 Full-coverage path generation method and generation device, intelligent robot and storage medium
CN111723986A (en) * 2020-06-18 2020-09-29 上海海事大学 Region division and combination method based on depth-first traversal
CN112147998B (en) * 2020-08-24 2021-12-07 同济大学 Mobile robot path planning method based on region growing method
CN112025715A (en) * 2020-09-21 2020-12-04 中国矿业大学 Mobile robot double-layer path planning method with unknown local environment
CN112504273B (en) * 2020-10-29 2022-05-24 广东杜尼智能机器人工程技术研究中心有限公司 Seamless connection planning method for arcuate path
CN113359699B (en) * 2021-05-08 2024-01-12 五邑大学 Full-coverage path planning method, device and storage medium
CN113741425B (en) * 2021-08-04 2023-11-24 中山大学 Full-coverage path planning method and navigation system
CN113759907A (en) * 2021-08-30 2021-12-07 武汉理工大学 Method, device, equipment and storage medium for planning regional full-coverage path
CN113935516B (en) * 2021-09-07 2022-10-04 亚太卫星宽带通信（深圳）有限公司 Airborne antenna rapid satellite finding, tracking and positioning method
CN113867355B (en) * 2021-10-11 2023-07-21 复旦大学 Full-coverage path planning method suitable for desert complex terrain
CN113792109B (en) * 2021-11-18 2022-02-11 上海景吾智能科技有限公司 Path planning method, path planning system, robot, and storage medium
CN114089765A (en) * 2021-11-22 2022-02-25 江苏科技大学 Urban green land-oriented traversing path planning method for mowing robot
CN114489068A (en) * 2022-01-24 2022-05-13 上海利淘豪斯机器人有限公司 Routing method and device for routing inspection task path of inspection robot under complex path
CN117490703B (en) * 2023-12-29 2024-03-08 博创联动科技股份有限公司 Agricultural machinery operation path planning method and system
CN117804463B (en) * 2024-03-01 2024-06-28 广东众聚人工智能科技有限公司 Intelligent investigation robot and path planning method

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102890507A (en) * 2011-07-21 2013-01-23 鸿奇机器人股份有限公司 Self-walking robot, cleaning robot and positioning method thereof
CN105922267A (en) * 2016-06-28 2016-09-07 山东理工大学 Design method of full-coverage traversal path planning algorithm of cleaning robot
CN106843216A (en) * 2017-02-15 2017-06-13 北京大学深圳研究生院 A kind of complete traverse path planing method of biological excitation robot based on backtracking search
CN106979785A (en) * 2017-03-24 2017-07-25 北京大学深圳研究生院 A kind of complete traverse path planing method of multi-robot system

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9266611B2 (en) * 2013-06-20 2016-02-23 University Of Florida Research Foundation, Inc. Flight path development for remote sensing vehicles in a moving reference frame

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102890507A (en) * 2011-07-21 2013-01-23 鸿奇机器人股份有限公司 Self-walking robot, cleaning robot and positioning method thereof
CN105922267A (en) * 2016-06-28 2016-09-07 山东理工大学 Design method of full-coverage traversal path planning algorithm of cleaning robot
CN106843216A (en) * 2017-02-15 2017-06-13 北京大学深圳研究生院 A kind of complete traverse path planing method of biological excitation robot based on backtracking search
CN106979785A (en) * 2017-03-24 2017-07-25 北京大学深圳研究生院 A kind of complete traverse path planing method of multi-robot system

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
Complete coverage path planning based on bioinspired neural network and pedestrian location prediction;Chunyu Yang etc.;《Proceedings of 2018 IEEE 8th Annual International Conference》;20180715;第528-530页 *
移动机器人增量地图分解与覆盖路径技术研究;闫帅;《中国优秀硕士学位论文全文数据库 信息科技辑》;20181015(第10期);正文第14-21、36-42页 *

Also Published As

Publication number Publication date
CN110502006A (en) 2019-11-26

Similar Documents

Publication Publication Date Title
CN110502006B (en) Full-coverage path planning method for mobile robot in abandoned land of mining area
Shrestha et al. Learned map prediction for enhanced mobile robot exploration
Yang et al. A neural network approach to complete coverage path planning
Moorehead et al. Autonomous exploration using multiple sources of information
CN112925315A (en) Crawler path planning method based on improved ant colony algorithm and A-star algorithm
Sadat et al. Recursive non-uniform coverage of unknown terrains for uavs
US6636847B1 (en) Exhaustive search system and method using space-filling curves
Guo et al. An improved a-star algorithm for complete coverage path planning of unmanned ships
CN112025715A (en) Mobile robot double-layer path planning method with unknown local environment
Paull et al. An information gain based adaptive path planning method for an autonomous underwater vehicle using sidescan sonar
Bin et al. Recurrent neural network for robot path planning
Zedadra et al. S-MASA: A stigmergy based algorithm for multi-target search
Zhao et al. A fast robot path planning algorithm based on bidirectional associative learning
Yoon et al. Collaborative mission and route planning of multi-vehicle systems for autonomous search in marine environment
Vckay et al. Solving a Path Planning Problem in a Partially Known Environment using a Swarm Algorithm
Liu et al. The path planning study of autonomous patrol robot based on modified Astar algorithm and genetic algorithm
CN109977455B (en) Ant colony optimization path construction method suitable for three-dimensional space with terrain obstacles
Kalaivanan et al. Coverage path planning for an autonomous robot specific to agricultural operations
Kang et al. A research on area coverage algorithm for robotics
Dhiman et al. A review of path planning and mapping technologies for autonomous mobile robot systems
Castellini et al. Online Monte Carlo Planning for Autonomous Robots: Exploiting Prior Knowledge on Task Similarities.
Nair Efficient Path Planning Algorithm for Mobile Robots Performing Floor Cleaning Like Operations
Morin et al. A hybrid algorithm for coverage path planning with imperfect sensors
Ahmad Distributed navigation of multi-robot systems for sensing coverage
Wurm et al. Improved Simultaneous Localization and Mapping using a Dual Representation of the Environment.

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant