CN113627646A - Path planning method, device, equipment and medium based on neural network - Google Patents
Path planning method, device, equipment and medium based on neural network Download PDFInfo
- Publication number
- CN113627646A CN113627646A CN202110771368.8A CN202110771368A CN113627646A CN 113627646 A CN113627646 A CN 113627646A CN 202110771368 A CN202110771368 A CN 202110771368A CN 113627646 A CN113627646 A CN 113627646A
- Authority
- CN
- China
- Prior art keywords
- neural network
- target
- agv
- path planning
- region
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 182
- 238000000034 method Methods 0.000 title claims abstract description 52
- 210000002569 neuron Anatomy 0.000 claims abstract description 68
- 238000010276 construction Methods 0.000 claims description 29
- 238000003860 storage Methods 0.000 claims description 25
- 238000001514 detection method Methods 0.000 claims description 4
- 230000006870 function Effects 0.000 description 11
- 230000004913 activation Effects 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 8
- 238000012545 processing Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 5
- 238000012549 training Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000013178 mathematical model Methods 0.000 description 4
- 230000006872 improvement Effects 0.000 description 3
- 238000009825 accumulation Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 210000002856 peripheral neuron Anatomy 0.000 description 1
- 230000035479 physiological effects, processes and functions Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000000638 stimulation Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
- G06Q10/047—Optimisation of routes or paths, e.g. travelling salesman problem
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
- G06Q10/083—Shipping
- G06Q10/0835—Relationships between shipper or supplier and carriers
- G06Q10/08355—Routing methods
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a path planning method, a device, equipment and a medium based on a neural network, wherein the method comprises the following steps: acquiring regional information; dividing the region into a plurality of grids with consistent areas and shapes according to the region information; determining a neural network structure corresponding to the grids according to the grids, wherein each neuron in the neural network structure corresponds to each grid position one by one; determining a target point and a plurality of AGVs from the plurality of neurons; constructing a target neural network corresponding to the target point; obtaining the state value of each AGV position according to the target neural network; and when the state values of the target point and the AGV positions are not equal to 0, finishing constructing the target neural network so as to determine the target AGV and the corresponding path. The efficiency of constructing the neural network is improved by changing the termination condition of constructing the neural network; the problem of long time consuming in constructing a neural network to reach a stable state to complete path planning in the prior art is solved.
Description
Technical Field
The present invention relates to the field of path planning technologies, and in particular, to a method, an apparatus, a device, and a medium for path planning based on a neural network.
Background
With the development of scientific technology and the gradual maturity of automated logistics technology, when an Automated Guided Vehicle (AGV) faces a complex activity area, the characteristics of automation and automation can be fully embodied no matter warehouse logistics transportation is carried out or the AGV is applied to an automatic production line, and the multi-AGV multi-task allocation and path planning are the key to whether a multi-AGV system can safely and efficiently complete a transportation task.
Aiming at the problems that multiple AGV multiple target points have unreasonable task allocation and unsmooth path planning results to generate more broken lines, path evaluation indexes of path length, path smoothness and path risk degree are integrated, redundant points in path points are identified according to the navigation mode of the AGV, the planned path is subjected to path smoothing processing and optimization, a multiple AGV path planning algorithm based on a biological heuristic neural network is provided, the task allocation is more reasonable, the planned path is shorter and safer, and after the smoothing processing and the optimization, the path is more optimal, and the high-efficiency navigation of the AGV is favorably realized. However, when the area is large, the number of two-dimensional grids is large after rasterization, the process of training the GBNN calculation according to the GBNN mathematical model is complex, and a long time is required for training the state values of each grid position to be not changed any more, so that the GBNN network construction is completed when determining that the neural network reaches a stable state, the state values of each grid position are not changed any more to serve as judgment conditions for completing the construction of the network, a large amount of time is consumed for completing the calculation process, and the time cost is wasted.
Therefore, it is needed to provide a path planning method that can quickly complete task allocation and path planning tasks and can achieve the shortest path to the target point to solve the above technical problems.
Disclosure of Invention
In order to solve the technical problem, the invention provides a path planning method based on a neural network. The problem of among the prior art reach steady state through constructing neural network and accomplish path planning, consuming time overlength is solved.
The technical effects of the invention are realized as follows:
a neural network based path planning method, the method comprising:
acquiring regional information;
dividing the region into a plurality of grids with consistent areas and shapes according to the region information;
determining a neural network structure corresponding to the grids according to the grids, wherein each neuron in the neural network structure corresponds to each grid position one by one;
determining a target point and positions of a plurality of AGVs from a plurality of neurons;
constructing a target neural network corresponding to the target point;
obtaining the state value of each AGV position according to the target neural network;
and when the state value of the target point and the state values of the AGV positions are not equal to 0, finishing constructing the target neural network so as to determine the target AGV and the corresponding path.
Further, the grids include a free grid and a non-free grid, the region is divided into a plurality of grids with consistent areas and shapes according to the region information, and then the method includes:
dividing the region into a free passing region and an obstacle region according to the region information;
determining a corresponding free grid according to the free passing area;
determining a corresponding non-free grid according to the obstacle region;
further, obtaining the state value of each AGV position according to the target neural network includes:
obtaining state values of eight adjacent neuron positions around the target point according to the target neural network;
and obtaining the state value of each neuron position based on the state values of eight adjacent neuron positions around the target point so as to obtain the state value of each AGV position.
Further, the determining a target AGV and a corresponding path includes:
comparing the state values of the AGV positions;
determining a target AGV according to the AGV with the largest state value;
and the path of the target AGV reaching the target point is the path corresponding to the target AGV.
Further, when the state value of the target point and the state value of each AGV position are not equal to 0, completing the construction of the target neural network, and then:
judging whether the target neural networks respectively corresponding to the other remaining target points are established;
and if so, finishing the construction of the target neural networks respectively corresponding to all the target points so as to determine the target AGV corresponding to each target point and the corresponding path.
In addition, a path planning device based on a neural network is also provided, the device includes:
an area information acquisition module: for obtaining area information;
a grid region construction module: the grid is used for dividing the region into a plurality of grids with consistent areas and shapes according to the region information;
a neural network structure determination module: the neural network structure is used for determining a neural network structure corresponding to a plurality of grids according to the grids, and each neuron in the neural network structure corresponds to each grid position one by one;
a target point determination module: the target point and the AGVs are determined from the neurons;
a target neural network construction module: the target neural network is used for constructing a target neural network corresponding to the target point;
AGV status value gets module: the AGV position detection device is used for obtaining a state value of each AGV position according to the target neural network;
the single task path planning determination module: and when the state values of the target point and the AGV positions are not equal to 0, finishing constructing the target neural network so as to determine a target AGV and a corresponding path.
Further, the apparatus further comprises:
a target neural network judgment module: the target neural network is used for judging whether the target neural networks respectively corresponding to the other remaining target points are established;
the multi-task path planning determination module: and when the target neural networks corresponding to the other remaining target points are established, the target neural networks corresponding to all the target points are established, so as to determine the target AGV corresponding to each target point and the corresponding path.
In addition, an apparatus is provided, which includes a processor and a memory, where at least one instruction, at least one program, a set of codes, or a set of instructions is stored in the memory, and the at least one instruction, the at least one program, the set of codes, or the set of instructions is loaded and executed by the processor to implement one of the above-mentioned neural network-based path planning methods.
Additionally, a computer storage medium is provided having at least one instruction, at least one program, set of codes, or set of instructions stored therein, which is loaded and executed by a processor to implement a neural network based path planning method as described above.
As described above, the present invention has the following advantageous effects:
1) when the neural network is constructed, the termination condition for constructing the neural network is changed, the learning time of the neural network is shortened, the time performance is better, and the problem that the time consumption is too long when the path planning is completed by constructing the neural network to reach the stable state in the prior art is solved.
2) A shorter and safer path is planned by judging the position with the maximum state value in real time, so that the task allocation is more reasonable, and the work efficiency is improved.
Drawings
In order to more clearly illustrate the technical solution of the present invention, the drawings used in the description of the embodiment or the prior art will be briefly described below. It is obvious that the drawings in the following description are only some embodiments of the invention, and that for a person skilled in the art it is also possible to derive other drawings from these drawings without inventive effort.
Fig. 1 is a flowchart of a path planning method based on a neural network according to an embodiment of the present disclosure;
FIG. 2 is a schematic diagram of a neural network structure provided in an embodiment of the present disclosure;
FIG. 3 is a schematic diagram of a warehousing area construction neural network provided by an embodiment of the present disclosure;
FIG. 4 is a diagram illustrating state values of positions of a target neural network according to an embodiment of the present disclosure;
FIG. 5 is a flowchart illustrating steps provided in an embodiment of the present disclosure to obtain a status value of each AGV position according to the target neural network;
FIG. 6 is a flowchart illustrating an implementation of determining an end of building a target neural network according to an embodiment of the present disclosure;
fig. 7 is a flowchart of another neural network-based path planning method provided in an embodiment of the present disclosure;
FIG. 8 is a flowchart illustrating another implementation of determining the end of building a target neural network according to an embodiment of the present disclosure;
fig. 9 is a block diagram illustrating a path planning apparatus based on a neural network according to an embodiment of the present disclosure;
fig. 10 is a block diagram of another neural network-based path planning apparatus provided in an embodiment of the present disclosure;
fig. 11 is a schematic structural diagram of a server device provided in an embodiment of the present specification.
Detailed Description
The technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the drawings in the embodiments of the present disclosure, and it is obvious that the described embodiments are only a part of the embodiments of the present disclosure, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or server that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
In the technical scheme of constructing a neural network based on a grid method for path planning, GBNN construction is completed when the neural network reaches a stable state, but the GBNN construction is taken as a judgment condition for completing the construction of the network, and the calculation of the state value of the neuron corresponding to each grid position can be completed within a long time. When the area is large, the number of two-dimensional grids is large after rasterization, the calculation process of training the GBNN according to the mathematical model is complex, and a long time is required for training the state values of all grid positions to be not changed any more, so that the time cost for training the GBNN is too high, and the efficiency for completing a path planning task is reduced.
Therefore, the present specification provides a path planning method for constructing a neural network by changing a termination condition for constructing the neural network, which can shorten the learning time of the neural network by determining that state values of all AGVs and other target points are no longer zero as the termination condition for constructing the neural network when constructing the biological heuristic neural network, has a better time performance, improves the efficiency of completing a path planning task, and solves the problem that the time consumption is too long when completing the path planning by constructing the neural network to reach a stable state in the prior art.
An embodiment of the present specification provides a path planning method based on a neural network, and as shown in fig. 1, the method includes:
s110: acquiring regional information;
s210: dividing the region into a plurality of grids with consistent areas and shapes according to the region information;
in a specific embodiment, the grids include a free grid and a non-free grid, and the step S200 is to divide the region into several grids with consistent areas and shapes according to the region information, and then includes:
dividing the region into a free passing region and an obstacle region according to the region information;
determining a corresponding free grid according to the free passing area;
determining a corresponding non-free grid according to the obstacle region;
specifically, the region is a search region of a two-dimensional space, which refers to a storage region in this embodiment, the storage region is divided into a plurality of grids having the same area and shape by obtaining information of the storage region, that is, the region constructed by the divided grids is represented by a rectangular coordinate system method, and a horizontal and vertical coordinate form is adopted for recording each grid. Two types of grids are typically included in the rasterized region: a free grid representing a spatial position without obstacles, i.e. a region through which the robot can pass, and a non-free grid representing a spatial position in which obstacles are present, which the robot must bypass.
When rasterizing a region, not all grids can be mapped in grid space intact. For some irregular-shaped obstacles, the obstacle cannot occupy an integral multiple of the grid space, and the obstacle adopts a binary expansion method for the edge part of the obstacle so as to fill the integral multiple of the grid space. The binary expansion method belongs to the prior art, and is not described in detail in the application.
S310: determining a neural network structure corresponding to the grids according to the grids, wherein each neuron in the neural network structure corresponds to each grid position one by one;
a Neural network structure (GBNN) is an adaptive artificial Neural network, which is composed of a plurality of nodes (also called neurons) and is formed by connecting the neurons, each neuron has an output, called a state value, each neuron outputs a special output function, the function is called an activation function, each connection between two neurons needs a connection signal, the connection signal is called a weight, the state value of each neuron is calculated by receiving the product of the state value and the weight of all other neurons and adding a bias term, and the final output result of each neuron in the Neural network is determined by the state values of other neurons, the weights and the activation functions, so that each neuron is controlled by all other neurons, the outputs of the individual neurons are then mutually constrained, each neuron has local lateral connections to its neighbors, much like neural networks in physiology, and each neuron reacts only to stimuli in its sensory neighborhood, with state values for each location calculated iteratively.
Wherein the grid positions correspond to neuron positions in a neural network structure, as shown in fig. 2, each neuron only responds to the stimulation of eight peripheral neurons adjacent to the neuron positions, and in a rectangular coordinate system constructed with the neuron positions as an origin, the eight peripheral adjacent neurons are neurons closest to the neurons in directions of 0 degrees, 45 degrees, 90 degrees, 135 degrees, 180 degrees, 225 degrees, 270 degrees and 315 degrees to the neuron positions and a Y axis of the rectangular coordinate system, respectively.
S410: determining a target point and positions of a plurality of AGVs from a plurality of neurons;
the target point and the AGVs are located at the free grid position.
S510: constructing a target neural network corresponding to the target point;
the mathematical model of the target neural network GBNN is as follows (1):
formula (1) is a formula for calculating neuron state values, c(i，j)Representing the state value of the neuron at position (I, j), t the time, g (x) is the activation function, I(i，j)And (3) a bias item representing the position of (i, j), (p, q) another neuron coordinate position, m, n represent the scale of the neural network, and the output value of each neuron at the next moment is obtained according to the accumulation and the bias item of the state value and the weight value of other neurons in the network at the last moment.
Formula (2) is a weight calculation formula, and (i, j) and (p, q) are coordinates of two neurons,
expressing Euclidean distance of two neurons, and calculating weight value in the neural network, when two neurons of (i, j) and (p, q) are adjacent neurons, the weight value between the two neurons is e-2dWhen p is equal to i and q is equal to j, the weight of the neuron and itself is 0.
Equation (3) is a calculation equation of an activation function, and each neuron obtains a final state value through the activation function.
Equation (4) is a calculation equation of the bias term for controlling the output after passing through the activation function.
In order to satisfy the condition after the activation function is passed, the state value is between 0 and 1. E must be greater than 1. At the starting moment, the state value of the position of a target point in the target neural network is initialized to 1, and the state values of the positions of the rest positions, namely the position of an obstacle, the initial position of the AGV and the positions capable of navigating are all 0. The next-in-time state of each neuron in the target neural networkThe value is obtained according to the accumulation and bias item of other neuron state values and weight value at the last moment. The state value of each neuron in the target neural network is changed from the target point position to gradually expand towards the periphery. At the first time, the state values of eight neighboring positions around the target point change from 0 due to the influence of the state value of the target point being 1. According to the formula for calculating the weight of the formula (2), the coordinates of the target point position are (p, q), and (i, j) is the coordinates of one of eight grid positions around the target point. Target point and weight W of eight surrounding positions((i，j)，(p，q))Is not 0 and C(p，q)When the state value is calculated according to equation (1), W is 1((i，j)，(p，q))C(p，q)Instead of 0, the eight neighborhood position bias terms around the target point are 0, so the state values of the eight surrounding positions change from 0 to valued after the activation function g (x). From the initial time to the first time, the state values of not only the target point positions are not 0, but also the state values of the surrounding eight positions are changed from 0 to the corresponding values according to the mathematical model, and at this time, the number of positions in the target neural network, the state values of which are not 0, is changed from only one target point to nine. At the second time, the state values of the 24 positions around the target point are not 0 because of the influence of the state values of the eight positions around the target point being not 0. By analogy, the target neural network is constructed according to the method, and the weight of the neuron in the target neural network are 0, so that the target neural network tends to be stable after a limited number of cycles. Namely, when the state values of the last time and the next time of each neuron position are not changed, the target neural network construction is finished.
For the warehousing area as shown in FIG. 3, a target neural network is constructed, and black dots TjThe target point position is indicated, the obstacle is indicated by a black grid, and the feasible position is indicated by a white grid.
The effect of the state values of the respective positions in the target neural network is shown in fig. 4, in the three-dimensional coordinate system of fig. 4, the horizontal axis X and the vertical axis Y represent the coordinates of the respective positions in the storage area, the vertical axis Z represents the state values of the respective positions, and the state value (i.e., Z-axis value) of the target point position (40, 90) is 1. In the three-dimensional coordinate system, there are two recessed positions representing the state values of the obstacle positions, and since the state values of the obstacle positions are always 0, the values of these places on the vertical axis are also 0. The closer the position state value to the target point in the target neural network is, the larger the value of the vertical axis in the three-dimensional coordinate system is.
S610: obtaining the state value of each AGV position according to the target neural network;
in a specific embodiment, the step S600 of obtaining the status value of each AGV position according to the target neural network, as shown in fig. 5, includes:
s611: obtaining state values of eight adjacent neuron positions around the target point according to the target neural network;
s612: and obtaining the state value of each neuron position based on the state values of eight adjacent neuron positions around the target point so as to obtain the state value of each AGV position.
S710: and when the state value of the target point and the state values of the AGV positions are not equal to 0, finishing constructing the target neural network so as to determine the target AGV and the corresponding path.
In one specific embodiment, the step S700 of determining a target AGV and a corresponding path includes:
comparing the state values of the AGV positions;
determining a target AGV according to the AGV with the largest state value;
and the path of the target AGV reaching the target point is the path.
Specifically, in the prior art, the GBNN neural network is constructed when the neural network reaches a stable state, but the GBNN neural network is used as a determination condition for completing the construction of the network, and it takes a long time to complete the calculation of the state value of each position neuron. Considering that the important concern in the GBNN network construction process is the comparison of the state values, which is mainly to find out which position state value is the largest, and the final result is not necessarily calculated, therefore, only when the state values of the positions where all AGVs are spread by the neural network are no longer 0, the construction of the neural network can be stopped, and the neural network does not need to be stopped until the neural network is stable, the state of the neural network at this time can still complete task allocation and path planning tasks, namely, the judgment condition for completing the construction of the network is improved accordingly, and a flow chart of the condition for judging the completion of the construction of the neural network after the improvement is shown in fig. 6, which is the state value of the position where the ith AGV is located, and is the number of AGVs in the area. With the construction of the GBNN neural network, the GBNN construction is ended when the state values of all AGVs are no longer 0.
An embodiment of the present specification provides a path planning method based on a neural network, and as shown in fig. 7, the method includes:
s120: acquiring regional information;
s220: dividing the region into a plurality of grids with consistent areas and shapes according to the region information;
s320: determining a neural network structure corresponding to the grids according to the grids, wherein each neuron in the neural network structure corresponds to each grid position one by one;
s420: determining a target point and positions of a plurality of AGVs from a plurality of neurons;
s520: constructing a target neural network corresponding to the target point;
s620: obtaining the state value of each AGV position according to the target neural network;
s720: when the state value of the target point and the state values of all the AGV positions are not equal to 0, finishing constructing the target neural network so as to determine a target AGV and a corresponding path;
s820: judging whether the target neural networks respectively corresponding to the other remaining target points are established;
s920: and if so, finishing the construction of the target neural networks respectively corresponding to all the target points so as to determine the target AGV corresponding to each target point and the corresponding path.
Specifically, the application scenario of the method is that a plurality of target points need to perform task allocation on the plurality of target points, that is, when the state values of the positions where the neural network spreads to all AGVs and other target points are no longer 0, the construction of the neural network can be stopped, the neural network does not need to be stopped until the neural network is stable, the state of the neural network at this time can still complete task allocation and path planning tasks, the determination condition for completing the construction of the network is improved accordingly, and a flow chart of the condition for determining the completion of the construction of the neural network after the improvement is shown in fig. 8, which is the state value of the position where the ith AGV is located, and is the number of AGVs in the area. Is the state value of the position of the jth target point. Is the number of target points in the area. With the construction of the GBNN neural network, the construction of the GBNN network is ended when the state values of all the AGVs and the target point positions are no longer 0.
When the neural network is constructed, the termination condition for constructing the neural network is changed, the learning time of the neural network is shortened, the time performance is better, and the problem that the time consumption is too long when the path planning is completed by constructing the neural network to reach the stable state in the prior art is solved.
An embodiment of the present specification provides a path planning apparatus based on a neural network, as shown in fig. 9, the apparatus includes:
the area information acquisition module 901: for obtaining area information;
grid region construction module 902: the grid is used for dividing the region into a plurality of grids with consistent areas and shapes according to the region information;
neural network structure determination module 903: the neural network structure is used for determining a neural network structure corresponding to a plurality of grids according to the grids, and each neuron in the neural network structure corresponds to each grid position one by one;
the target point determination module 904: the target point and the AGVs are determined from the neurons;
the neural network building block 905: the target neural network is used for constructing a target neural network corresponding to the target point;
AGV status value obtaining module 906: the AGV position detection device is used for obtaining a state value of each AGV position according to the target neural network;
the single task path plan determination module 907: and when the state values of the target point and the AGV positions are not equal to 0, finishing constructing the target neural network so as to determine a target AGV and a corresponding path.
An embodiment of the present specification provides a path planning apparatus based on a neural network, and as shown in fig. 10, the method includes:
the area information acquisition module 1001: for obtaining area information;
grid region building module 1002: the grid is used for dividing the region into a plurality of grids with consistent areas and shapes according to the region information;
the neural network structure determination module 1003: the neural network structure is used for determining a neural network structure corresponding to a plurality of grids according to the grids, and each neuron in the neural network structure corresponds to each grid position one by one;
the target point determination module 1004: the target point and the AGVs are determined from the neurons;
the neural network building module 1005: the target neural network is used for constructing a target neural network corresponding to the target point;
AGV status value obtaining module 1006: the AGV position detection device is used for obtaining a state value of each AGV position according to the target neural network;
single-task path planning determination module 1007: when the state values of the target point and each AGV position are not equal to 0, finishing constructing the target neural network so as to determine a target AGV and a corresponding path;
the target neural network determination module 1008: the target neural network is used for judging whether the target neural networks respectively corresponding to the other remaining target points are established;
multitask path plan determination module 1009: and when the target neural networks corresponding to the other remaining target points are established, the target neural networks corresponding to all the target points are established, so as to determine the target AGV corresponding to each target point and the corresponding path.
The single-task path planning determining module 1007 and the multi-task path planning determining module 1009 may be integrated into one path planning determining module, and may implement the functions of single-task path planning and multi-task path planning.
The present specification embodiments provide an apparatus comprising a processor and a memory, wherein the memory stores at least one instruction, at least one program, code set, or instruction set, and the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the neural network-based path planning method according to the above method embodiments.
Specifically, please refer to fig. 11 for a schematic structural diagram of a server device provided in an embodiment of the present specification. The server is used for implementing the neural network-based path planning method provided in the above embodiment. Specifically, the method comprises the following steps:
the server 2000 includes a Central Processing Unit (CPU)2001, a system memory 2004 including a Random Access Memory (RAM)2002 and a Read Only Memory (ROM)2003, and a system bus 2005 connecting the system memory 2004 and the central processing unit 2001. The server 2000 also includes a basic input/output system (I/O system) 2006 to facilitate transfer of information between devices within the computer, and a mass storage device 2007 to store an operating system 2013, application programs 2014, and other program modules 2015.
The basic input/output system 2006 includes a display 2008 for displaying information and an input device 2009 such as a mouse, keyboard, etc. for a user to input information. Wherein the display 2008 and the input devices 2009 are coupled to the central processing unit 2001 through an input-output controller 2010 coupled to the system bus 2005. The basic input/output system 2006 may also include an input/output controller 2010 for receiving and processing input from a number of other devices, such as a keyboard, mouse, or electronic stylus. Similarly, the input-output controller 2010 also provides output to a display screen, a printer, or other type of output device.
The mass storage device 2007 is connected to the central processing unit 2001 through a mass storage controller (not shown) connected to the system bus 2005. The mass storage device 2007 and its associated computer-readable media provide non-volatile storage for the server 2000. That is, the mass storage device 2007 may include a computer-readable medium (not shown) such as a hard disk or CD-ROM drive.
Without loss of generality, the computer-readable media may comprise computer storage media and communication media. Computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Computer storage media includes RAM, ROM, EPROM, EEPROM, flash memory or other solid state memory technology, CD-ROM, DVD, or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices. Of course, those skilled in the art will appreciate that the computer storage media is not limited to the foregoing. The system memory 2004 and mass storage device 2007 described above may be collectively referred to as memory.
The server 2000 may also operate as a remote computer connected to a network via a network, such as the internet, according to various embodiments of the present invention. That is, the server 2000 may be connected to the network 2012 through a network interface unit 2011 that is coupled to the system bus 2005, or the network interface unit 2011 may be utilized to connect to other types of networks or remote computer systems (not shown).
The memory also includes one or more programs stored in the memory and configured to be executed by one or more processors; the one or more programs include instructions for performing the method of the backend server side.
The embodiment of the present invention further provides a computer storage medium, where the computer storage medium may store a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the steps of the neural network processing method according to the embodiment of the present invention, and are not described herein again.
It should be noted that: the precedence order of the above embodiments of the present invention is only for description, and does not represent the merits of the embodiments. And specific embodiments thereof have been described above. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, as for the device and server embodiments, since they are substantially similar to the method embodiments, the description is simple, and the relevant points can be referred to the partial description of the method embodiments.
It will be understood by those skilled in the art that all or part of the steps for implementing the above embodiments may be implemented by hardware, or may be implemented by a program instructing relevant hardware, where the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, etc.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (9)
1. A method for neural network based path planning, the method comprising:
acquiring regional information;
dividing the region into a plurality of grids with consistent areas and shapes according to the region information;
determining a neural network structure corresponding to the grids according to the grids, wherein each neuron in the neural network structure corresponds to each grid position one by one;
determining a target point and positions of a plurality of AGVs from a plurality of neurons;
constructing a target neural network corresponding to the target point;
obtaining the state value of each AGV position according to the target neural network;
and when the state value of the target point and the state values of the AGV positions are not equal to 0, finishing constructing the target neural network so as to determine the target AGV and the corresponding path.
2. The neural network-based path planning method according to claim 1, wherein the grids include a free grid and a non-free grid, the region is divided into several grids with consistent areas and shapes according to the region information, and then the method comprises:
dividing the region into a free passing region and an obstacle region according to the region information;
determining a corresponding free grid according to the free passing area;
determining a corresponding non-free grid from the obstacle region.
3. The neural network-based path planning method of claim 1, wherein obtaining the state value of each AGV position according to the target neural network comprises:
obtaining state values of eight adjacent neuron positions around the target point according to the target neural network;
and obtaining the state value of each neuron position based on the state values of eight adjacent neuron positions around the target point so as to obtain the state value of each AGV position.
4. The neural network-based path planning method of claim 3, wherein said determining a target AGV and a corresponding path comprises:
comparing the state values of the AGV positions;
determining a target AGV according to the AGV with the largest state value;
and the path of the target AGV reaching the target point is the path corresponding to the target AGV.
5. The neural network-based path planning method according to claim 1, wherein when the state values of the target points and the AGV positions are not equal to 0, the construction of the target neural network is completed, and thereafter comprising:
judging whether the target neural networks respectively corresponding to the other remaining target points are established;
and if so, finishing the construction of the target neural networks respectively corresponding to all the target points so as to determine the target AGV corresponding to each target point and the corresponding path.
6. A neural network-based path planning apparatus, the apparatus comprising:
an area information acquisition module: for obtaining area information;
a grid region construction module: the grid is used for dividing the region into a plurality of grids with consistent areas and shapes according to the region information;
a neural network structure determination module: the neural network structure is used for determining a neural network structure corresponding to a plurality of grids according to the grids, and each neuron in the neural network structure corresponds to each grid position one by one;
a target point determination module: the target point and the AGVs are determined from the neurons;
a target neural network construction module: the target neural network is used for constructing a target neural network corresponding to the target point;
AGV status value gets module: the AGV position detection device is used for obtaining a state value of each AGV position according to the target neural network;
the single task path planning determination module: and when the state values of the target point and the AGV positions are not equal to 0, finishing constructing the target neural network so as to determine a target AGV and a corresponding path.
7. The neural network-based path planning apparatus of claim 6, wherein the apparatus further comprises:
a target neural network judgment module: the target neural network is used for judging whether the target neural networks respectively corresponding to the other remaining target points are established;
the multi-task path planning determination module: and when the target neural networks corresponding to the other remaining target points are established, the target neural networks corresponding to all the target points are established, so as to determine the target AGV corresponding to each target point and the corresponding path.
8. An apparatus comprising a processor and a memory, the memory having stored therein at least one instruction, at least one program, a set of codes, or a set of instructions, the at least one instruction, the at least one program, the set of codes, or the set of instructions being loaded and executed by the processor to implement a neural network-based path planning method as claimed in any one of claims 1 to 5.
9. A computer storage medium having stored therein at least one instruction, at least one program, set of codes, or set of instructions, which is loaded and executed by a processor to implement a neural network-based path planning method as claimed in any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110771368.8A CN113627646B (en) | 2021-07-08 | Path planning method, device, equipment and medium based on neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110771368.8A CN113627646B (en) | 2021-07-08 | Path planning method, device, equipment and medium based on neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113627646A true CN113627646A (en) | 2021-11-09 |
| CN113627646B CN113627646B (en) | 2024-07-02 |
Family
ID=
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118003342A (en) * | 2024-04-09 | 2024-05-10 | 北京科技大学 | Path planning method for space crawling mechanical arm |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109471446A (en) * | 2018-11-08 | 2019-03-15 | 淮阴师范学院 | A kind of complete coverage path planning method neural network based |
| CN110543171A (en) * | 2019-08-27 | 2019-12-06 | 华中科技大学 | Storage multi-AGV path planning method based on improved BP neural network |
| CN112015174A (en) * | 2020-07-10 | 2020-12-01 | 歌尔股份有限公司 | Multi-AGV motion planning method, device and system |
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109471446A (en) * | 2018-11-08 | 2019-03-15 | 淮阴师范学院 | A kind of complete coverage path planning method neural network based |
| CN110543171A (en) * | 2019-08-27 | 2019-12-06 | 华中科技大学 | Storage multi-AGV path planning method based on improved BP neural network |
| CN112015174A (en) * | 2020-07-10 | 2020-12-01 | 歌尔股份有限公司 | Multi-AGV motion planning method, device and system |
Non-Patent Citations (2)
| Title |
|---|
| 朱博;邓三鹏;王英飞;邢嘉珍;: "基于生物激励神经网络的移动机器人遍历路径规划", 装备制造技术, no. 12 * |
| 赵德云;杨厚华;王哲;: "基于模糊神经网络控制的AGV避障路径规划仿真", 机电工程, no. 09 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118003342A (en) * | 2024-04-09 | 2024-05-10 | 北京科技大学 | Path planning method for space crawling mechanical arm |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11970161B2 (en) | Apparatus, method and article to facilitate motion planning of an autonomous vehicle in an environment having dynamic objects | |
| CN107169608B (en) | Distribution method and device for multiple unmanned aerial vehicles to execute multiple tasks | |
| Smith et al. | Distributed inference-based multi-robot exploration | |
| JP7394853B2 (en) | Devices, methods and articles that facilitate motor planning in environments with moving objects | |
| Feyzabadi et al. | Risk-aware path planning using hirerachical constrained markov decision processes | |
| EP3627250B1 (en) | Method and system for free space detection in a cluttered environment | |
| WO2019141223A1 (en) | Path planning method and system for mobile robot | |
| KR20200068534A (en) | Operation management method, apparatus and system using machine learning and transfer learning | |
| Dharmasiri et al. | Novel implementation of multiple automated ground vehicles traffic real time control algorithm for warehouse operations: djikstra approach | |
| CN116523155A (en) | Convex relaxation-based full-coverage path planning method | |
| CN110222824B (en) | Intelligent algorithm model autonomous generation and evolution method, system and device | |
| CN115014328A (en) | Dynamic loading method, device, equipment and medium for grid map | |
| CN115461199A (en) | Task-oriented 3D reconstruction for autonomous robotic operation | |
| Dundar | Dynamic path finding method and obstacle avoidance for automated guided vehicle navigation in Industry 4.0 | |
| Zhao et al. | A multi-robot cooperative exploration algorithm considering working efficiency and working load | |
| Rottmann et al. | Adaptive autonomous control using online value iteration with gaussian processes | |
| Wang et al. | A partitioning-based approach for robot path planning problems | |
| CN113627646B (en) | Path planning method, device, equipment and medium based on neural network | |
| Zhang et al. | Reinforcement learning–based tool orientation optimization for five-axis machining | |
| CN113627646A (en) | Path planning method, device, equipment and medium based on neural network | |
| CN110749325A (en) | Flight path planning method and device | |
| Mukhopadhyay et al. | Multi-robot Map Exploration Based on Multiple Rapidly-Exploring Randomized Trees | |
| US20220300002A1 (en) | Methods and systems for path planning in a known environment | |
| Švogor et al. | Multi-criteria software component allocation on a heterogeneous platform | |
| CN113587931B (en) | Path planning method, device, equipment and storage medium |
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 |