CN113837049A - Intelligent road surface cleaning method based on convolutional neural network and genetic algorithm - Google Patents
Intelligent road surface cleaning method based on convolutional neural network and genetic algorithm Download PDFInfo
- Publication number
- CN113837049A CN113837049A CN202111090784.8A CN202111090784A CN113837049A CN 113837049 A CN113837049 A CN 113837049A CN 202111090784 A CN202111090784 A CN 202111090784A CN 113837049 A CN113837049 A CN 113837049A
- Authority
- CN
- China
- Prior art keywords
- road surface
- grid
- path
- garbage
- neural network
- 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
- 238000004140 cleaning Methods 0.000 title claims abstract description 65
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 51
- 238000000034 method Methods 0.000 title claims abstract description 49
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 47
- 230000002068 genetic effect Effects 0.000 title claims abstract description 43
- 238000009826 distribution Methods 0.000 claims abstract description 59
- 238000012549 training Methods 0.000 claims abstract description 19
- 240000007651 Rubus glaucus Species 0.000 claims abstract description 11
- 235000011034 Rubus glaucus Nutrition 0.000 claims abstract description 11
- 235000009122 Rubus idaeus Nutrition 0.000 claims abstract description 11
- 230000006870 function Effects 0.000 claims description 17
- 238000012360 testing method Methods 0.000 claims description 15
- 230000035772 mutation Effects 0.000 claims description 11
- 238000003860 storage Methods 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 8
- 238000010586 diagram Methods 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 6
- 230000004888 barrier function Effects 0.000 claims description 5
- 238000005457 optimization Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000005520 cutting process Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 claims description 3
- 230000009191 jumping Effects 0.000 claims description 3
- 238000002372 labelling Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000007639 printing Methods 0.000 claims description 3
- 238000010187 selection method Methods 0.000 claims description 3
- 238000013473 artificial intelligence Methods 0.000 abstract description 12
- 238000000605 extraction Methods 0.000 description 6
- 238000011176 pooling Methods 0.000 description 6
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 210000002569 neuron Anatomy 0.000 description 3
- 238000011160 research Methods 0.000 description 2
- 238000010408 sweeping Methods 0.000 description 2
- 241001417527 Pempheridae Species 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000000349 chromosome Anatomy 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000009396 hybridization Methods 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
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
- G05D1/02—Control of position or course in two dimensions
- G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
- G05D1/0212—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory
- G05D1/0221—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory involving a learning process
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
- G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Physiology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Genetics & Genomics (AREA)
- Aviation & Aerospace Engineering (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Automation & Control Theory (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an intelligent road surface cleaning method based on a convolutional neural network and a genetic algorithm, which belongs to the technical field of artificial intelligence, and aims to solve the technical problem of how to clean a street surface by using artificial intelligence to achieve the aim of cleaning the street surface without the artificial intelligence, wherein the adopted technical scheme is as follows: s1, constructing a pavement garbage distribution model through a convolutional neural network; s2, recognizing the garbage distribution condition on the road surface by using a binary classification method in the convolutional neural network, and training a road surface garbage distribution model; s3, planning a path through a genetic algorithm according to the distribution condition of the pavement garbage; s4, embedding the road surface garbage model and the path plan into the raspberry pi, embedding the raspberry pi and the unmanned camera into a mobile road surface cleaning machine, and moving the road surface cleaning machine to clean the road.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to an intelligent road surface cleaning method based on a convolutional neural network and a genetic algorithm.
Background
Neural networks (neural networks) are part of the field of artificial intelligence research, and the most popular neural network at present is deep Convolutional Neural Networks (CNNs), which are rarely used for reasons of accuracy, expressiveness, and the like, although they also have a shallow structure. Referring to CNNs and convolutional neural networks, there is no specific distinction between academic and industrial fields, and they generally refer to convolutional neural networks with deep structures, and the number of layers varies from "several layers" to "tens to hundreds". CNNs are currently enjoying tremendous success in many areas of research, such as speech recognition, image segmentation, natural language processing, etc. Although the problems addressed in these areas are not the same, these application methods can be generalized in that CNNs can automatically learn features from (usually large-scale) data and generalize the results to the same type of unknown data. The general steps for implementing the task of image recognition using CNN are: the input layer reads in a regularized (uniform size) image, each neuron of each layer takes a group of small local adjacent units of the previous layer as input, namely local receptive field and weight sharing, the neuron extracts some basic visual features such as edges, angular points and the like, and the features are used by the neurons of the higher layer. The convolutional neural network obtains a feature map by a convolution operation, and at each position, the units from different feature maps obtain different types of features respectively. A convolutional layer usually contains a plurality of feature maps with different weight vectors, so that richer features of the image can be retained. The back of the convolutional layer is connected with the pooling layer for down-sampling operation, so that the resolution of the image can be reduced, the parameter quantity can be reduced, and the robustness of translation and deformation can be obtained. The alternating distribution of the convolution layer and the pooling layer leads the number of the characteristic maps to be gradually increased and the resolution to be gradually reduced, thus the structure is a double pyramid.
Genetic Algorithm (GA) is a search Algorithm used in computational mathematics to solve optimization, and is one of evolutionary algorithms. Evolutionary algorithms were originally developed by using some phenomena in evolutionary biology, including inheritance, mutation, natural selection, and hybridization. Genetic algorithms are typically implemented as a computer simulation. For an optimization problem, a population of abstract representations (called chromosomes) of a certain number of candidate solutions (called individuals) evolves towards better solutions. Traditionally, solutions are represented in binary (i.e., strings of 0's and 1's), but other representations are possible. Evolution starts with a population of completely random individuals, followed by one generation. In each generation, fitness of the entire population is evaluated, a number of individuals are randomly selected from the current population (based on their fitness), and a new life population is created through natural selection and mutation, which becomes the current population in the next iteration of the algorithm.
Many existing lifestyles are mainly artificial, but the artificial intelligence technology is rapidly developed in recent years. The artificial intelligence needs more service life, so that people can enjoy higher life quality by saving time for life, for example, the time-consuming and labor-consuming work of cleaning the street pavement needs to be solved only by using the artificial intelligence.
Therefore, how to clean the street pavement by using artificial intelligence and achieve the aim of cleaning the street pavement without artificial intelligence is a technical problem to be solved urgently at present.
Disclosure of Invention
The invention provides an intelligent road surface cleaning method based on a convolutional neural network and a genetic algorithm, and aims to solve the problem of how to clean street surfaces by using artificial intelligence and realize self-service road surface cleaning.
The technical task of the invention is realized in the following way, the invention relates to an intelligent road surface cleaning method based on a convolutional neural network and a genetic algorithm, the method is characterized in that the garbage distribution condition on the road surface is identified through the convolutional neural network, the path planning is carried out through the genetic algorithm, and the road surface cleaning is carried out by utilizing a mobile road surface cleaning machine; the method comprises the following specific steps:
s1, constructing a pavement garbage distribution model through a convolutional neural network;
s2, recognizing the garbage distribution condition on the road surface by using a binary classification method in the convolutional neural network, and training a road surface garbage distribution model;
s3, planning a path through a genetic algorithm according to the distribution condition of the pavement garbage;
s4, embedding the road surface garbage model and the path plan into the raspberry pie, embedding the raspberry pie and the unmanned camera into a mobile road surface cleaning machine, and moving the road surface cleaning machine to clean the road surface so as to achieve the aim of cleaning the road surface.
Preferably, the road surface garbage distribution model in step S2 is trained as follows:
s201, constructing a road surface data set: dividing the training set into a training set and a testing set;
s202, training a pavement garbage distribution model: putting the garbage distribution model into a processed horizontal plane training set, and performing pavement garbage distribution model training through a server;
s203, testing a pavement garbage distribution model: putting the trained model into a test set for polarity test;
s204, optimizing a road surface garbage distribution model: and according to the test result of the pavement garbage distribution model, acquiring the pavement data set again to train the pavement garbage distribution model or finely adjust the pavement garbage distribution model until the model is fitted.
More preferably, the road surface data set is constructed in step S201 as follows:
s20101, collecting pictures: under different light environments, photographing and recording the road surface through an unmanned aerial vehicle to obtain a road surface picture;
s20102, labeling: observing whether the photos of the road surface contain the garbage or not by naked eyes, and printing a label on the pictures, wherein the content of the label is whether the garbage exists or not;
s20103, picture preprocessing: selectively cutting, rotating, amplifying or reducing the existing picture;
s20104, making a road surface data set: all the labeled photos are made into a road surface data set.
Preferably, the path planning by the genetic algorithm in step S3 is specifically as follows:
s301, establishing a map: constructing a grid map by adopting a grid algorithm;
s302, initializing a population: randomly generating a plurality of feasible paths, wherein the feasible paths represent paths which do not collide with the barrier grids;
s303, fitness function calculation: the path is optimized by iterating the fitness function.
Preferably, the grid map is a rectangular coordinate system established by taking the first grid at the lower left corner of the map as a coordinate origin, each grid is represented in the form of (x, y) coordinates, and each grid is numbered, and a path is represented by the grid number.
Preferably, the initialization population in step S302 is specifically as follows:
s30201, generating a discontinuous path: randomly taking out an obstacle-free grid from each row in sequence to form a discontinuous path; the first grid of the path and the last grid of the path are respectively a starting position and a target position;
s30202, connecting the discontinuous path to a continuous path, specifically as follows:
s3020201, determining whether two adjacent grids are continuous grids from the first grid:
if the grid is a continuous grid, executing the step S3020202;
s3020202, judging whether the new grid is an obstacle grid:
step S3020203 is executed if the new grid is an obstacle grid;
if the new grid is a barrier-free grid, jumping to the step S3020205;
s3020203, taking the adjacent grids of the new grid in the order of top, bottom, left and right, and judging whether the adjacent grids of the new grid are already in the path (preventing falling into a dead loop):
if the adjacent grid of the new grid is an obstacle-free grid and is not in the path, executing step S3020204;
s3020204, inserting into the path, and executing the step S3020207;
s3020205, inserting the two discontinuous grids, and executing a step S3020206;
s3020206, continuously judging whether the newly inserted grid and the previous grid of the newly inserted grid are continuous:
if not, looping step S3020201 to step S3020206 to search for a new grid until the two grids are continuous, and next executing step S3020207;
and S3020207, when two grids are continuous, taking the next grid, and looping the step S3020201 to the step S3020206 to search the next grid until the whole path is continuous.
Preferably, the path optimization by iterating the fitness function in step S303 is specifically as follows:
s30301, a selection method: calculating the sum of fitness functions of all individuals, calculating the ratio of each individual, and selecting the next generation of individuals in a roulette mode according to the probability of each individual;
s30302, a cross mode: determining a cross probability pc, generating a random number between 0 and 1, and comparing the random number with the cross probability pc; if the number of the detection signals is less than pc, performing cross operation; the method comprises the following specific steps:
finding out all the same points in the two paths, randomly selecting one of the points, and performing cross operation on the subsequent paths;
s30303, variation mode: determining a variation probability pm, generating a random number between 0 and 1, and comparing the random number with the variation probability pm; if the pm is smaller than pm, performing mutation operation; the method comprises the following specific steps:
randomly selecting two grids except the starting point and the end point in the path, removing the path between the two grids, and then taking the two grids as adjacent points to connect the discontinuous path into a continuous path for continuous operation by using the step S30202; when the continuous path can not be generated, two points are selected again to carry out the mutation operation until the mutation operation is completed.
Preferably, the road cleaning performed by moving the road surface cleaning device in step S4 is as follows:
s401, will clean the road surface machine and settle the suitable position at the road that needs cleaned, unmanned aerial vehicle carries out the whole shooting in road surface to spread into the raspberry with the photo and send into, judge whether there is rubbish through road surface rubbish distribution model:
firstly, if a garbage model exists, executing a step S402;
if no garbage exists, no treatment is carried out;
s402, the road surface garbage distribution model processes the processed result through a genetic algorithm to obtain a routing graph, and the routing graph is sent to a road surface cleaning machine;
s403, the road surface cleaning machine performs cleaning work according to the path planning diagram until the road surface is free of garbage;
and S404, after the cleaning work is finished, returning the road surface cleaning machine to the original position.
An electronic device, comprising: a memory and at least one processor;
wherein the memory has stored thereon a computer program;
the at least one processor executes the memory-stored computer program, causing the at least one processor to perform the intelligent road cleaning method based on a convolutional neural network and a genetic algorithm, as described above.
A computer-readable storage medium, in which a computer program is stored, the computer program being executable by a processor for implementing an intelligent road cleaning method based on a convolutional neural network and a genetic algorithm as described above.
The intelligent road surface cleaning method based on the convolutional neural network and the genetic algorithm has the following advantages:
the method comprises the steps of (I) respectively judging the distribution of road surface garbage and planning a path based on a convolutional neural network and a genetic algorithm in deep learning, so as to achieve the aim of cleaning the road surface;
the invention relates to two events, one is to confirm the distribution of the rubbish on the road surface and to process the specific route analysis of the rubbish after confirming the distribution of the rubbish; the two events respectively relate to two artificial intelligence techniques: convolutional neural networks and genetic algorithms; the method has the advantages that the garbage distribution on the road surface is identified through the convolutional neural network, then the route planning is carried out through the genetic algorithm, and finally the road cleaning is carried out through the machine until no garbage exists on the road surface, so that the labor intensity is reduced, the labor force is saved, and the road surface cleaning efficiency is improved.
Drawings
The invention is further described below with reference to the accompanying drawings.
FIG. 1 is a flow diagram of an intelligent road cleaning method based on a convolutional neural network and a genetic algorithm;
FIG. 2 is a schematic diagram of a pavement refuse distribution model;
FIG. 3 is a block diagram of a process for path planning by genetic algorithm;
FIG. 4 is a block diagram of a process for moving a road sweeper to sweep a road.
Detailed Description
The intelligent road surface cleaning method based on the convolutional neural network and the genetic algorithm is described in detail below with reference to the drawings and specific embodiments.
Example 1:
as shown in the attached figure 1, the intelligent road surface cleaning method based on the convolutional neural network and the genetic algorithm identifies the garbage distribution condition on the road surface through the convolutional neural network, carries out path planning through the genetic algorithm and carries out road surface cleaning by utilizing a mobile road surface cleaning machine; the method comprises the following specific steps:
s1, constructing a pavement garbage distribution model through a convolutional neural network; as shown in fig. 2, the road surface garbage distribution model includes an input unit, a feature extraction unit and a feature output unit; the characteristic extraction unit comprises a convolution layer, a pooling layer, a convolution layer, a pooling layer and a full-communication layer; the road surface picture input by the input unit is processed by the feature extraction unit, feature extraction is sequentially carried out on the road surface picture through the convolution layer, the pooling layer, the convolution layer, the pooling layer and the full-communication layer, and after feature extraction, the feature extraction unit outputs the features through the feature output unit.
S2, recognizing the garbage distribution condition on the road surface by using a binary classification method in the convolutional neural network, and training a road surface garbage distribution model;
s3, planning a path through a genetic algorithm according to the distribution condition of the pavement garbage;
s4, embedding the road surface garbage model and the path plan into the raspberry pie, embedding the raspberry pie and the unmanned camera into a mobile road surface cleaning machine, and moving the road surface cleaning machine to clean the road surface so as to achieve the aim of cleaning the road surface.
The road surface garbage distribution model in step S2 of this embodiment is trained as follows:
s201, constructing a road surface data set: dividing the training set into a training set and a testing set;
s202, training a pavement garbage distribution model: putting the garbage distribution model into a processed horizontal plane training set, and performing pavement garbage distribution model training through a server;
s203, testing a pavement garbage distribution model: putting the trained model into a test set for polarity test;
s204, optimizing a road surface garbage distribution model: and according to the test result of the pavement garbage distribution model, acquiring the pavement data set again to train the pavement garbage distribution model or finely adjust the pavement garbage distribution model until the model is fitted.
The road surface data set is constructed in step S201 of this embodiment specifically as follows:
s20101, collecting pictures: under different light environments, photographing and recording the road surface through an unmanned aerial vehicle to obtain a road surface picture;
s20102, labeling: observing whether the photos of the road surface contain the garbage or not by naked eyes, and printing a label on the pictures, wherein the content of the label is whether the garbage exists or not;
s20103, picture preprocessing: selectively cutting, rotating, amplifying or reducing the existing picture;
s20104, making a road surface data set: all the labeled photos are made into a road surface data set.
As shown in fig. 3, the path planning by the genetic algorithm in step S3 in this embodiment is specifically as follows:
s301, establishing a map: constructing a grid map by adopting a grid algorithm;
s302, initializing a population: randomly generating a plurality of feasible paths, wherein the feasible paths represent paths which do not collide with the barrier grids;
s303, fitness function calculation: the path is optimized by iterating the fitness function.
The smaller the grid area is, the more accurate the representation of various environment information in the space is, but the storage space occupied at the same time is increased, and the searching time used by the algorithm is increased. If the grid area is larger, various environmental information in the space cannot be accurately represented, and the collision problem is easy to occur. Therefore, when the environment model is established, the following rules need to be made: the problem of the height of the barrier is not considered, and the walking space of the robot is a two-dimensional plane space; the size, position of the obstacle are known, and no dynamic obstacle exists; the robot can be regarded as particle processing during planning. To facilitate the grid area, we usually represent the walking space of the robot by squares. If the area of the obstacle is smaller than that of the square, the obstacle can be enlarged into the square; if the area of the obstacle is larger than the area of the square, the area of the obstacle can be represented by two squares; if the obstacle area is larger, it can be represented by a plurality of squares. The grid map is characterized in that a rectangular coordinate system is established by taking the first grid at the lower left corner of the map as a coordinate origin, each grid is represented in the form of (x, y) coordinates, numbers are compiled for each grid, and a path is represented by grid numbers.
The initialization population in step S302 of this embodiment is specifically as follows:
s30201, generating a discontinuous path: randomly taking out an obstacle-free grid from each row in sequence to form a discontinuous path; the first grid of the path and the last grid of the path are respectively a starting position and a target position;
s30202, connecting the discontinuous path to a continuous path, specifically as follows:
s3020201, determining whether two adjacent grids are continuous grids from the first grid:
if the grid is a continuous grid, executing the step S3020202;
s3020202, judging whether the new grid is an obstacle grid:
step S3020203 is executed if the new grid is an obstacle grid;
if the new grid is a barrier-free grid, jumping to the step S3020205;
s3020203, taking the adjacent grids of the new grid in the order of top, bottom, left and right, and judging whether the adjacent grids of the new grid are already in the path (preventing falling into a dead loop):
if the adjacent grid of the new grid is an obstacle-free grid and is not in the path, executing step S3020204;
s3020204, inserting into the path, and executing the step S3020207;
s3020205, inserting the two discontinuous grids, and executing a step S3020206;
s3020206, continuously judging whether the newly inserted grid and the previous grid of the newly inserted grid are continuous:
if not, looping step S3020201 to step S3020206 to search for a new grid until the two grids are continuous, and next executing step S3020207;
and S3020207, when two grids are continuous, taking the next grid, and looping the step S3020201 to the step S3020206 to search the next grid until the whole path is continuous.
For example: the start position is grid 0 and the target position is grid 99. Initializing the population requires randomly generating a plurality of feasible paths, which represent paths that do not collide with the barrier grid.
In this embodiment, the path optimization through the iterative fitness function in step S303 is specifically as follows:
s30301, a selection method: the sum of fitness functions of all individuals is calculated, the ratio of each individual is calculated, next generation individuals are selected in a roulette mode according to the probability of each individual, partial non-optimal individuals are guaranteed in the roulette mode, and the algorithm can be effectively prevented from falling into a local optimal solution. The fitness function is divided into two parts which are respectively used for judging the length and the smoothness degree of a path. Due to the constraints of kinematics and dynamics, the robot is not suitable for turning too much during traveling, and a relatively smooth path is beneficial to the traveling of the robot, so that the generated path has the requirement of smoothness.
S30302, a cross mode: determining a cross probability pc, generating a random number between 0 and 1, and comparing the random number with the cross probability pc; if the number of the detection signals is less than pc, performing cross operation; the method comprises the following specific steps:
finding out all the same points in the two paths, randomly selecting one of the points, and performing cross operation on the subsequent paths;
s30303, variation mode: determining a variation probability pm, generating a random number between 0 and 1, and comparing the random number with the variation probability pm; if the pm is smaller than pm, performing mutation operation; the method comprises the following specific steps:
randomly selecting two grids except the starting point and the end point in the path, removing the path between the two grids, and then taking the two grids as adjacent points to connect the discontinuous path into a continuous path for continuous operation by using the step S30202; when the continuous path can not be generated, two points are selected again to carry out the mutation operation until the mutation operation is completed.
As shown in fig. 4, the road sweeping operation performed by the mobile road-sweeping machine in step S4 of the present embodiment is as follows:
s401, will clean the road surface machine and settle the suitable position at the road that needs cleaned, unmanned aerial vehicle carries out the whole shooting in road surface to spread into the raspberry with the photo and send into, judge whether there is rubbish through road surface rubbish distribution model:
firstly, if a garbage model exists, executing a step S402;
if no garbage exists, no treatment is carried out;
s402, the road surface garbage distribution model processes the processed result through a genetic algorithm to obtain a routing graph, and the routing graph is sent to a road surface cleaning machine;
s403, the road surface cleaning machine performs cleaning work according to the path planning diagram until the road surface is free of garbage;
and S404, after the cleaning work is finished, returning the road surface cleaning machine to the original position.
Example 2:
an embodiment of the present invention further provides an electronic device, including: a memory and a processor;
wherein the memory stores computer execution instructions;
the processor executes the computer-executable instructions stored in the memory, so that the processor executes the intelligent road surface cleaning method based on the convolutional neural network and the genetic algorithm in any embodiment of the invention.
Example 3:
embodiments of the present invention further provide a computer-readable storage medium, in which a plurality of instructions are stored, and the instructions are loaded by a processor, so that the processor executes the intelligent road surface cleaning method based on the convolutional neural network and the genetic algorithm in any embodiment of the present invention. Specifically, a system or an apparatus equipped with a storage medium on which software program codes that realize the functions of any of the above-described embodiments are stored may be provided, and a computer (or a CPU or MPU) of the system or the apparatus is caused to read out and execute the program codes stored in the storage medium.
In this case, the program code itself read from the storage medium can realize the functions of any of the above-described embodiments, and thus the program code and the storage medium storing the program code constitute a part of the present invention.
Examples of the storage medium for supplying the program code include a floppy disk, a hard disk, a magneto-optical disk, an optical disk (e.g., CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RYM, DVD-RW, DVD + RW), a magnetic tape, a nonvolatile memory card, and a ROM. Alternatively, the program code may be downloaded from a server computer via a communications network.
Further, it should be clear that the functions of any one of the above-described embodiments may be implemented not only by executing the program code read out by the computer, but also by causing an operating system or the like operating on the computer to perform a part or all of the actual operations based on instructions of the program code.
Further, it is to be understood that the program code read out from the storage medium is written to a memory provided in an expansion board inserted into the computer or to a memory provided in an expansion unit connected to the computer, and then causes a CPU or the like mounted on the expansion board or the expansion unit to perform part or all of the actual operations based on instructions of the program code, thereby realizing the functions of any of the above-described embodiments.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. An intelligent road surface cleaning method based on a convolutional neural network and a genetic algorithm is characterized in that the method is to identify the garbage distribution condition on the road surface through the convolutional neural network, plan the path through the genetic algorithm and clean the road surface by using a mobile road surface cleaning machine; the method comprises the following specific steps:
s1, constructing a pavement garbage distribution model through a convolutional neural network;
s2, recognizing the garbage distribution condition on the road surface by using a binary classification method in the convolutional neural network, and training a road surface garbage distribution model;
s3, planning a path through a genetic algorithm according to the distribution condition of the pavement garbage;
s4, embedding the road surface garbage model and the path plan into the raspberry pie, embedding the raspberry pie and the unmanned camera into a mobile road surface cleaning machine, and moving the road surface cleaning machine to clean the road surface so as to achieve the aim of cleaning the road surface.
2. An intelligent road cleaning method based on a convolutional neural network and a genetic algorithm as claimed in claim 1, wherein the road garbage distribution model in step S2 is trained as follows:
s201, constructing a road surface data set: dividing the training set into a training set and a testing set;
s202, training a pavement garbage distribution model: putting the garbage distribution model into a processed horizontal plane training set, and performing pavement garbage distribution model training through a server;
s203, testing a pavement garbage distribution model: putting the trained model into a test set for polarity test;
s204, optimizing a road surface garbage distribution model: and according to the test result of the pavement garbage distribution model, acquiring the pavement data set again to train the pavement garbage distribution model or finely adjust the pavement garbage distribution model until the model is fitted.
3. An intelligent road surface cleaning method based on a convolutional neural network and a genetic algorithm as claimed in claim 2, wherein the road surface data set is constructed in step S201 as follows:
s20101, collecting pictures: under different light environments, photographing and recording the road surface through an unmanned aerial vehicle to obtain a road surface picture;
s20102, labeling: observing whether the photos of the road surface contain the garbage or not by naked eyes, and printing a label on the pictures, wherein the content of the label is whether the garbage exists or not;
s20103, picture preprocessing: selectively cutting, rotating, amplifying or reducing the existing picture;
s20104, making a road surface data set: all the labeled photos are made into a road surface data set.
4. An intelligent road surface cleaning method based on a convolutional neural network and a genetic algorithm as claimed in claim 1, wherein the path planning through the genetic algorithm in step S3 is specifically as follows:
s301, establishing a map: constructing a grid map by adopting a grid algorithm;
s302, initializing a population: randomly generating a plurality of feasible paths, wherein the feasible paths represent paths which do not collide with the barrier grids;
s303, fitness function calculation: the path is optimized by iterating the fitness function.
5. An intelligent road surface cleaning method based on a convolutional neural network and a genetic algorithm as claimed in claim 4, characterized in that the grid map is a rectangular coordinate system established by taking the first grid at the lower left corner of the map as the origin of coordinates, each grid is represented by the coordinate form of (x, y), and each grid is numbered, and a path is represented by the grid number.
6. An intelligent road surface cleaning method based on a convolutional neural network and a genetic algorithm as claimed in claim 4, wherein the initialization population in step S302 is specifically as follows:
s30201, generating a discontinuous path: randomly taking out an obstacle-free grid from each row in sequence to form a discontinuous path; the first grid of the path and the last grid of the path are respectively a starting position and a target position;
s30202, connecting the discontinuous path to a continuous path, specifically as follows:
s3020201, determining whether two adjacent grids are continuous grids from the first grid:
if the grid is a continuous grid, executing the step S3020202;
s3020202, judging whether the new grid is an obstacle grid:
step S3020203 is executed if the new grid is an obstacle grid;
if the new grid is a barrier-free grid, jumping to the step S3020205;
s3020203, sequentially fetching the adjacent grids of the new grid in the upper, lower, left and right directions, and judging whether the adjacent grids of the new grid are in the path:
if the adjacent grid of the new grid is an obstacle-free grid and is not in the path, executing step S3020204;
s3020204, inserting into the path, and executing the step S3020207;
s3020205, inserting the two discontinuous grids, and executing a step S3020206;
s3020206, continuously judging whether the newly inserted grid and the previous grid of the newly inserted grid are continuous:
if not, looping step S3020201 to step S3020206 to search for a new grid until the two grids are continuous, and next executing step S3020207;
and S3020207, when two grids are continuous, taking the next grid, and looping the step S3020201 to the step S3020206 to search the next grid until the whole path is continuous.
7. An intelligent road cleaning method based on a convolutional neural network and a genetic algorithm as claimed in claim 4 or 5 or 6, wherein the path optimization through the iterative fitness function in step S303 is specifically as follows:
s30301, a selection method: calculating the sum of fitness functions of all individuals, calculating the ratio of each individual, and selecting the next generation of individuals in a roulette mode according to the probability of each individual;
s30302, a cross mode: determining a cross probability pc, generating a random number between 0 and 1, and comparing the random number with the cross probability pc; if the number of the detection signals is less than pc, performing cross operation; the method comprises the following specific steps:
finding out all the same points in the two paths, randomly selecting one of the points, and performing cross operation on the subsequent paths;
s30303, variation mode: determining a variation probability pm, generating a random number between 0 and 1, and comparing the random number with the variation probability pm; if the pm is smaller than pm, performing mutation operation; the method comprises the following specific steps:
randomly selecting two grids except the starting point and the end point in the path, removing the path between the two grids, and then taking the two grids as adjacent points to connect the discontinuous path into a continuous path for continuous operation by using the step S30202; when the continuous path can not be generated, two points are selected again to carry out the mutation operation until the mutation operation is completed.
8. An intelligent road cleaning method based on the convolutional neural network and the genetic algorithm as claimed in claim 1, wherein the mobile road cleaning machine in step S4 is specifically as follows:
s401, will clean the road surface machine and settle the suitable position at the road that needs cleaned, unmanned aerial vehicle carries out the whole shooting in road surface to spread into the raspberry with the photo and send into, judge whether there is rubbish through road surface rubbish distribution model:
firstly, if a garbage model exists, executing a step S402;
if no garbage exists, no treatment is carried out;
s402, the road surface garbage distribution model processes the processed result through a genetic algorithm to obtain a routing graph, and the routing graph is sent to a road surface cleaning machine;
s403, the road surface cleaning machine performs cleaning work according to the path planning diagram until the road surface is free of garbage;
and S404, after the cleaning work is finished, returning the road surface cleaning machine to the original position.
9. An electronic device, comprising: a memory and at least one processor;
wherein the memory has stored thereon a computer program;
the at least one processor executing the memory stored computer program causes the at least one processor to perform the intelligent road surface cleaning method based on convolutional neural network and genetic algorithm as set forth in any one of claims 1 to 8.
10. A computer-readable storage medium, in which a computer program is stored, the computer program being executable by a processor to implement the intelligent road surface cleaning method based on a convolutional neural network and a genetic algorithm as set forth in any one of claims 1 to 8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111090784.8A CN113837049B (en) | 2021-09-17 | 2021-09-17 | Intelligent pavement cleaning method based on convolutional neural network and genetic algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111090784.8A CN113837049B (en) | 2021-09-17 | 2021-09-17 | Intelligent pavement cleaning method based on convolutional neural network and genetic algorithm |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113837049A true CN113837049A (en) | 2021-12-24 |
CN113837049B CN113837049B (en) | 2023-06-20 |
Family
ID=78959715
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111090784.8A Active CN113837049B (en) | 2021-09-17 | 2021-09-17 | Intelligent pavement cleaning method based on convolutional neural network and genetic algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113837049B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116681195A (en) * | 2023-06-06 | 2023-09-01 | 深圳启示智能科技有限公司 | Robot road-finding device based on artificial intelligence |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10242264B1 (en) * | 2018-04-25 | 2019-03-26 | Imagry (Israel) Ltd. | System and method for training a machine-learning model to identify real-world elements |
CN110097139A (en) * | 2019-05-13 | 2019-08-06 | 济南浪潮高新科技投资发展有限公司 | A kind of intelligence rice washing method and device based on convolutional neural networks |
CN111985316A (en) * | 2020-07-10 | 2020-11-24 | 上海富洁科技有限公司 | Road surface garbage sensing method for intelligent road cleaning |
-
2021
- 2021-09-17 CN CN202111090784.8A patent/CN113837049B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10242264B1 (en) * | 2018-04-25 | 2019-03-26 | Imagry (Israel) Ltd. | System and method for training a machine-learning model to identify real-world elements |
CN110097139A (en) * | 2019-05-13 | 2019-08-06 | 济南浪潮高新科技投资发展有限公司 | A kind of intelligence rice washing method and device based on convolutional neural networks |
CN111985316A (en) * | 2020-07-10 | 2020-11-24 | 上海富洁科技有限公司 | Road surface garbage sensing method for intelligent road cleaning |
Non-Patent Citations (1)
Title |
---|
孙干余: "基于深度学习技术的巡检视频图像智能理解算法研究", 《中国优秀硕士论文全文数据库》 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116681195A (en) * | 2023-06-06 | 2023-09-01 | 深圳启示智能科技有限公司 | Robot road-finding device based on artificial intelligence |
Also Published As
Publication number | Publication date |
---|---|
CN113837049B (en) | 2023-06-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Jeong et al. | Image preprocessing for efficient training of YOLO deep learning networks | |
CN111444821B (en) | Automatic identification method for urban road signs | |
CN109446970B (en) | Transformer substation inspection robot road scene recognition method based on deep learning | |
CN109785385B (en) | Visual target tracking method and system | |
CN113110482B (en) | Indoor environment robot exploration method and system based on priori information heuristic method | |
JP6980289B2 (en) | Learning method and learning device that can detect lanes using a lane model, and test method and test device using this {LEARNING METHOD, LEARNING DEVICE FOR DETECTING LANE USING LANE MODEL AND TEST METHOD, TEST DEVICE | |
US6668084B1 (en) | Image recognition method | |
CN113110513A (en) | ROS-based household arrangement mobile robot | |
CN111931703B (en) | Object detection method based on human-object interaction weak supervision label | |
CN112766170B (en) | Self-adaptive segmentation detection method and device based on cluster unmanned aerial vehicle image | |
CN114511077A (en) | Training point cloud processing neural networks using pseudo-element based data augmentation | |
CN113837049A (en) | Intelligent road surface cleaning method based on convolutional neural network and genetic algorithm | |
CN113485373A (en) | Robot real-time motion planning method based on Gaussian mixture model | |
CN113284228B (en) | Indoor scene room layout dividing method based on point cloud | |
Zhao et al. | Boundary regularized building footprint extraction from satellite images using deep neural network | |
Hwang et al. | Object Detection for Cargo Unloading System Based on Fuzzy C Means. | |
CN112428271B (en) | Robot real-time motion planning method based on multi-mode information feature tree | |
CN115454096B (en) | Course reinforcement learning-based robot strategy training system and training method | |
CN116541701A (en) | Training data generation method, intelligent body training device and electronic equipment | |
CN115809751A (en) | Two-stage multi-robot environment coverage method and system based on reinforcement learning | |
WO2022064610A1 (en) | Object detection device, trained model generation method, and recording medium | |
CN115258509A (en) | Method and device for selecting items and computer readable storage medium | |
CN109215049B (en) | Roof segmentation method, system and equipment based on multi-scale three-dimensional prior information | |
CN116863509B (en) | Method for detecting human-shaped outline and recognizing gesture by using improved polar mask | |
US20240028784A1 (en) | Segmenting a building scene |
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 |