JP2000172664A - Optimum route and optimum circulation route searching method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To search in a short time an optimum route through all designated via points with little calculation quantity. SOLUTION: A prescribed number of routes (individuals), which are expressed in expanded genetic types, from a start S to a goal G are prepared as an initial individual group and by repeatedly applying a genetic algorithm to that individual group expressed with the expanded genetic types, the optimum route from the start through all designated via points A and B to the goal is searched. In this case, the weight is determined corresponding to the number of designated via the points included in the individual (route), and the inverse of a value multiplying the weight to a distance from the start of that individual to the goal is defined as the evaluation function of the genetic algorithm.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a genetic algorithm
The present invention relates to an optimal route search method for searching for an optimal route using (GA: Genetic Algorithm), and particularly to an optimal route search method for searching for an optimal route from a departure point to a destination through all designated via points with a small amount of computation, Also, the present invention relates to an optimal traveling route searching method for searching for an optimal traveling route that passes through all designated transit points with the same departure point and destination, or an optimal traveling route that passes through a maximum number of transit points with a specified upper limit cost or less.

[0002]

2. Description of the Related Art A vehicle navigation system is provided with a route guidance function for guiding a vehicle from a departure place to a destination. According to the route guidance function, when a destination is set, an optimum route (for example, a route of the shortest distance) from the departure place (current vehicle position) to the destination is searched, and the searched shortest route is displayed on a monitor screen. Guide the driver to the destination. Conventionally, as a search method that can be used for such a shortest path problem, a depth-first search (dfs) method, which is a type of full search method,
A breadth-first search (bfs) method is known. However, they have the disadvantage that the search takes time. Therefore,
A Dijkstra method, which is a kind of breadth-first search method, has been proposed. Since the Dijkstra method has a short search time, many methods that apply this method have been proposed and become mainstream.

However, in the conventional route search method such as the Dijkstra method, it is difficult to select a plurality of route candidates having the shortest point-to-point distance (the shortest route and a route similar thereto) by a single search. For this reason, if the driver is not satisfied with the selected shortest route, there is a problem that a new route search must be performed from the beginning to find the next candidate. Further, in the conventional route search method, there is a problem that even if the route search is performed again from the beginning, a route having a shorter second rank is not always selected. Therefore, the applicant of the present application searched for an optimal route using a genetic algorithm (GA) in Japanese Patent Application No. 9-285154 (filing date: October 17, 1997, title of the invention: optimal route searching method). An optimal route search method has been proposed in which the shortest route and a route similar thereto can be simultaneously searched in a single route search (first proposed method).

On the other hand, there is known an optimal route search when arriving at a destination through a plurality of transit points. In the conventional route search that passes through multiple waypoints, the user inputs the order of the waypoints and the final destination, and the navigation device searches for the shortest route from the starting point to the first waypoint using the Dijkstra method, Then, the shortest route from the first stopover to the second stopover is searched for, and by following the same procedure, the optimum route that reaches the destination through a plurality of stopovers is searched for. However, in this method, the order of the route points is fixed, and there is no guarantee that the obtained result is optimal for all routes. In particular, in the case of efficiently traveling around a nearby sightseeing spot, it is actually difficult to input in advance the order of travel around an unfamiliar sightseeing spot. Therefore, the present inventors,
Using a genetic algorithm (GA), we have proposed an optimal route search method when arriving at a destination through multiple transit points from a departure point (the second proposed method, IEICE December 1997 IE)
Reported at the study group, Technical report of the Institute of Telecommunications, CAS97-6, VLD
97-6, DSP97-21: Refer to the paper title "Consideration on route search with multiple via designation").

[0005]

However, in the second proposed method, in the search process, all but the individual expressing the route passing through all the waypoints are used as lethal genes. For this reason, at the time of generation or crossover of the initial population, in order to repeat the process of individual generation until an individual that is not a lethal gene is generated,
Requires a lot of computation. In addition to the optimal route search from the departure point to the destination through a plurality of designated transit points, the so-called optimal route search of a so-called tour route returning to the departure point over several designated transit points (for example, sightseeing spots). Is also an indispensable function in terms of effective use of the navigation device. or,
If it is not possible to traverse all designated transit points at the upper limit cost (for example, the maximum mileage), it is also effective to use the navigation device to search for an optimal patrol route that returns to the departure point through the maximum number of designated transit points within the upper limit cost. It is an indispensable function in terms of. Therefore, a route search method that realizes these functions is demanded.

As described above, an object of the present invention is to provide an optimal traveling route search method capable of searching for an optimal route passing through all designated via points in a short time with a small amount of calculation. It is another object of the present invention to determine the weights to be an optimal value, so that an optimal solution can be obtained without falling into a local solution. It is another object of the present invention to provide an optimal traveling route searching method capable of searching for an optimal traveling route passing through all designated route points. It is another object of the present invention to provide an optimal route search method capable of searching for an optimal route that returns to a departure place after passing through a maximum number of designated transit points with an upper limit cost or less.

[0007]

When searching for a route from a first point (departure point) to a second point (destination) through N designated waypoints, all of the real nodes The smaller of the maximum number M of connected nodes and (N + 1) obtained by adding 1 to the number of designated via points is set as the number k of virtual nodes, and k virtual nodes are prepared to enable a return path for each real node. , An extended genotype having a virtual node corresponding to an all-passing real node from the first point to the second point as a gene and having a structure capable of performing crossover processing in a genetic algorithm, and designing the extended genotype; A predetermined number of routes (individuals) from the first point to the second point represented by are prepared as an initial population, and a genetic algorithm is repeatedly applied to the population having the expanded genotype structure, By genetic algorithm Searching for the optimum route from the first point which passes all over point to the second point. In this case, a weight is determined according to the designated number of via points included in the individual (route), and the reciprocal of a value obtained by multiplying the distance from the first point to the second point of the individual by the weight is determined by a genetic algorithm. And the evaluation function in. In this way,
The optimal route search can be efficiently performed by utilizing the individual that has been the lethal gene, the amount of calculation can be reduced, and the optimal route can be searched in a short time. In addition, the weight is initially set to be small, and the shortest path length L _{n of the} path including n designated via points obtained each time the genetic algorithm is applied, and the shortest path length L _{n of the} path including (n-1) designated via points comparing the magnitude of the path length _{L n-1 · α n-} 1 for the shortest route length L _n-1 multiplied by the weight α _n-1, if _{L n> L n-1 ·} α a _n-1, The weight is increased by a predetermined ratio, and control is performed so that the weight becomes an optimum value. As described above, since the weight is controlled to be the optimum value, the optimum solution can be obtained with extremely high probability without falling into the local solution.

In the above-described optimal route search, when searching for an optimal circuit route in which the first point and the second point are the same point (departure point), (N-1) designated routes from the departure point An extended genotype having, as a gene, a virtual node corresponding to an all-passing real node of the route to the last waypoint through the point,
A predetermined number of traveling routes (individuals) expressed in combination with genotypes having, as genes, all the passing actual nodes of the route from the last designated waypoint to the departure point as a gene are prepared as an initial population. A genetic algorithm is repeatedly applied to search for an optimal traveling route that returns to the starting point through all designated waypoints.

In addition, a predetermined number of the first
, A predetermined number of second populations passing through (N-1) designated transit points, a predetermined number of third populations passing through (N-2) designated transit points, ... Prepared, the genetic circuit is repeatedly applied to each population, and the optimal circuit path that passes through all the waypoints, the optimal circuit path that passes through (N-1) waypoints, and (N-2) .. Are determined, and a search is made for a route that passes through the maximum number of route points with the specified upper limit cost or less.

[0010]

(A) Outline of the present invention (a) First configuration of a navigation device to which the present invention can be applied FIG. 1 is a schematic configuration diagram of a navigation device for explaining an optimum route search according to the present invention. is there. In the figure, reference numeral 1 denotes a map information storage medium such as a CD-ROM or a DVD for storing map information;
Reference numeral 2 denotes a vehicle position detection unit for detecting the position of the own vehicle, which is composed of a self-contained navigation sensor (vehicle speed sensor, direction sensor, etc.) and GPS. 3 is a destination and waypoint setting, various instructions by menu item selection, A remote control unit for performing operations such as enlargement / reduction of the map, 4 is a route search unit in the navigation control unit, 5 is a map and an optimal route, for example, an optimal route MSP from the departure point S to the destination through designated via points A and B. Is a monitor that displays.

When the route search unit 4 inputs a departure place S, a destination G, and waypoints A and B and instructs a route search, the route search unit 4 starts from the departure place S based on the genetic algorithm (GA). The search for the optimum route to the destination G through A and B is started, and the obtained optimum route MSP is displayed on the monitor 5. That is, the route search unit 4 prepares a predetermined number of routes (individuals) from the departure point to the destination represented by the extended genotype as an initial population, and genetically prepares a population having the extended genotype structure. The algorithm is repeatedly applied, and the genetic algorithm searches for the optimal route from the starting point to the destination that passes through all the waypoints. In this case, a weight is determined according to the designated via points included in the individual (route), and the reciprocal of a value obtained by multiplying the distance from the departure place to the destination of the individual by the weight is used as an evaluation function in the genetic algorithm. I do. The weight is initially set to be small, and the shortest path length L _n of a path including n designated via points obtained each time the genetic algorithm is applied, and the shortest path length of a path including (n-1) designated via points comparing the magnitude of the path length _{L n-1 · α n-} 1 to the length L _n-1 multiplied by the weight α _n-1, if _{L n> L n-1 ·} α a _n-1, the weight Control is performed so that the weight increases by a predetermined ratio and the weight becomes an optimum value.

As described above, the route search method of the present invention uses the temporary building block (DEGoldberg, "Generic Algorithms
in Search Optimization and Machine Learning "Addis
on. Wesley, 1989), which is a search method that introduces a weight depending on the number of via points of an individual (route) into an evaluation function. By controlling the weights, it is possible to efficiently search without losing effective skimmers (plural forms of schemata) contained in an individual that has been a lethal gene, and to reduce the amount of calculation. Skimata means "sequence", and GA means a meaningful pattern in chromosome or arrangement of genes.

FIG. 2 is a schematic block diagram of a navigation device for explaining the optimal traveling route search according to the present invention.
The same reference numerals are given to the same parts, and the monitor 5 displays a map and an optimal circuit route, for example, an optimal circuit route MSRP that returns from the departure point S to the departure point through the designated waypoints A, B, and C.
Is displayed. When the departure point S and the waypoints A, B, and C are input and the traveling route search is instructed, the route search unit 4 sends the designated waypoints A, B, and the like from the departure point S based on the genetic algorithm (GA). A search for the optimal route MSRP which returns to the departure point through C is started, and the obtained optimal route MSRP is obtained.
Is displayed on the monitor 5. That is, when searching for the optimal circuit route, the route search unit 4 determines (N-1)
The extended genotype structure of the route from the specified (N = 3 in the above example) waypoint to the last Nth waypoint, and the last N
A predetermined number of tour routes (individuals) expressed by combining the genotype structures of the routes from the designated route point to the departure point are prepared as an initial population, and the genetic algorithm is repeatedly applied to the population to optimize the population. Search for the route MSRP.

When the upper limit cost is specified, the route search unit 4 determines that the predetermined number of first individuals passing through all the designated waypoints and the predetermined number of (N-1) designated waypoints passing through the designated waypoints A number of second populations, a predetermined number of third populations passing through (N-2) designated waypoints,... Are prepared, and the genetic algorithm is repeatedly applied to each population to obtain a total. Optimal circuit path passing through waypoints, optimal circuit path passing through (N-1) waypoints, optimal circuit path passing through (N-2) waypoints,
··· is searched for a route that passes through the maximum number of waypoints with the specified upper limit cost or less.

(B) Outline of Genetic Algorithm FIG. 3 is a schematic flowchart of the genetic algorithm.
FIG. 3 is an explanatory diagram of a general individual expression in a genetic algorithm. In the genetic algorithm GA, the trait of each individual (search point) is represented as a chromosome (see FIG. 4).
The chromosome is composed of genes, and each gene represents a partial trait of an individual. The chromosome of each individual needs to be expressed so that each GA operation described later is effectively performed. Further, it is necessary that all solution candidates can be expressed in order to make all solution candidates in the search range, and it is desirable that all individuals that can be represented by chromosomes are solution candidates in order to prevent redundant search.

In the genetic algorithm GA, an initial population of individuals is created using random numbers (step 10).
1). However, when there is some prior knowledge, an individual considered to have high fitness is created as an initial group.
Genes that are not present in the initial population only arise from mutations. For this reason, the method of creating the initial group may greatly affect the search efficiency. Next, the fitness of the individual is evaluated (step 102). Fitness is an evaluation of the trait represented by the chromosome of each individual. One of the features of the GA is that the fitness evaluation function can be arbitrarily determined. In addition, the setting of the evaluation function for making a selection using this fitness greatly affects the efficiency of the search. Once the fitness is determined for each individual, the individual that is the basis for the next generation of individuals is selected from the set of search points (initially the initial population) according to the individual fitness and the progress of the search (selection, Step 103). Depending on the selection method, there may be a problem of initial convergence in which a specific gene spreads rapidly and is occupied by the same individual who is not a local solution.

Thereafter, two parent individuals are selected at a predetermined occurrence frequency from a population of a plurality of individuals selected by the selection operation, and chromosomes are rearranged to form child chromosomes (crossover, step 104). It is important that new individuals created by crossover inherit the traits of the underlying parents. After the crossover process, the gene is changed with a certain probability by a mutation operation (mutation, step 105).
If mutations occur too frequently, they are randomly searched. However, the creation of a chromosome other than the combination of genes in the initial population requires a change in the gene due to mutation. The above steps 102 to 105 are repeated until the termination condition is satisfied (step 106), and when the condition is satisfied, the search process is terminated.

The above is an outline of a general genetic algorithm GA. To apply the genetic algorithm GA to an optimal route search, the following items (various rules and parameters) must be modified or modified in advance for the route search. It needs to be realized. If these parameters are not properly selected, problems such as a long calculation time required to converge to an optimal solution and a problem of falling into a local solution and failing to obtain the optimal solution occur. -Individual trait expression-Initial population creation method-Fitness evaluation function-Selective selection method-Crossover method-Mutation method-Search termination condition-Number of individuals

(C) Design of genotype applicable to search for shortest path between two points The following genotypes are used for the purpose of suppressing the amount of calculation at the time of crossover, suppressing lethal genes, and maintaining the diversity of individuals of the next generation. Designed. FIG. 5 is an explanatory diagram of a structure (genotype structure) expressing a path as an individual, (a) is map information, (b)
FIG. 7 (c) is an example of a genotype having a route 0 → 5 → 9 → 10 → 4 from a starting point 0 to a destination 4 as a phenotype. This genotype has a passing node number as a gene. This satisfies the crossover condition described below,
It is a genotype of equal length using path expression. Specifically, the path from the node number m to the node number n (hereinafter, referred to as path m → n) is represented by storing n in the m-th locus, and the path from the start point to the arrival point This process is repeated, and for a node that does not pass through even once, one is randomly selected from the set of nodes to which the node is connected and stored. For example, FIG.
If the map information shown in FIG.
An example of a genotype having a phenotype of → 5 → 9 → 10 → 4 is
As shown in FIG. 5C, 0 1 2 3 4 5 6 7 8 9 10 ... Node numbers ↓ ↓ ↓ ↓... Arrows indicate routes [5 2 6 8 3 9 7 8 5 10 4]・ ・ It becomes a locus. The upper row is the node number, and the lower row is the locus.

It can be confirmed that 5 is stored at the 0th locus, 9 is stored at the 5th locus, 10 is stored at the 9th locus, and 4 is stored at the 10th locus. The gene at the locus indicated by the node number without underline stores a randomly selected one from the nodes that can go next from the gene (node) of the node number,
It can be seen that different genotypes having the same phenotype are generated depending on how to select this node. In the genotypes designed as described above, the phenotype and the genotype correspond one-to-many.

(B) Search for Shortest Path Between Two Points (First Proposed Method) FIG. 6 is a configuration diagram for executing the search for the shortest path between two points based on a genetic algorithm GA. It should be noted that the configuration for executing “the shortest route search with designation of a plurality of via points”, “optimal tour route search through all designated via points”, and “optimum tour route search when an upper limit cost is set” described later is also illustrated in FIG. 6, but the processing method is different. In the figure, reference numeral 11 denotes an initial group creation unit for creating an initial group of routes (individuals) from a departure point to a destination calculated by a computer, and 12 denotes a fitness calculation for calculating fitness of each individual based on an evaluation function. The unit 13 is a search end determination unit, and when the fitness is no longer changed,
Alternatively, when the number of searches reaches the set number, it is determined that the search has ended. Reference numeral 14 outputs the individual searched by the end of the search, that is, the route having the highest fitness (for example, the shortest route), or if necessary. A shortest path output unit that outputs the second and third rank paths together with the shortest path, 15 is a selection and selection processing unit that selects an individual with high fitness, and 16 replaces the gene (node) of the selected individual (parent). The crossover processing unit 17 is a mutation processing unit that mutates an individual gene (node) according to a predetermined rule.

(A) Initial group creation unit In the shortest path search between two points, as shown in FIG. 5, a road intersection (node) on a map is used as a gene, and a one-dimensional array at the locus is treated as a chromosome. As the number of individuals increases, the number of search points increases, so that the diversity of solution candidates increases, and good results are obtained, but the search time increases. As a result of the experiment, it was confirmed that stable search was not performed with the number of individuals of about 40 or less. Therefore, in this experiment, the number of individuals was 5
Set to 0. As described above, the initial group creation unit 11 creates an initial group consisting of 50 individuals.

(B) Fitness Calculation Unit The fitness calculation unit 12 calculates the fitness of each individual by using an evaluation function represented by the following equation: Fitness = 1 / (Σrlength (i)) i = 1 to M (1) calculate. Where rlength (i) is the number of nodes that pass through M from the starting point to the arriving point.
The path length is between the i-th node and the (i-1) -th node. In the shortest path problem, it is necessary to set an evaluation function in which the fitness becomes higher as the path length is shorter, and is defined by the reciprocal of the sum of the path lengths, as in Equation (1).

(C) Search Completion Judgment Unit The conditions for terminating the search are that the maximum fitness of the individual in each generation is not improved, for example, for 50 consecutive generations, and that the search has reached 1000 generations. Note that only the search termination condition may be used. FIG. 7 is a flowchart of a search end determination process by the search end determination unit 13. When the fitness of each individual is received from the fitness calculating unit 12 (step 301), i is incremented (i + 1 → i, step 302), and then the maximum fitness of the individual in the current generation is higher than the previous maximum fitness. It is checked whether it is large, in other words, whether the searched route has been improved in terms of distance (step 3).
03).

If not improved, the count value c (initial value is 0) is counted up by 1 (step 304).
Next, it is checked whether c ≧ 50 (step 305).
If c ≧ 50, the search processing is terminated assuming that the individual search for the shortest route has been completed (step 306). c <
In the case of 50, it is checked whether i = 1000, in other words, whether the search has reached 1000 generations (step 307). If i <1000, step 30
1 and the subsequent processing is repeated. If i = 1000,
The search processing ends (step 306). On the other hand, if it is determined in step 303 that the search route has been improved in terms of distance, the count value c is cleared (step 308), and then the processing in step 307 and subsequent steps is executed.

(D) Selective Selection Processing Unit Selective selection is an elite preservation strategy that leaves the individual with the best evaluation value (fitness) unconditionally to the next generation, and other individuals with a probability proportional to the evaluation value. To adopt a roulette strategy. That is, the selection / selection processing unit 15 performs the selection / selection processing by combining the elite preservation strategy with the general fitness proportional strategy in 60% of individuals having high fitness. However, if maximum fitness is reduced,
Instead of adding the individual with the highest fitness in the previous generation as a new individual, replacing the individual with the lowest fitness prevents an increase in the number of individuals. This is because a stable search may not be performed if the value is 60% or less as a result of an experiment performed when the value is 60%. FIG. 8 is an explanatory diagram of the selection, crossover, and mutation processes. The selection / selection processing unit 15 (FIG. 6) takes in N individuals constituting the individual group, and assigns individual 1 (I ₁ ), individual 2 (I ₂ ),. N
(I _N ) and the number of individuals (= N). Of the individuals arranged in this way, only 60% of the individuals with the highest fitness are selected, and the remaining 40% of the individuals are selected.

The crossover processing unit 16 and the mutation processing unit 17
Using the selected and remaining individuals, a new individual is created by performing crossover and mutation described later, and the total number of individuals is set to N. Here, the use frequency of the original individual used for crossover is selected with a probability proportional to the fitness (roulette strategy).
Next, fitness is calculated by the fitness calculator 12 for all N individuals of the new population obtained by performing the predetermined crossover and mutation, and each individual 1 (I ′ ₁ ), Individual 2 (I ′ ₂ ),..., Individual N
(I ′ _N ) and the number of individuals (= N). Then, the magnitude of the fitness is compared between the individual 1 (I ′ ₁ ) having the highest fitness in the current population and the individual 1 (I ₁ ) having the highest fitness in the previous population. I do. If I ₁ > I ′ ₁ , that is, if the current maximum fitness I ′ ₁ is smaller than the previous maximum fitness I _{1, the} individual 60 (I _'60 ) is discarded, and instead, the individual 1 (I ₁ ) having the highest fitness in the previous population is inserted as an individual in the new population. This ensures that the maximum fitness of the new population is always greater than or equal to the maximum fitness of the original population. Thus, a new population is created.
By repeating this operation, the genetic algorithm is executed.

FIG. 9 is a processing flow of the selection process. The selection / selection processing unit 15 fetches N individuals constituting the individual group (step 401), and assigns individual 1 (I ₁ ), individual 2 (I ₂ ),. N
( _IN ) and the number of individuals (= N) are rearranged (step 40).
2). Next, of the individuals arranged in this way, only those having a total fitness of 60% are selected in descending order of fitness, and the remaining 40% of the individuals are eliminated (step 403). The population selected in this manner is used as a next generation parent individual according to a predetermined frequency of occurrence. In addition, since the selection can be performed by the roulette strategy, 100% selection can be performed in the selection and selection unit. In practice, the selection is set to 100%.

(E) Crossover processing section The general crossover includes one-point crossover, multipoint crossover, uniform crossover, and the like. The crossover processing unit 16 executes a crossover process based on one of the uniform crossovers. Uniform crossover is
It is used to create a new individual from two separate individuals, and creates another individual whose chromosome contains exactly the same gene as either parent to inherit the properties of each parent individual. FIG. 10 is an explanatory diagram of uniform crossover. In the figure, P1 and P2 are individuals (parent 1 and parent 2) to be subjected to uniform crossover processing, MK is a mask used for uniform crossover processing,
CH is a new individual (child) obtained by uniform crossover. In the portion where 1 stands in the mask MK, the gene of the parent individual P2 is inherited, and in the portion of 0, the gene of the parent individual P1 is inherited, and a new child individual CH is created.

The mask MK used for uniform crossing is not always the same, and is changed using a random number.
In the method of determining a parent individual, a parent is randomly selected at an occurrence frequency corresponding to the fitness. For simplicity, the number of individuals is 3, and the fitness of each individual is a for individual 1, b for individual 2, and c for individual 3. The probability that each individual is selected at this time (the overall probability is 1) is: individual 1: a / (a + b + c) individual 2: b / (a + b + c) individual 3: c / It is represented by (a + b + c).

FIG. 11 is an explanatory diagram for uniformly crossing over individuals represented by genotypes. Crossover is an operation that generates one child individual from two parent individuals. In this method, a route search is performed, and after the crossover, the obtained child individual must be in a set of routes starting and ending with the designated starting point and reaching point. Therefore, the following genotypes were prepared, and the following crossover method based on uniform crossover was considered. First, from two individuals (parent 1 and parent 2) selected by the roulette strategy, one of the node numbers at the locus of the starting node number is randomly selected, and the same gene of the child is selected. Store in Zodiac. Randomly select one of the node numbers at the two parent loci with the same number as the stored node number, store it in the same loci of the child, repeat this until reaching the destination point, via For those nodes that did not, either one of the genes of the parent individual is randomly selected and stored.

The crossover method will be specifically described with reference to FIG. 11 (node number 0: starting point, node number 4: destination point). First, 5 at the 0th locus of parent 1 and 1 of parent 2
Of the parent 1 and 6 of the parent 2 at the locus of the node number 5 according to the selected 5.
Select one of the two. This is repeated until the arrival point (node number 4) is reached. Nodes that did not go through (nodes not enclosed in squares in the child of FIG. 11)
Is randomly selected from the gene of the parent individual.

FIG. 12 is a flowchart of the uniform crossover process, and the crossover process is executed in cooperation with the mutation process described later. First, a mask MK is created using random numbers (step 501), and then parent individuals P1 and P2 are randomly selected from the selected and selected individuals at an occurrence frequency corresponding to the fitness (step 502). Thereafter, a uniform crossover process is performed using the parent individuals P1 and P2 and the mask MK to execute the child individuals C
H is created (step 503). After the offspring is generated, the offspring is subjected to a mutation process as required (step 50).
4). Next, it is checked whether a required number of child individuals have been generated (step 505). If the child individuals have been generated, the crossover processing is terminated. If the required number of child individuals has not been generated, the process returns to step 501 and the above processing is repeated.

(F) Mutation processing section Mutation is performed not only in accordance with the mutation probability, but also in the case where an individual having the exact same gene already exists in the same generation. In other words, in the early stage of the search, mutation occurs only at a rate almost in accordance with the mutation probability while the solution diversity is high, but when the diversity decreases in the latter half of the search, many mutations occur and the solution diversity increases. Nature will be maintained. A large number of experiments were performed on the mutation probability, and it was confirmed that the search results did not significantly change in the range of 3% to 20%. Therefore, it was set to 5%.

FIG. 13 is an explanatory diagram of the mutation process. At the initial stage of the search, the offspring populations rarely have exactly the same genetic structure (chromosomes). In such a case, the mutation is performed according to a predetermined mutation probability. For example, 5 mutations are performed for 100 crossovers. FIG. 14 is a flowchart of the mutation processing by the mutation processing unit 17, and details the mutation processing step (step 504) in FIG. Two parent individuals P
1, P2 are selected, and a uniform crossover process is executed using the parent individuals P1, P2 and the mask to create child individuals (steps 502, 503). After the generation of the child individual, it is checked whether another child individual having the same genetic structure as the child individual exists (step 504a). If it exists, the gene (node) of the new child individual is changed and mutated (step 5).
04b).

On the other hand, if there is no separate child individual having the same gene structure in step 504a, mutation is performed according to the mutation probability. For example, 5 mutations are performed for 100 crossovers as described above. Therefore, if "NO" in step 504a, it is checked whether or not to make a mutation based on the mutation probability (step 504c).
The process of step 04b is executed to change and mutate the gene (node) of the child individual, and if it is not to be mutated, the mutation process is terminated. Thereafter, the fitness calculation process, the search end determination process, the selection process, the crossover process, and the mutation process are repeated for the newly generated population of N individuals.

(G) Simulation Results Genetic algorithm (G
A simulation experiment of the optimal route search between two points using A) was performed. The number of intersections (number of nodes) is 72. or,
The number of individuals was 50, the termination condition was 50 generations non-evolution, and the mutation probability was 5%. As a result of conducting an experiment using the above parameters, the shortest path MSP shown in FIG. 15 was obtained. This is the shortest path between the starting point S and the arriving point D, which coincides with the optimum path obtained by the Dijkstra method.

To search and display only the shortest route,
A conventional method represented by the Dijkstra method is also possible.
However, in the conventional method, only the optimum route is obtained, and the second and third-rank routes according to the optimum route cannot be obtained at the same time. For this reason, if another route that is different from the shortest route and does not have a large difference in distance is requested,
In the conventional method, the parameter must be changed and the search must be restarted from the beginning. Further, there is no guarantee that the second and third rank routes can be reliably obtained even when the search is restarted from the beginning. On the other hand, in the optimal route search method using GA, many solution candidates are generated in the process of the optimal route search. For this reason, by storing some of these solution candidates, it is possible to search and output the top m solutions (which can be made as large as the memory allows) by one optimal route search process. FIGS. 16 (a) and (b) show the second- and third-ranked individuals (routes) MSP1 and MSP2 simultaneously obtained during the search for the optimal route MSP shown in FIG.
Is shown. As described above, according to the optimal route search by the GA, the n-th
You can search for individuals up to the rank.

(C) Shortest Route Search with Specifying Multiple Via Points (Second Proposed Method) The route search method for searching for the shortest route between two points has been described above. Will be described. In the conventional method for finding the shortest route passing through a plurality of designated waypoints, it is necessary to search for the shortest route between the starting point and each waypoint, between the waypoints, and between each waypoint and the arrival point. Therefore, assuming that the total number of designated transit points is N, a maximum of N (N + 3) / 2 searches are required. Further, using the obtained shortest paths and path lengths, the path lengths of all permutations (N! Streets) from the starting point through all the waypoints to the arrival point are examined, and the one having the shortest path length is determined. Is the solution. For the above reasons, the conventional method has a problem that a large number of calculation points are required when the number of designated waypoints increases.

The genetic algorithm GA generates individuals as many solution candidates in the search process. Of these, generation switching is performed using only individuals passing through all the waypoints, thereby making it possible to search for the shortest route that passes through all the points specified in one search. However, the shortest path that passes through a waypoint may pass through the same node a plurality of times and cannot be represented by the genotype shown in FIG.
Therefore, an extended genotype is designed by adding the following changes to the genotype.

(A) Design of extended genotype Consider a route that passes the same node k times. Assuming that k nodes exist virtually at the position of one node,
A different node number is given to all the virtual nodes. FIG.
FIG. 7 shows an example in the case of k = 2. Virtual nodes 0, 1, 3 and 4 actually exist at the positions of real nodes a and b, respectively. The genotype in FIG. 5 (c) cannot encode a path that passes through the same node multiple times. Therefore, the route S → a → b → a → G using the actual node number cannot be expressed. However, the route S → 0 → 3 → 1 using the virtual node number
G has no nodes that pass multiple times and can be coded. In this way, the shortest path can be obtained by performing GA processing using the genotype coded using the virtual node number and returning to the real node number when generating a phenotype from the obtained individual .

When the number k of virtual nodes increases, the gene length increases and the amount of calculation increases. Therefore, it is necessary to find the minimum necessary k, but the value of k depends on the number of nodes to which one node is connected and the number of waypoints.
Since a path that makes two or more round trips of a path connecting any two nodes is not the shortest path, considering the group of nodes as shown in FIG. 18, the number of times of passing through the point c is determined by the path S → c → d →
It becomes the maximum in the case of c → e → c → f → c → S and is equal to the number of nodes connected to the point c (d, c, f are not limited to the designated waypoints). The number of connected nodes is checked for all nodes included in the search range, and the maximum value is set to M. Focusing on an arbitrary node, the maximum number of passes through the node is (the number of designated via points + 1) times when the vehicle goes to all designated via points and when it reaches the arrival point. For example, when all designated via points are arranged as points d, e, and f in FIG. 18, the number of times passing point c is determined by the route S → c → d →
It becomes the maximum in the case of c → e → c → f → c → S, and if the designated number of via points is N, it is (N + 1) times. This value does not change even if there is another node connected to the point c.

Therefore, the value of the number k of virtual nodes is M, (N +
It is sufficient to adopt the smaller one of 1), and k = min {M, (N + 1)} (2). For example, the map information (M
= 5), if the departure point is node 0, the destination is node 4, and the designated waypoint is node 9 (N = 1), the number of virtual nodes is k = 2, and the route 0 → 1 → 9 → 1 →
An example of an expanded genotype having a phenotype of 2 → 3 → 4 is shown in FIG.
9 (b). The upper part is a gene expressing a virtual node corresponding to each real node, and the lower part is a value of a gene existing at each locus. A genotype using k virtual nodes is advantageous in that a route search allowing k times of passage for the same real node can be performed without changing the GA processing itself.

(B) Simulation Results The three route points A,
B and C are designated, and the shortest path that passes through all three points is searched. In this map data, M = 4 and N
Since k = 3, k = 4 from equation (2). Four virtual nodes for one real node are prepared, and an extended genotype is designed. FIG. 2 shows the results of an experiment in which the number of individuals was 50, the termination condition was 50 generations non-evolution, and the mutation probability was 5%.
The shortest path MSP shown as 0 was obtained. From the figure, it can be seen that the shortest route that passes through all three designated waypoints A, B, and C has been obtained. Further, the route passing through the points B and C has nodes that pass in an overlapping manner, but the shortest route having such a route can be searched.

As described above, it has become possible to search for the shortest route that passes through all of a plurality of designated transit points in a single process, which is difficult with the conventional method. In addition, it is possible to encode a route having a node that passes a plurality of times by designing a genotype in which a virtual node is introduced. In addition,
In order to perform GA processing only on individuals passing through all the transit points in the process of GA processing, individuals are randomly generated, and those other than individuals passing through all the transit points are killed as lethal genes. For this reason, there is a problem that the amount of calculation increases.

(D) Optimal route search through all designated via points according to the present invention (a) Shortest route search by evaluation function introducing weight (a-1) Problem of second proposed method The second proposed method is as follows. As described above, among many solution candidates generated in the process of the GA, the search is performed using only the individual having a route that passes through all the designated waypoints. FIG. 21 shows a result of applying the second proposed method to map information of Iwaki-shi according to the second proposed method in order to compare with the present invention. Figure 2 shows the parameters used.
As shown in FIG. Points A and B in the figure are via points specified in advance, and the shortest route passing through these two points is obtained. The calculation time for this search is based on Sun SuperS
It is 34.41 seconds with parc 85MHz + GCC (-O3). Of these, the generation of the initial population takes 14.19 seconds, and this method has a problem that a large amount of computation is required for processing involving the generation of individuals.

(A-2) Route Searching Method of the Present Invention The problem of the second proposed method is that, when an initial population is generated or crossed, all the individuals other than the individuals passing through all the set waypoints are regarded as lethal genes. This is because the process of ontogenesis is repeated until an individual that is not a lethal gene is born. However, individuals having a lethal gene include individuals having effective skimmers (for example, individuals passing through some of the designated waypoints). We propose a search method based on the building block hypothesis in order to perform an efficient search using such an effective skimmer in the search and reduce the amount of calculation. The building block hypothesis is a hypothesis that, in the GA processing process, skimmers with high fitness are combined like a building block to generate individuals with higher fitness and perform efficient searches. In many problems, the formation of building blocks is one of the indications that GA can search efficiently.

However, it is generally difficult to set an evaluation function that forms a building block because it is unknown which skimmer is a building block. However, since the present invention searches for a route that passes through a set waypoint, the skimmer that expresses the passage of the waypoint is clearly a part of the building block. Therefore, by giving higher fitness to an individual having many waypoints on the route, the selection probability of a skimmer expressing passage of a waypoint is increased, and a search using this as a building block becomes possible. Specifically, the following equation Fitness = 1 / (Σrlength (i) × αn) i = 1 to M (3) is used as the number of evaluation intervals used in the genetic algorithm. That is, the weight α _n is determined according to the designated number of via points n included in the individual (route), and the reciprocal of a value obtained by multiplying the distance from the starting point to the destination of the route by the weight α _n is used as an evaluation function in the genetic algorithm. And Where α _n
Is a weight assigned according to the number of passing points n, and is set such that 1 ≦ α _N <α _N−1 <... <Α ₀ where N is the total number of passing points.

By introducing this evaluation function and applying the present invention to the map information of Iwaki City, the same route as in FIG. 21 could be obtained. Each parameter used, selection selection,
The method of crossover and mutation is the same as the second proposed method.
The magnitudes of the weights were α ₀ = 3.0, α ₁ = 2.0, and α ₂ = 1.0.
The calculation time required for the search of the present invention is 3.
This is 40 seconds, a 90.1% reduction compared to the second proposed method, and is effective.

(B) Method of Determining Weight As described above, determining the number of evaluations makes it possible to greatly reduce the amount of calculation. However, by appropriately setting the value of α _n , It is possible to obtain a route that does not pass or to reduce the probability of an erroneous search that falls into a local solution even when passing through all the waypoints, thereby enabling a better search. Theoretically, if the weight is set so that (the shortest path length in the case of n via points) <(the shortest path length in the case of n-1 via points x α _n-1 ) (4), the optimal solution can be obtained. can get. However, since the shortest path length is unknown before the search, the value of the weight cannot be determined from this equation alone. In the following, some simulations are performed with different weights,
From the results obtained, the cause of the erroneous search and the method of setting an appropriate weight value are considered.

(B-1) The magnitude of the weight and the ability to search for the optimum solution α is changed in the range of 1 to 100 for the map information used in the route search shown in FIG. Each search is performed 100 times, and the number of times the optimum solution is obtained is checked. Where α _{0 to} α ₂ are α ₀ = α, α ₁ = (1 + α) / 2, α
₂ = 1. Figure 2 shows the relationship between α and the probability of obtaining the optimal solution.
3 is shown. The dotted line in the figure satisfies 3.0 ≦ α ≦ 10.5 in the range of α where the search probability of the optimal solution exceeds 95%. As shown in the figure, when α increases, the search probability of the optimal solution rapidly increases and then decreases again. The cause of the erroneous search in each of the range of α <3.0 and the range of α> 10.5 is described below.

(B-2) Cause of Local Solution When α is Small FIG. 24 shows an example of a search result when α ₀ = 1.5, α ₁ = 1.25, and α ₂ = 1.0. As can be seen from this figure, a route is obtained that passes through only one point (point A) out of two preset waypoints. This is because α is small, so that (the shortest path length passing through two points)> (the shortest path length passing through one point x α ₁ ), the optimal function can be obtained with the evaluation function using the above weights. Can not.

When α is large If α is made sufficiently large, the evaluation function will not be such that an optimal solution cannot be obtained logically, but the probability of falling into a local solution will increase. FIG. 25 shows an example of a local solution when a search is performed with α = 100.0, α ₀ = α, α ₁ = (1 + α) / 2, and α ₂ = 1. From FIG. 25, it can be seen that although the vehicle has passed all of the designated transit points A and B, it has fallen into a local solution. The reason for the local solution is that the diversity of the population is lost in the search stage. To verify this, the change in the variance of the actual path length of the population due to the generation change is examined. The result is shown in FIG.

The solid line in the figure shows the variance when the optimal solution is obtained at α = 3.0, and the dotted line shows the variance when the local solution falls at α = 100.0. From FIG. 26, it can be seen that when α = 100.0, the variance is small at an early stage of the search, and the diversity of the population is rapidly lost. Consider the reasons for the loss of diversity. In the present invention, the initial population is randomly generated, and individuals having many transit points have a lower occurrence probability than individuals having few transit points. If the setting of α is too large, there will be a large difference in fitness depending on the number of via points. For this reason, the selection probability of a small individual having many via points increases. Therefore, the population is occupied by the same individual, and the diversity of the population is lost. When the diversity of the population is small, the types of skimmers that can be building blocks are reduced, and the probability that an optimum building block is formed decreases. Furthermore, since the search solution converges to a local solution at an early stage, the process of searching for an optimum solution is only a mutation. However, even if an individual having an effective skimmer is generated by mutation, it is highly likely that the individual will be eliminated due to a large difference in fitness, and it is difficult to escape from a local solution.

On the other hand, when α is optimally set, no extreme bias occurs in the selection probabilities of individuals. As a result, diversity in the search process is maintained, and effective skimmers in the individual remain as building blocks, so that an optimal solution can be obtained. FIG. 27 and FIG. 28 show an example of how building blocks are formed when α = 3.0 and α = 100.0. The dotted line in the figure indicates the number of individuals having the via point A, the broken line indicates the number of individuals having the via point B, and the solid line indicates the change in the number of individuals having both of them. From the figure, it can be seen that when an appropriate α (= 3.0) is set, as the generation change progresses, effective skimmers are combined to form a building block. Conversely, if α is too large, it can be seen that no building block is formed.

(B-3) Procedure for Determining Weight As described above, the value of α is determined within a range where a route passing through all the waypoints is selected in order to maintain the diversity of the population and reduce the probability of falling into a local solution. , May be set as small as possible.
That is, the initial value of α is set small, and each individual is observed in the search process (the path length of the best individual with the number of via points n)> (the number of via points n
When an individual having the best individual path length of −1 × α _n−1 ) appears, the value of α may be increased. FIG. 29 is a flowchart of the weight determination process. At the initial stage, the value of the weight α is set smaller (step 551), the genetic algorithm is applied to the i-th generation (step 552), and it is checked whether the termination condition is satisfied (step 553). If the termination condition is not satisfied, n
Seeking the shortest path length L _n-1 through the shortest path length L _n and (n-1) pieces of waypoints through the number of all _{waypoints, L n <L n-1} · α n-1 is (5) It is checked whether the condition is satisfied (step 554). If the condition is satisfied, i is incremented without changing the weight (step 555), and the processing from step 552 is repeated. Step 5
At 54, if equation (5) does not hold, the weight is increased by 25%.
(Step 556) Steps i and 552 are performed by incrementing i. If the termination condition is satisfied in step 553, the process ends.

(B-4) Example of Realizing Weight Determination An example of realizing the above-described weight setting method using specific numerical values will be described. In this implementation example, a value that satisfies the constraint on the weight α is set using the distance between the points, and is applied to the map information in FIG. First, the linear distances SA, AB, BG, AG, SG between the starting point S, the waypoints A, B, and the arriving point G are checked. Using this, the linear distance SG between the starting point and the destination point, the shortest linear distance SAG via one point (= SA + AG), the shortest linear distance SABG via two points (= SA + AB + BG)
And search is started with α ₀ = (SA + AB + BG) / SG and α ₁ = (SA + AG) / SG as initial values. In the search process, if an individual with (path length of the best individual with the number of via points n)> (path length of the best individual with the number of via points n-1 × α _n-1 ) appears, α ₀ is 25% of the initial value. increments, alpha ₁ updates such that _{α 1 = (α 0 -1)} × ( initial values of / alpha ₀ of alpha ₁₎ +1. As a result of the search, an optimal solution was obtained. The search time is 3.49 seconds, 89.compared to the second proposed method.
9% reduction. In addition, when the search was performed 100 times, the weight values at the end of the search were all within the range shown by the dotted line in FIG. 23, and the number of times the optimal solution was obtained was 99 times. Therefore, it can be said that this weight determination method is effective.

(E) Searching for optimal traveling route through all designated waypoints The method of searching for the shortest route passing through all the designated waypoints when the departure point and the destination are different has been described above. A description will be given of a shortest route search that passes through all designated waypoints when the destination is the same. (A) Design of genotype The extended genotype of FIG. 19B can be expressed only when the starting point and the ending point are different. For this reason, it is necessary to design a genotype when the starting point and the destination are the same. In the present invention, the circuit route is a first route from the departure point to the last N-th way point after passing through (N-1) designated waypoints, and a departure point from the last designated waypoint. The route is divided into a second route, each route is expressed by a genotype, and a cyclic route (individual) is expressed by combining the genotypes. That is, FIG. 19 (b) having, as a gene, a virtual node corresponding to all the passing real nodes of the first route from the departure place to the last Nth waypoint through (N-1) designated waypoints ) and extended genotype {G _1} of the two last all-pass real node of the second path leading to departure from the specified route point genotypes shown in FIG. 5 (c) with a gene {G _2} A traveling route (individual) is represented by a connected genotype {G ₁ {G ₂ }}.

(B) Simulation Results Three route points A,
B and C are designated, and the shortest route that goes around all three points from the starting point S is searched. The fitness is calculated by the reciprocal of the total length of the traveling route, and the selection is performed using a roulette strategy and an elite conservation strategy. The crossover is performed so that a new gene having a similar order of the circulating nodes is generated from the two genotypes. Mutation is performed based on the set mutation probability. As a result of an experiment in which the number of individuals is 50, the termination condition is 50 generations non-evolution, and the mutation probability is 5%, the shortest path M shown in FIG.
SRP was obtained. From the figure, the three designated waypoints A,
It can be seen that the shortest route that passes through all of B and C is obtained. Further, the route passing through the point B has nodes that pass in duplicate, and the shortest route having such a route can be searched. As described above, it has become possible to search for the shortest traveling route that traverses all of a plurality of designated transit points in a single process, which is difficult with the conventional method.

(C) Optimum tour route search when an upper limit cost is set In an optimum tour search that traverses a designated waypoint, if an upper limit of the tour cost (for example, the maximum traveling distance) is set, the upper limit is set. This is a case where all designated waypoints cannot be visited within the cost. In such a case, it is necessary to search for and output a tour route that goes around the maximum number of designated waypoints at or below the upper limit cost. FIG. 31 is a flowchart of a search process for an optimal traveling route that passes through the maximum number of designated via points at or below the upper limit cost. If the number of designated waypoints is N, a predetermined number m (for example, m = 25) of first individuals passing through all N waypoints, a predetermined number m passing through (N-1) designated waypoints , A predetermined number m of third individuals passing through (N-2) designated waypoints,... A predetermined number m of Nth populations passing through one designated waypoint To obtain an initial group (step 601). Each individual is represented by the genotype {G ₁ {G ₂ } described in the above (a).

The shortest (optimal) circuit path passing through all the waypoints, the shortest circuit path passing through (N-1) waypoints, and (N-2) ) Shortest round path through the via points, ...,
The shortest traveling route passing through one designated waypoint is determined (step 602). In this case, a new individual is generated in the process of applying GA to the population passing through the (N-1) waypoints, but the individual not passing through the (N-1) waypoints is a lethal gene. Need to kill as. When the optimal tour route for each individual group is obtained, it is checked whether there is an optimal tour route that is equal to or less than the set upper limit cost (maximum traveling distance) (step 603). Whether the cost is below the maximum cost is determined by the fitness
The running distance of the traveling route is obtained from the reciprocal of (Fitness), and it is determined whether the running distance is equal to or less than the upper limit cost (maximum running distance). If there is an optimal route that is equal to or less than the upper limit cost (maximum traveling distance), a route that passes through the maximum number of waypoints is searched for from among the optimal routes, and is displayed on a display screen, for example (step 604).

(D) Simulation Results An experiment was performed in which the present method was applied to the above-described map information of Iwaki City. The designated waypoints are three points A to C in the map.
All individuals are divided into three groups,
GA processing is performed using only individuals passing through one, two, or three points. The parameters used in this experiment are that the number of individuals per individual group is 50, the termination condition is 50 generations non-evolution, and the mutation probability is 5%. When the upper limit cost is 25 km, the shortest route that passes through all the waypoints is obtained as shown in FIG. 32A, and the same result as the above-described optimal route search method that goes through all the designated waypoints is obtained. When the upper limit cost is 10 km, a shortest route that passes through two points A and B among the waypoints is obtained as shown in FIG. That is, when the upper limit cost is 10 km, the traveling distance of the shortest route that passes through three points exceeds the upper limit cost, and the optimal route that passes through the two points A and B is used as a solution, so that the travel distance is within the upper limit cost. The shortest path through the maximum number of waypoints can be obtained. As described above, the present invention can be applied to a tour route search, a home delivery route search, and the like in a car navigation system.
As described above, the present invention has been described with reference to the embodiments. However, the present invention can be variously modified in accordance with the gist of the present invention described in the claims, and the present invention does not exclude these.

[0063]

As described above, according to the present invention, a weight is determined according to the number of designated via points included in an individual (route), and the reciprocal of the value obtained by multiplying the distance from the starting point to the destination of the individual by the weight is determined. Was used as the evaluation function in the genetic algorithm, so that it was possible to efficiently search for the optimal route using individuals that had previously been lethal genes, reduce the amount of calculation, and search for the optimal route in a short time. Was. According to the present invention, the shortest path length L of the path including the n designated via points, which is obtained each time the genetic algorithm is applied, is set by setting the weight smaller at first.
_n and the shortest path length L of the path including the (n-1) designated via points
comparing the magnitude of the path length _{L n-1 · α n-} 1 multiplied by the weight alpha _n-1 to _n-1, if _{L n> L n-1 ·} α a _n-1, a predetermined percentage weights Since the weight is increased and the weight is controlled to be an optimum value, an optimum solution can be obtained without falling into a local solution. Further, according to the present invention, when searching for an optimal circuit route, from the starting point
An extended genotype having as a gene a virtual node corresponding to all the passing real nodes of the route to the last waypoint after passing through (N-1) specified waypoints, and from the last specified waypoint to the starting point A predetermined number of traveling paths (individuals) expressed by combining genotypes having genes that have all the passing real nodes of a path as genes are prepared as initial populations, and a genetic algorithm is repeatedly applied to the populations. Therefore, it is possible to search for an optimal traveling route that passes through all the designated waypoints.

Further, according to the present invention, a predetermined number of first populations passing through all designated transit points, a predetermined number of second populations passing through (N-1) designated transit points, (N- 2) Provide a predetermined number of third populations passing through the designated waypoints,..., And apply the genetic algorithm repeatedly to each population to optimally travel through all the waypoints; -1) Optimum circuit route through (N-2) waypoints, etc., and the maximum number of waypoints with less than or equal to the specified upper limit cost are determined. Since it is configured to search for a traveling route, it is possible to search for a traveling route that returns to the departure place after passing through the maximum number of designated transit points at or below the upper limit cost.

[Brief description of the drawings]

FIG. 1 is a schematic configuration diagram of a navigation device for explaining an optimal route search according to the present invention.

FIG. 2 is a schematic configuration diagram of a navigation device for explaining an optimal traveling route search according to the present invention.

FIG. 3 is an explanatory diagram of a GA processing procedure.

FIG. 4 is an explanatory diagram of a general individual expression.

FIG. 5 is an explanatory diagram of a genotype applicable to a shortest path search between two points.

FIG. 6 is a configuration diagram for executing a shortest route search based on GA.

FIG. 7 is a search end determination processing flow.

FIG. 8 is an explanatory diagram of a selection process.

FIG. 9 is a flowchart of a selection process.

FIG. 10 is an explanatory diagram of uniform crossover.

FIG. 11 shows an example in which genotype individuals are uniformly crossed.

FIG. 12 is a crossover processing flow.

FIG. 13 is an explanatory diagram of a mutation process.

FIG. 14 is a flowchart of a mutation process.

FIG. 15 is an explanatory diagram of a shortest route search result.

FIG. 16 is an explanatory diagram of the second and third ranks obtained simultaneously with the shortest route in the shortest route search.

FIG. 17 is an explanatory diagram of a route having a virtual node.

FIG. 18 is an example of a connection node.

FIG. 19 is an explanatory diagram of an extended genotype.

FIG. 20 is an explanatory diagram of a search result of a shortest route passing through all designated via points.

FIG. 21 is an explanatory diagram of a search result of a route including two waypoints.

FIG. 22 is an explanatory diagram of each parameter.

FIG. 23 is an explanatory diagram of an ability to search for an optimum solution for a weight α.

FIG. 24 is an example of a local solution (α = 1.5).

FIG. 25 is an example of a local solution (α = 100.0).

FIG. 26 is a diagram showing the relationship between generational changes and the variance of individual groups.

FIG. 27 is an explanatory diagram of the formation of a building block (α = 3.0).

FIG. 28 is an explanatory diagram of building block formation (α = 100.0).

FIG. 29 is a flowchart of a weight determination process.

FIG. 30 is an explanatory diagram of a search result of the shortest traveling route passing through all designated via points.

FIG. 31 is a flowchart of a search processing flow for the shortest route that is equal to or less than the upper limit cost.

FIG. 32 is an explanatory diagram of a search result of an optimal circuit route that passes through a maximum number of designated waypoints at or below the upper limit cost.

[Description of Signs] 1. Map information storage medium 2. Vehicle position detection unit 3. Mocon unit 4. Route search unit that searches for the optimal route by GA 5. Monitor S. Departure point G. Purpose Location A, B ... Via point MSP ... Shortest route

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Jun Inagaki 15-17-17 Ark Court Kita-Onishi 303 Chuo-ku, Sapporo-shi No. 5-28 F term (reference) 2F029 AA02 AB01 AB07 AB13 AC02 AC09 AC14 5H180 AA01 BB13 CC12 EE02 FF04 FF05 FF22 FF32

Claims (5)

[Claims] 1. When searching for a route from a first point to a second point through N designated via points, 1 is set to the maximum number M of connected nodes in all real nodes and the designated via point. The smaller one of the added (N + 1) is set as the number k of virtual nodes, and k virtual nodes are prepared to enable a return route for each real node, and the number of virtual nodes from the first point to the second point is set. An extended genotype having a virtual node corresponding to an all-passed real node as a gene and having a structure capable of performing cross-processing in a genetic algorithm is designed, and a second genotype is designed from a first point represented by the extended genotype. A predetermined number of routes (individuals) to a point are prepared as an initial population, and a genetic algorithm is repeatedly applied to the population represented by the extended genotype to pass from the first point to the second point passing through all designated via points. Find the best route to a point In the optimal route search method, a weight is determined according to the number of designated via points included in the individual (route), and a reciprocal of a value obtained by multiplying a distance from a first point to a second point of the individual by the weight is determined. An optimal path search method, wherein is an evaluation function in a genetic algorithm. 2. The shortest path length L _n of a path including n designated via points obtained each time the genetic algorithm is applied by setting the initial value of the weight to be small, and (n-1) designated paths Compare the path length L _n−1 · α _{n−1 obtained} by multiplying the shortest path length L _{n−1 of the} path including the via point by the weight α _n−1, and calculate L _n > L _n−1 · α _{n -1}
2. The optimal route search method according to claim 1, wherein the weight is increased. 3. When searching for a route from the first point to the second point through the specified N waypoints, the maximum number M of connected nodes in all the real nodes and 1 for the specified waypoint are set. The smaller one of the added (N + 1) is set as the number k of virtual nodes, and k virtual nodes are prepared to enable a return route for each real node, and the number of virtual nodes from the first point to the second point is set. An extended genotype having a virtual node corresponding to an all-passed real node as a gene and having a structure capable of performing cross-processing in a genetic algorithm is designed, and a second genotype is designed from a first point represented by the extended genotype. A predetermined number of routes (individuals) to a point are prepared as an initial population, and a genetic algorithm is repeatedly applied to the population represented by the extended genotype to pass from the first point to the second point passing through all designated via points. Find the best route to a point In the method of searching for an optimal route, the first point and the second point are searched for an optimal circuit route where the first point and the second point are the same point (departure point), from the departure point through (N-1) designated via points. An extended genotype having, as a gene, a virtual node corresponding to an all-passing real node of the route to the last waypoint, and a genotype having, as a gene, an all-passing real node of the route from the last designated waypoint to the starting point A predetermined number of tour routes (individuals) expressed by combining the above are prepared as an initial population, and the optimal tour route returning to the departure point through all designated via points by repeatedly applying a genetic algorithm to the population is provided. The optimal circuit search method to search. 4. A predetermined number of first individuals passing through all designated via points, a predetermined number of second individuals passing through (N-1) designated via points, and (N-2) designated vias A predetermined number of third populations passing through the points are prepared, and an optimal circuit route that passes through all the transit points by repeatedly applying the genetic algorithm to each population is provided, (N-1)
, The optimal circuit route passing through the (N-2) number of waypoints, and so on, and the maximum number of waypoints that pass through the maximum number of waypoints at or below the specified upper limit cost 4. The method according to claim 3, wherein the search is performed. 5. The genetic algorithm, comprising the steps of: calculating a fitness of each individual using an evaluation function; selecting some of the individuals having a large fitness and selecting the remaining individuals; Generating a new individual by performing crossover processing and mutation processing to generate a population consisting of a total of N individuals, until no improvement in fitness is seen, or the number of times the genetic algorithm is applied Applying the above-described processing to each of the individuals constituting the individual group until the set number of times is reached, when the improvement of the fitness is no longer observed, or when the number of applications of the genetic algorithm reaches the set number of times, Outputting the best individual as the optimal route,
5. The optimal circuit search method according to claim 1, further comprising: