JP2006194603A - Route search device and method using genetic algorithm - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a route search device and a method using a genetic algorithm capable of generating a slave route by rearranging a part of a master route after a node, even if the node is not common. <P>SOLUTION: In this route search device, a common node used in common by the first route and the second route is extracted as a crossable node in cross processing, and when a connection route linking from a node on one route which is not a common node to the other route is searched for, the node is added to the extracted crossable node group. In the route search device, when the crossable node selected from the crossable node group exists only on one route, one route is rearranged partially after the crossable node by a connection route using the crossable node as a start end and a route which is a part of the other route, from a node on the other route which is a terminal end of the connection route to a destination, to thereby generate one slave route. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a route search device represented by an in-vehicle navigation device and a route search method for searching for a route from a departure place to a destination.

  A route search device is a device that obtains an optimum route from a departure point to a destination on a route network composed of a large number of nodes and links. As is well known, the in-vehicle navigation device is the most widely used route search device in society, and includes a recording medium in which map data including node and link information is recorded, vehicle information acquisition means using GPS, and the like. And a display means using an LCD or the like, and a control means for processing the data sent from the vehicle information acquisition means or the like, or searching for a route. In the vehicle-mounted navigation device, a node corresponds to a reference point on a route network such as an intersection or an intersection. A link represents a connection between nodes (that is, a road connecting the nodes), and generally has a direction.

  In an in-vehicle navigation system, when a departure point (departure node) and a destination (destination node) are input or determined, an optimal route (or an optimal route candidate consisting of a plurality of routes) connecting these points is controlled. And presented to the user. In the in-vehicle navigation device, map data around the vehicle is read from the recording medium based on the position information sequentially obtained by the vehicle information acquisition means. Then, in a state where the current position of the vehicle and the road to be selected are clearly indicated, a map around the current position is displayed on the display means as needed, thereby guiding the user along the optimum route.

  In a general vehicle-mounted navigation device, an optimum route is obtained by the Dijkstra method. In the Dijkstra method, since the basic algorithm is relatively simple, it does not place an excessive burden on a microcomputer mounted as a control means (hereinafter referred to as a “microcomputer”), and the route search can be performed in a short time. There is an advantage to end. However, on the other hand, a situation in which reliability is low and a truly or sufficiently appropriate route is not required occurs with great frequency. In order to solve this problem, several in-vehicle navigation devices that search for an optimum route using a genetic algorithm have been proposed in recent years (see Patent Documents 1 to 3).

  In an in-vehicle navigation device, a route search using a genetic algorithm is performed as follows, for example. First, an initial population having a chromosome from the starting point to the destination is prepared. In a genetic algorithm, each chromosome is composed of a population of genes. In an in-vehicle navigation device, a link or a node is usually used as a gene, and each route is represented by a link string or a node string.

  Next, selection processing is performed based on the evaluation cost (fitness) calculated for each route, and two parent routes are selected from the initial group. Although two parent routes may be selected at random, the selection process is usually performed so that a route with a high evaluation cost is selected with a higher probability. The evaluation cost is a travel time, a travel cost, a route length, or the like.

Next, genetic manipulation processing is performed on the selected parent route, and generally, crossover processing and further mutation processing are performed to create a new route. The crossover process is to recombine a part of the chromosome, that is, a part of the link string or the node string between the selected paths to generate a new path (child path), and the mutation process is generated. A part of the link sequence or node sequence of the child path is replaced with an appropriate link sequence or node sequence. The child path thus generated is added to the initial population. Then, a dredging process is performed to remove two routes with low evaluation costs from the initial group. Thereafter, after the selection process, genetic manipulation process, and selection process are repeated a predetermined number of times for the path group, the path with the best evaluation cost is identified as the optimal path.
JP-A-9-178500 Japanese Patent Laid-Open No. 11-118501 JP 2000-172664 A

  However, the route search using the genetic algorithm as described above has the following problems. FIG. 7 shows nodes and links near the interchange (IC) on the highway. In the figure, nodes are indicated by white circles and links are indicated by arrows (the same applies hereinafter). Black arrows correspond to expressways, white arrows correspond to general roads, and gray arrows correspond to connecting roads.

Consider a case where two routes E and F are selected in the selection process described above. The route E is a route toward the destination using the highway, and passes this IC. Path E, the link string _{{L e (1), ···} , L e (i), L e (i + 1), L e (i + 2), ···, L e (n)} in Expressed. The route F is a route toward a destination through a general road (at least in the vicinity of this IC), and is a link string {L _{f (1)} ,..., L _{f (j)} , L _{f (j + 1 ), L f (j + 2} ), ···, it is expressed by L _{f (m)}.} In order to create a new route by performing crossover processing on routes E and F, it is necessary that these routes intersect in this IC, in other words, there are common nodes in these routes in this IC. Needed.

However, when routes E and F as shown in FIG. 7 are selected, there is no common node for routes E and F in this IC, so as a result of the crossover process, get off the highway at this IC. Route to the general road {L _{e (1)} , ..., L _{e (i)} , L _{r (1)} , L _{f (j + 2)} , ..., L _{f (m)} } (L _{r (1)} is a link that represents a connection path). For the same reason, a route {L _{f (1)} ,..., L _{f (j)} , L _{r (2)} , L _{e (j + 2)} that moves from a general road to a highway in this IC. ,..., L _{e (n)} } (L _{r (2)} is a link representing a connection path) is not generated. Therefore, when it is required for the optimum route to move from a highway to a general road or from a general road to a highway in this IC, crossover processing functions effectively in the route search using the conventional genetic algorithm. I couldn't. As a result, even when a genetic algorithm is used, there has been a situation in which an appropriate route is not always required as the optimum route.

Similar problems occur for crossroads that include viaducts. FIG. 8A shows nodes and links near such an intersection. In the selection process, for example, two routes of route G and route H are selected, and route G is a route toward the destination through a main road (indicated by a black arrow (link)) on the viaduct. there are, link row _{{L g (1), ···} , L g (i), L g (i + 1), L g (i + 2), ···, L g (n)} expressed in Is done. A route H is a route toward a destination through a road under a viaduct (indicated by a white arrow), and is a link string {L _{h (1)} ,..., L _{h (j)} , L _{h (j +1)} , L _{h (j + 2)} ,..., L _{h (m)} }. Since there is no common node (indicated by white circles) in the routes G and H near the intersection, as a result of the crossover process, the main road passes through the connecting road (indicated by the gray arrow) from the road under the viaduct. migration routes _{{L h (1), ···} , L h (j), L r (3), L g (j + 2), ···, L g (n)} to (L _{r (3 )} Is not a link that represents a connection path).

  Similar problems also occur on parallel roads composed of main roads and side roads. FIG. 8B shows such parallel road nodes and links. In the selection process, when the route I passing the main road (indicated by the black arrow (link)) and the route J passing through the side road (indicated by the white arrow) are selected near the illustrated node, the crossover process is performed. Thus, a route that moves from the main road to the side road through the connecting path (indicated by the gray arrow) or a path that moves from the side road to the main road is not generated.

  The present invention solves the above problems, and even when a certain node is not a common node, a path using a genetic algorithm that can generate a child path by recombining a part of the parent path after that node. A search apparatus and method are provided.

  The route search apparatus of the present invention selects a first route and a second route from a route group composed of a plurality of routes from a starting point to a destination, and performs a crossover process on the first route and the second route. And a route search device using a genetic algorithm that repeatedly generates one or two child routes.

  The route search apparatus according to the present invention extracts, in the crossover process, a common node shared by the first route and the second route as a crossover-capable node, and from the nodes on the first route that are not common nodes. When a first connected route satisfying a predetermined condition connected to two routes is searched, this node is added to the extracted crossable node group, and a node on the second route that is not a common node is changed to the first route. When a second connected route satisfying the predetermined condition to be connected is searched, this node is added to the crossable node group.

  Furthermore, in the route search device according to the present invention, when the crossover node selected from the crossover node group is a common node, the first route and the second route are partially separated after the selected crossover node. If the crossable node selected from the crossable node group exists only on the first route, the crossover node is selected after the selected crossable node. A first connection route having a crossover-capable node as a start node, a route from a node on the second route that is a terminal node of the first connection route and is a terminal node of the first connection route to the destination Then, when the first path is partially recombined to generate one child path, and the crossable node selected from the crossable node group exists only on the second path, the selected crossover is performed. After possible nodes A second connection route having the crossover node as a start node and a node on the first route which is a part of the first route and is a terminal node of the second connection route to the destination. The second route is partially recombined to generate a single child route.

  The route search method of the present invention selects a first route and a second route from a route group composed of a plurality of routes from a departure point to a destination, and performs a crossover process on the first route and the second route. And a route search method using a genetic algorithm that repeatedly generates one or two child routes.

  In the route search method of the present invention, a step of extracting a common node shared by the first route and the second route as a crossable node and a node on the first route that is not a common node are connected to the second route. When a first connected route satisfying a predetermined condition is searched, a step of adding this node to the extracted crossable node group, and a node on the second route that is not a common node is connected to the first route. And a step of adding this node to the crossable node group when a second connection route satisfying a predetermined condition is searched.

  Furthermore, in the route search method according to the present invention, when the crossover node selected from the crossover node group is a common node, the first route and the second route are partially separated after the selected crossover node. And the step of generating two child paths by recombination with each other and the crossover possible node selected from the group of crossover possible nodes exist only on the first path after the selected crossover possible node The first connection route having the crossover node as a start node and a part of the second route that is a terminal node of the first connection route from the node on the second route to the destination A path that partially recombines the first path to generate one child path, and a crossable node selected from the crossable node group exists only on the second path, Selected dating After the possible node, from a second connection route having the crossover node as a start node and a node on the first route that is a part of the first route and is a terminal node of the second connection route And a step of partially recombining the second route with the route to the destination to generate one child route.

  According to the present invention, even if a certain node is not a common node, when a predetermined condition is satisfied, after that node, one parent route is a connected route and the other from the end of the connected route to the destination. It is recombined with a part of the parent pathway to generate a child pathway. Therefore, even if the selected parent route does not intersect at the IC, for example, a child route that moves from the IC to the general road or a child route that moves from the general road to the IC is generated by the crossover process. Path search by genetic algorithm functions more efficiently.

  The predetermined condition is that a distance between a node on the first or second route that is not a common node and a terminal node of the first or second connection route that uses this node as a start node is equal to or less than a predetermined distance. Is preferably included. The predetermined condition preferably includes that the length of the first or second connection path is equal to or less than a predetermined length. By these, suitable 1st and 2nd connection paths are searched.

  The route group is preferably prepared using the Dijkstra method, and the first connection route and the second connection route are preferably searched using the Dijkstra method. By combining Dijkstra's method with a genetic algorithm in this way, it is possible to speed up the route search using the genetic algorithm.

  Embodiments in which the present invention is applied to an in-vehicle navigation device will be described below. FIG. 1 is a block diagram showing an outline of an in-vehicle navigation device (hereinafter referred to as “navigation device”) that is an embodiment of the present invention. The navigation apparatus of this embodiment includes vehicle information acquisition means for acquiring the current position of a vehicle, recording means for reading map data recorded on a recording medium, display means for displaying a navigation screen, various information, and the like, a driver Voice output means for outputting a voice for guiding the sound, operation means including various keys, and control means connected to these means for performing various processes and controls necessary for operation.

  The vehicle information acquisition means includes a GPS receiver (1), a gyroscope (3), a gyro microcomputer (5), and the like. The GPS receiver (1) receives radio waves sent from GPS satellites via the GPS antenna (7), and calculates the absolute position of the vehicle based on the received signals. The calculated absolute position of the vehicle (for example, this position is given by latitude and longitude) is sent to the control microcomputer (9) as control means. The control microcomputer (9) is composed of a CPU, RAM, ROM, etc., not shown. The gyroscope (3) is used to obtain the vehicle orientation, and the gyro microcomputer (5) determines the vehicle orientation based on the signal sent from the gyroscope (3) and controls the information. Send to microcomputer (9). Further, a vehicle speed pulse signal is sent to a gyro microcomputer (5) from a vehicle speed sensor (not shown) provided in the vehicle. The gyro microcomputer (5) determines the speed and travel distance of the vehicle based on the vehicle speed pulse signal and sends them to the control microcomputer (9). The control microcomputer (9) determines the relative position of the vehicle from the vehicle direction and travel distance information sent from the gyro microcomputer (5), and travels from the relative position information and the absolute position information sent from the GPS receiver (1). The trajectory is obtained and matched with the road (link) on the map data.

  The recording means includes a recording medium (11) and a driver (13) that reads information from the recording medium (11) and sends it to the control microcomputer (9). As the recording medium (11), for example, a DVD, a CD-ROM, or a hard disk is used, and map data is recorded. The map data includes shape information (background data and road data), connection information (road network data), attribute information (location search data and service search data), and the like. The connection information includes information regarding nodes and links on the map, and is used for route search. The information regarding each node includes, for example, a node ID number, coordinates, road type information, and adjacent link information. The information regarding each link is composed of, for example, a link ID number, road type information, restriction information, link length, adjacent node information, deployable link number, and the like.

  The control microcomputer (9) appropriately acquires information necessary for processing from the recording medium (11) via the driver (13). For example, when performing a navigation operation for displaying a navigation image on the display means, the navigation device acquires nearby nodes, links, and other information based on position information obtained from the vehicle information acquisition means. . The control microcomputer (9) processes the information obtained from the vehicle information acquisition means and the recording means, and sends instructions necessary for creating a navigation image to the display means. The display means includes a drawing IC (15), an LCD I / F (17), an LCD (19), and the like. The drawing IC (15) has a memory storing character data representing the current position and traveling direction of the vehicle, various roads and facilities, and character data for displaying addresses, road names, and the like. Then, image data of the navigation image is created based on an instruction from the control microcomputer (9). The image data created by the drawing IC (15) is sent to the LCD (19) via the LCD I / F (17), and a navigation image is displayed on the LCD (19). The control microcomputer (9) processes the information obtained from the vehicle information acquisition means and the recording means during the navigation operation, and creates voice data for guiding the driver by voice. The audio data is created in a digital format and sent to the audio output means. The audio output means is composed of a digital-analog converter (21), an amplifier (23), a speaker (25), etc., and the audio data sent from the control microcomputer (9) is a digital-analog converter ( After being converted into an analog by 21) and amplified by the amplifier (23), it is output from the speaker (25).

  The operation means includes various operation keys (27) provided on the casing of the navigation device, a touch panel (29) disposed on the LCD (19), an attached remote controller (not shown), and the remote controller The remote control light receiving unit (31) that receives the infrared signal sent from the computer and the key microcomputer (33) that processes the input from these are configured. When a certain operation key (27) is pressed, the key microcomputer (33) discriminates the pressed operation key (27) and sends an operation command corresponding thereto to the control microcomputer (9). The key microcomputer (33) determines the area where the touch panel (29) is pressed while the setting screen is displayed on the LCD (19), and sends an operation command corresponding to the operation to the control microcomputer ( Send to 9). The remote control light receiving unit (31) converts the received infrared signal into an electrical signal, further demodulates the converted electrical signal and sends it to the key microcomputer (33). The key microcomputer (33) is based on the transmitted signal. The specified operation command is sent to the control microcomputer (9).

  In addition to the vehicle information acquisition means and storage means described above, the navigation apparatus is provided with means for using infrastructure such as an information system for providing road traffic information and an automatic toll collection system for highways. The navigation device can acquire road traffic information such as traffic jam information, accident / construction information, and speed regulation information using VICS (Vehicle Information Communication System). The VICS tuner (35) receives FM multiplex broadcasting on which road traffic information is superimposed via the FM antenna (37), and transmits it from the radio beacon provided on the road via the RF antenna (39). The received 2.5 GHz RF signal related to road traffic information is received and demodulated. The VICS microcomputer (41) is connected to an optical beacon light receiving unit (43) that receives an infrared signal related to road traffic information transmitted from an optical beacon provided on the road. 43) converts the received infrared signal into an electrical signal, and further demodulates the electrical signal. The VICS microcomputer (41) processes the demodulated signal sent from the VICS tuner (35) and the optical beacon light receiving unit (43) to obtain road traffic information. The obtained road traffic information is processed by the control microcomputer (9) and displayed on the LCD (19). In addition, during the navigation operation, a navigation screen is created based on road traffic information sent from the VICS microcomputer (41) in addition to vehicle information and map data. For example, a congested road is clearly shown on the navigation screen based on the traffic jam information.

  Further, an ETC device (45) for using an ETC (Electronic Toll Collection System) is connected to the navigation device. The ETC device (45) includes an RF antenna for transmitting and receiving a 5.8 GHz RF signal, a transmitting and receiving unit that performs modulation / demodulation processing, a reader / writer that reads and writes an ETC card, and the like (none of which are shown). ). For example, based on the information sent from the ETC device (45), the control microcomputer (9) guides the lane of the ETC gate with the voice of the speaker (25), or displays an enlarged view of the ETC gate on the LCD (19). indicate.

  Hereinafter, a route search operation performed by the navigation device of this embodiment will be described. FIG. 2 is a flowchart showing an outline of the route search operation. The program describing this operation is stored in the ROM in the control microcomputer (9) and is executed by the CPU in the control microcomputer (9). Is done. When the route search operation mode is selected via the operation means, a setting screen for prompting the driver to input the starting point and the destination is displayed on the LCD (19) according to an instruction from the control microcomputer (9). Setting is performed (S1). Note that the set departure point may be the current position of the vehicle obtained from the vehicle information acquisition unit. Then, the control microcomputer (9) reads the node and link information in a predetermined area including the departure place and the destination from the recording medium (11) and stores them in the built-in RAM (S2). In subsequent processing, the node and link information stored in the RAM are referred to as appropriate.

  After step S2, the control microcomputer (9) performs processing for generating an initial population using the Dijkstra method (S3). The initial population is a route population (gene pool) consisting of a plurality of routes, which is a departure node (or a node closest to the departure location, hereinafter referred to as a “departure node”) and a destination node (or destination). It is composed of a plurality of routes (for example, 20 routes) connecting nodes closest to the ground (hereinafter referred to as “destination nodes”). The generated route group is stored in the RAM. Note that the initial population may be generated using a BFS method other than the DFS method or the Dijkstra method.

After step S3, processing for selecting two parent routes from the route group is performed (S4). In step S4, any two parent routes may be selected, but a route with a high evaluation cost, which will be described later, is preferably selected as a parent route with a higher probability. In this embodiment, the link is a gene, and the path is expressed by a link string. Thereafter, the selected two parental route path _{A {L a (1),} ···, L a (n)} and the path _{B {L b (1),} ···, L b (m) }write. The route A is composed of n links from La ₍₁₎ to La _{(n) in} order from the departure point side. The route B is composed of m links from L _{b ( 1)} to L _{b (m) in} order from the departure side. In addition, after that, n + 1 nodes on the route A are sequentially arranged from the departure side, N _{a (0)} ,..., N _{a (n)} (N _{a (0)} is the departure node, N _{a (0 n)} is expressed as destination node). In addition, m + 1 nodes on the route B are arranged in order from the departure side, N _{b (0)} ,..., N _{b (m)} (N _{b (0)} = N _{a (0)} , N _{b ( m)} = N _{a (n)} ) The starting node of the link La _(i) is the node Na _(i-1) , and the terminal node of the link La _(i) is the node Na _(i) .

  After step S4, gene manipulation processing, specifically, crossover processing is performed on the selected parent route, and one or two child routes are generated (S5). The generated route is added to the route group, and wrinkle processing is performed (S6). In the selection process, the evaluation costs are compared for the routes constituting the route group, and one or two routes having the lowest evaluation cost are removed from the route group. When there is one child path, one path is removed, and when there are two child paths, two paths are removed. Note that there may be a plurality of identical routes in the route group. Therefore, when there is no identical route, it is determined whether or not the same route exists. If there is a route, step S6 may be configured to remove the overlapping portion of the route from the route group. In this embodiment, only the crossover process is performed in step S5. However, a process for a child path such as a mutation process that replaces a part of the link string of the generated child path is further performed. Also good. The present invention relates to a crossover process in a genetic algorithm, and the route search operation may be changed in any way as long as the effect of the present invention is not lost.

  In this embodiment, in step S6 (also step S4 as necessary), the route evaluation cost is obtained as: evaluation cost = Σ (required time × driver hour unit price) + fuel unit price × Σconsumed fuel + toll road fee It is done. In this equation, the required time, the driver time unit price, and the fuel consumption are values determined according to the link, and Σ means that the sum is taken for all the links constituting the route. The required time for each link is the required time = (average travel time x road type factor x road width factor x regulation speed factor) + 10 seconds x number of left / right turns + 30 seconds x number of signals + 5 seconds x number of intersections) x regional coefficient x bypass coefficient Is calculated as The average travel time, road type coefficient, regulation speed coefficient, number of signals, etc. are included in the map data for each link or are determined based on the information of the map radar. The average travel time is the average time required to move the link, and the road type coefficient, road width coefficient, regulation speed coefficient, area coefficient and bypass coefficient are coefficients set according to various features of the link. . For example, the area coefficient is 0.9 when the link is in the city, and 1 otherwise. The bypass coefficient is 1.1 if the link is bypass or part of the bypass, and 1 otherwise. The driver time unit price is a value for converting the required time into cost (price), and is set high near the departure point, low near the destination, and medium in the middle.

  The fuel consumption for each link is calculated as fuel consumption = link length ÷ fuel consumption × road type fuel consumption coefficient + 50 ml × number of right / left turns + 50 ml × number of signals. Further, in the above-described formula relating to the evaluation cost, the driver hour unit price is a value for converting the travel time into the cost, and the toll road fee is a toll for a highway or the like on the route.

  After step S6, the control microcomputer (9) determines whether or not an end condition is satisfied (S7). For example, the end condition is that the number of repetitions of steps S4 to S6 has reached a predetermined number (for example, 50 times). When it is determined in step S7 that the termination condition is not satisfied, the control microcomputer (9) performs the processes of steps S4 to S6 again. If it is determined that the condition is satisfied, the control microcomputer (9) displays the route search result on the LCD (19) (S8). On the LCD (19), for example, the route with the highest evaluation cost or a plurality of routes with the highest evaluation cost is displayed.

In the present invention, even if a node having one parent path is not a common node, if a predetermined condition is satisfied in the crossover process (S5), a child path is generated by partially recombining the link string after that node. It is possible to do. Hereinafter, the crossover process of the present embodiment will be described in detail with reference to the flowcharts of FIGS. Referring to FIG. 3, first, a variable g and a flag are initialized (S11). The variable g is set to 1 and the flag is turned off. Note that the values of variables, flags, processing results, etc. mentioned below are appropriately stored in the RAM in the control microcomputer (9) and read out. Next, a process of extracting a common node existing in both paths A and B is performed (S12). The extracted common node is stored in the control microcomputer (9) as a crossover capable node. Hereinafter, k extracted nodes capable of crossing are denoted as N ₍₁₎ ,..., N _(k) . At this time, the departure point and the destination node are excluded from the crossover possible nodes. Further, if the first few links are common in paths A and B and / or if the last few links are common in paths A and B, the common nodes on such links Are preferably excluded from the crossover possible nodes. There may be zero crossable nodes extracted in step S12. Even in this case, the route search operation shown in FIGS. 3 to 5 is executed without any problem.

After step S12, a process of setting a value of a variable i for specifying a node on the path A is performed in the subsequent process (S13). When step S13 is performed first, i is set to 1. Step S13 is the beginning of a loop that repeats the processing from steps S14 to S26 described later, and step S15 described later is the end of this loop. After step S13, it is determined whether or not the node N _{a (i)} is a common node (S14). If it is determined in step S14 that the node N _{a (i)} is a common node, it is determined in step S15 whether or not the node N _{a (i)} is a node immediately before the destination node. , I is determined whether it is equal to n-1 (S15). If it is determined in step S15 that the value of i is different from n-1, the process returns to step S13 and the value of i is incremented. Thereafter, the processing after step S14 is performed again.

If it is determined in step S14 that the node N _{a (i)} is not a common node, processing for setting the value of the variable j that identifies the node of the path B is performed in the subsequent processing (S16). When step S16 is performed first, j is initialized to 1. Step S16 is the beginning of a loop that repeats the processing from steps S17 to S24 described later, and step S18 described later is the end of this loop. After step S16, it is determined whether or not the node N _{b (j)} is a common node (S17). If it is determined in step S17 that the node N _{b (j)} is a common node, it is determined in step S18 whether the node N _{b (j)} is a node immediately before the destination node. , J is equal to m−1 or not (S18). If it is determined in step S18 that the value of j is different from m−1, the process returns to step S16, and the value of j is incremented. Thereafter, the processing after step S17 is performed again.

If it is determined in step S17 that the node N _{b (j)} is not a common node, whether or not the straight line distance between the node N _{a (i)} and the node N _{b (j)} is equal to or less than a predetermined distance X Is determined (S19). If it is determined in step S19 that the straight line distance exceeds the predetermined distance X, the processes in and after step S18 are performed. If it is determined in step S19 that the straight line distance is equal to or less than the predetermined distance X, a search for a route (hereinafter referred to as “connected route” ₎ from the node N _{a (i)} to the node N _{b (j)} is performed. This is performed using the Tykstra method (S20). Note that a BFS method other than the DFS method or the Dijkstra method may be used. After step S20, it is determined whether the route search is successful (S21). For example, if a connected route is not obtained even if route search is performed for a predetermined time, it is determined in step S21 that the route search has failed. If it is determined that the route search has failed, that is, if a connection route from the node N _{a (i)} to the node N _{b (j)} cannot be obtained, the processing from step S18 is performed. When a connected route from the node N _{a (i)} to the node N _{b (j)} is obtained in step S20 ₍ when it is determined that the route search is successful in step S21), the length of this connected route Is less than or equal to X + α (α is a predetermined constant) (S22).

When it is determined in step S22 that the length of the obtained connection route exceeds X + α, the processing after step S18 is performed. When it is determined that the length of the route obtained in step S20 is less than or equal to X + α, the node Na _(i) is stored as a crossover possible node N _{(k + g)} and added to the crossoverable node group. (S23). Then, the searched connection path is R _(g) : {L _{r (g) (1)} ,..., L _{r (g) (o)} } (o is a positive integer. If o = 1, R _(g) is stored as one link _{Lr (g) (1)} ), and the flag is turned on (S24).

Step S18 is performed after step S24. When Steps S17 and S19 to 24 are repeatedly performed for the node _{Nb (1)} to the node _{Nb (j-1)} , it is determined in Step S18 that the value of j is equal to m-1. In this case, after step S18, it is determined whether or not the flag is on (S25). If the flag is on, the value of the variable g is incremented and the flag is turned off (S26). ). After step S26, or when the flag is off in step S25, step S15 is performed.

Since steps S17 and S19 to 24 are repeatedly performed for the node N _{b (1)} to the node N _{b (j−1)} , a certain node N _{a (i)} on the route A is on the route B to be connected. It is possible that a plurality of connected routes having different nodes are searched. In this embodiment, when step S24 is repeatedly performed in such a case, the recorded connection route R _(g) is overwritten, and finally the connection route closer to the destination (on the connection route B). The connection route where the node is closest to the target value) is recorded. In step S24, when the length of the connection route R _(g) already stored is compared with the length of the newly obtained connection route, the latter is shorter than the length of the former. The latter may be overwritten and stored as the connection route R _(g) .

When Steps S14, S16 to S26 are repeatedly performed for the node Na ₍₁₎ to the node Na _(i-1) , the value of i is determined to be equal to n-1 in Step S15. Thereafter, a series of processes shown in FIG. 4 is performed. The processes in steps S31 to S44 shown in FIG. 4 are basically the same as the processes in and after step S13 shown in FIG. 3 except that the positions of the route A and the route B are changed. Therefore, the detailed description regarding the process of step S31 thru | or 44 shown in FIG. 4 is abbreviate | omitted. For example, step S31 corresponds to step S13, step S32 corresponds to step S14, step S35 corresponds to step S17, and step S37 corresponds to step S19.

When Steps S32 and S34 to 44 are performed for the node _{Nb (1)} to the node _{Nb (j-1)} , it is determined in Step S33 that the value of j is equal to m-1. Thereafter, a series of processes shown in FIG. 5 is performed. First, an arbitrary crossoverable node N _(w) is selected from the stored crossoverable node group (S61). Here, w = 1 to k + g _f , and g _f is extracted in step S12 as understood from the final value of g, that is, the above description and FIGS. It is equal to the total number of crossover nodes added later for the crossover node group. Next, it is determined whether or not the selected crossable node N _(w) is a node on the route A (S62). If it is determined in step S62 that the crossover possible node N _(w) is on the path A, it is determined whether or not the crossover possible node N _(w) is a node on the path B (S63). If it is determined in step S63 that the crossover possible node N _(w) is on the route B, that is, if the crossover possible node N _(w) is a common node of the routes A and B, this crossover possible node N _(w) After that, a crossover process for partially recombining the link strings of the paths A and B is performed, and child paths C and D are generated (S64). Crossable node N _(w) is node N _{a (p)} located on p + 1 from the departure node on route A, and is further located on q + 1 from route node on route B Assuming that N _{b (q)} , the child path C generated in step S64 is a link string {L _{a (1)} ,..., L _{a (p)} , L _{b (q + 1)} ,. , L _{b (m)} }, and the child path D is linked to the link string {L _{b (1)} ,..., L _{b (q)} , L _{b (p + 1) ,. )} }. After step S64, these child paths C and D are added to the path group (S65), and the crossover process ends.

If it is determined in step S62 that the crossover possible node N _(w) selected in step S61 is not a node on the route A, that is, a node that exists only on the route B, this crossable node After N _(w) , a crossover process for partially recombining the link sequence of the path B is performed to generate one child path D (S66). In step S66, the link sequence after the crossover possible node N _(w) is stored in step S42, and the link sequence { ₍ link ₎ of the connection route R _(g) connecting the crossover possible node N _(w) and the node on the route A L _{r (g) (1)} ,..., L _{r (g) (o)} } and on path A which is a part of route A and is a terminal node of link L _{r (g) (o)} And the link sequence of the route from the node to the destination node. For example, the node N _(w) that can be crossed is the node located at the q + 1th position on the route B counted from the departure node, and the node on the route A that is the terminal node of the link L _{r (g) (o)} Is the node located at the p + 1th position from the departure node, the child route D is linked to the link string {L _{b (1)} ,..., L _{b (q)} , L _{r (g) (1 ), ···, L r (g} ) (o), L a (p + 1), the _{···, L a (n)}} . After step S66, this child route D is added to the route group (S65), and the crossover process ends.

If it is determined in step S63 that the crossover possible node N _(w) selected in step S61 is not a node on the route B, that is, a node that exists only on the route A, this crossable node After N _(w) , a crossover process for partially recombining the link string of the path A is performed to generate one child path C (S67). At step S67, the crossover enabled node N _(w) after the link row is stored in step S24, the link row of connection channel R _(g) connecting the this crossover allows node N _(w) and on the path B node {L _{r (g) (1)} ,..., L _{r (g) (o)} } and a route B that is a part of the route B and is a terminal node of the link L _{r (g) (o)} This is a composite of the link sequence of the route from the upper node to the destination node. The crossover possible node N _(w) is the node located at the p + 1th position on the route A from the departure node, and the node on the route B which is the terminal node of the link L _{r (g) (o)} Assuming that the node is located at the q + 1th position from the ground node, the child path C has link strings {L _{a (1)} ,..., L _{a (p)} , L _{r (g) (1),.} .., L _{r (g) (o)} , L _{b (q + 1)} ,..., L _{b (m)} }. After step S67, this child route C is added to the route group (S65), and the crossover process is terminated.

FIG. 6 is a conceptual diagram illustrating the crossover process of this embodiment, and shows an example of the parent routes A and B selected in step S4 shown in FIG. In the figure, these routes are shown partially omitted. Each link constituting route A is indicated by a black arrow, each link constituting route B is indicated by a white arrow, and nodes are indicated by white circles. ing. In step S12 shown in FIG. 3, a common node is extracted and stored as a crossable node. In the example shown in FIG. 6, the step S12 is executed, the link L as _{a (2)} and L _{b (2)} of the terminating node crossover allows nodes N _(1), the link L _{a (n-1)} and L _{b (m-1)} starting node is stored as crossover enabled node N _(k) (Note that other common node located is also stored during the crossover possible nodes N ₍₁₎ and N _(k)).

The processing after step S13 shown in FIG. 3 searches for and stores a connected route connecting from the node to route B that satisfies a predetermined condition for each node on route A that is not a common node. . Further, the node for which such a connection route is searched is added to the crossable node group. In a loop starting from step S16 and ending at step S18, a node on the route B excluding the departure point and destination node is not a common node specified in step S13. For each, an attempt is made to find a connection path that satisfies a predetermined condition. For example, when the node N _{a (p)} on the route A shown in FIG. 6 is specified in step S13, the first condition is that the linear distance between the nodes at both ends of the connection route is a predetermined distance X or less. Therefore, in step S19, it is determined whether or not the node on the route B designated in step S17 is within a circle with a radius X centered on the node _{Na (p)} . In the example shown in FIG. 6, when the node N _{b (q)} on the route B is in this circle and the node N _{b (q)} is designated in step S17, the node N _a is added after step S19. A connection route from _(p) to node N _{b (q)} is searched (S20).

The second condition is that the length of the connection path is not more than a predetermined length. The connection route obtained in step S20 may be, for example, a route that protrudes greatly from the circle shown in FIG. 6, and such a substantially meaningless route is eliminated by imposing this condition. This condition is determined in step S22. The predetermined length X + α is naturally longer than X. For example, α is set to the same size as X. When the connection route satisfies the second condition, the node N _{a (p)} is stored as a crossable node (S23), and the connection route is stored as R _(g) (S24). In FIG. 6, the stored connection route R _(g) is indicated by a link with a gray arrow.

  The processing after step S31 shown in FIG. 4 satisfies the above first and second conditions in the same manner as described above for each of the nodes on the route B that are not the common nodes extracted in step S12. A connection route connecting from the node to the route A is obtained and stored. In addition, the node on the route B that is the starting end of such a connection route is added to the group of nodes that can be crossed.

In the example shown in FIG. 6, for example, when a crossover node that is a common node such as the node N ₍₁₎ or the node N _(k) shown in the figure is selected in step S61, in step S64, Link trains are recombined as in the conventional crossover process to create two child paths. On the other hand, for example, when the node N _{a (p)} shown in FIG. 6 is selected in step S61, as described above, the route reaches the node N _{a (p)} from the departure node as shown in FIG. Thereafter, one child route C from the node N _{b (q)} to the destination node is generated through the connection route R _(g) as in the route B.

  In the present embodiment, each route is represented by a link string having a link as a gene as described above, but may be represented by a node string having a node as a gene. In addition, the present invention has been described by taking an embodiment in which the present invention is applied to an in-vehicle navigation device. However, the present invention is not limited to other route search devices, for example, a device for searching for a transportation route such as luggage or a patrol route. It can be used for a searching device.

  The above description of the embodiments is for explaining the present invention, and should not be construed as limiting the invention described in the claims or reducing the scope thereof. Each part configuration of the present invention is not limited to the above embodiment, and various modifications can be made within the technical scope described in the claims.

It is a block diagram which shows the outline | summary of the vehicle-mounted navigation apparatus which is an Example of this invention. It is a flowchart which shows the outline | summary of the route search operation | movement which the navigation apparatus of a present Example of this invention performs. It is a flowchart which shows the crossover process which the navigation apparatus of the Example of this invention performs. It is a flowchart which shows the crossover process which the navigation apparatus of the Example of this invention performs. It is a flowchart which shows the crossover process which the navigation apparatus of the Example of this invention performs. It is a conceptual diagram explaining the crossover process which the navigation apparatus of the Example of this invention performs. It is explanatory drawing which shows the node and link of the interchange vicinity of a highway. FIG. 8A is an explanatory diagram showing nodes and links in the vicinity of an intersection, and FIG. 8B is an explanatory diagram showing nodes and links on a parallel road.

Explanation of symbols

(1) GPS receiver
(3) Gyroscope
(5) Gyro microcomputer
(9) Control microcomputer
(11) Recording medium
(19) LCD
(25) Speaker
(27) key

Claims (10)

One or two routes are selected by selecting a first route and a second route from a route group composed of a plurality of routes from a starting point to a destination, and performing a crossover process on the first route and the second route. A route search device using a genetic algorithm that repeatedly generates child routes of
In the crossover process,
A common node shared by the first route and the second route is extracted as a crossable node;
When a first connected route satisfying a predetermined condition connected to the second route is searched from a node on the first route that is not a common node, the nodes on the first route are extracted as a group of nodes that can be crossed. In addition to
When a second connected route that satisfies the predetermined condition connected to the first route from a node on the second route that is not a common node is searched, the node on the second route is added to the crossable node group. ,
If the crossable node selected from the crossable node group is a common node, two child routes are obtained by partially recombining the one route and the second route after the selected crossable node. Produces
A crossover node selected from the crossover node group is present only on the first route, a first connection route having the crossover node as a start node after the selected crossover node; A part of the second route that is a terminal node of the first connection route and a route from the node on the second route to the destination is partially recombined with the first route. Create a child path of
When the crossover possible node selected from the crossover possible node group exists only on the second route, the second connection route having the crossover possible node as the start node after the selected crossover possible node; A part of the first route that is a terminal node of the second connection route and a route from the node on the first route to the destination is obtained by partially recombining the second route. Search device that generates a child route.   The predetermined condition is that a distance between a node on the first route or the second route that is not a common node and a terminal node of the first connection route or the second connection route having the node as a start node is: The route search device according to claim 1, comprising being less than or equal to a predetermined distance.   The route search device according to claim 1 or 2, wherein the predetermined condition includes that a length of the first connection route or the second connection route is equal to or less than a predetermined length.   The route search device according to claim 1, wherein an initial population of the route population is prepared using a Dijkstra method.   The route search device according to any one of claims 1 to 4, wherein the first connection route and the second connection route are searched using a Dijkstra method. One or two routes are selected by selecting a first route and a second route from a route group composed of a plurality of routes from a starting point to a destination, and performing a crossover process on the first route and the second route. A path method using a genetic algorithm that repeatedly generates child paths of
Extracting a common node shared by the first route and the second route as a crossable node;
When a first connected route satisfying a predetermined condition connected to the second route is searched from a node on the first route that is not a common node, the nodes on the first route are extracted as a group of nodes that can be crossed. Adding to the process,
When a second connected route satisfying the predetermined condition connected to the first route from a node on the second route that is not a common node is searched, the node on the second route is added to the crossable node group Process,
If the crossable node selected from the crossable node group is a common node, two child routes are obtained by partially recombining the one route and the second route after the selected crossable node. Generating
A crossover node selected from the crossover node group is present only on the first route, a first connection route having the crossover node as a start node after the selected crossover node; A part of the second route that is a terminal node of the first connection route and a route from the node on the second route to the destination is partially recombined with the first route. Generating a child path of
When the crossover possible node selected from the crossover possible node group exists only on the second route, the second connection route having the crossover possible node as the start node after the selected crossover possible node; A part of the first route that is a terminal node of the second connection route and a route from the node on the first route to the destination is obtained by partially recombining the second route. Generating a child route of the route.   The predetermined condition is that a distance between a node on the first route or the second route that is not a common node and a terminal node of the first connection route or the second connection route having the node as a start node is: The route search method according to claim 6, comprising being equal to or less than a predetermined distance.   The route search method according to claim 6 or 7, wherein the predetermined condition includes that a length of the first connection route or the second connection route is equal to or less than a predetermined length.   The route search method according to claim 6, comprising preparing an initial population of the route population using a Dijkstra method.   The route search method according to claim 6, wherein the first connection route and the second connection route are searched using a Dijkstra method.

Images