Reservation and conventional bus hybrid compiling method based on unmanned driving

Abstract

The invention relates to the technical field of reservation bus taking, and discloses an unmanned driving-based reservation and conventional bus hybrid method, which comprises the following steps of S1, acquiring a reservation request by utilizing a reservation clustering algorithm model; s2, acquiring a reserved vehicle by using a fish swarm algorithm model and automatically driving the reserved vehicle to a reserved getting-on point; s3, switching vehicle states to realize vehicle deployment; s4, planning a reserved route by using the enhanced VPR algorithm model; and S5, the reservation task is completed, and the vehicles are allocated. The state switching between the normal state and the stress state of the vehicle is realized by using the fish swarm algorithm, the reservation route planning is realized based on the enhanced VRP algorithm model, the transport capacity of the idle reservation vehicle is put into the conventional route, and meanwhile, the idle conventional vehicle receives the reservation order, so that the space-time utilization rate of the unmanned vehicle can be effectively improved, and the space-time utilization rate of the unmanned vehicle is improved. The service efficiency of bus travel is improved, the idle phenomenon is reduced, the resource utilization rate is improved, and the operation cost is reduced.

Description

Technical Field

[0001] The invention relates to the technical field of bus reservation, and in particular to a method for mixing reservation and conventional public transportation based on unmanned driving. Background Art

[0002] With the continuous development of driverless technology, various places have gradually integrated the application of driverless technology with public transportation business. The current driverless buses have the characteristics of small vehicle capacity, mainly running on micro-circulation lines, long continuous operation time and high service reliability. Based on the above characteristics, they are very suitable for application scenarios of appointment services, thereby improving public transportation services for citizens.

[0003] Since the vehicle itself is relatively small, the requirements for boarding and alighting points are relatively low; vehicle operation is not affected by the driver's physiological reasons, which can reduce the time for stopping and preparing for departure, and the vehicle can be dispatched immediately upon receiving a reservation request with high service reliability; in addition, considering that the demand for public transportation reservations has a certain degree of uncertainty in space and time, if a special car is used for reservation services, it is easy for it to be idle, which leads to waste of resources and high operating costs. Summary of the invention

[0004] The purpose of the present invention is to provide a method for mixing reservation and conventional public transportation based on unmanned driving to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A method for mixing reservation and regular public transportation based on unmanned driving, comprising the following steps:

[0007] S1. Using the reservation clustering algorithm model, the discrete incentives are spatially clustered according to the pick-up point and the drop-off point, and the reservation pick-up point clustering model and the reservation drop-off point clustering model are obtained respectively, wherein the incentive is the passenger's reservation request for the vehicle;

[0008] S2. Using the fish school algorithm model to obtain a vehicle that is effective for the inducement, so that the vehicle switches from a normal state to a start-up state in a stress state, and automatically drives to the scheduled boarding point;

[0009] S3. On the way to the scheduled boarding point, determine whether the inducement is invalid; when the inducement is invalid, the vehicle switches from the start state to the waiting state. The vehicle in the waiting state stops on the spot and automatically enters the normal state after waiting for a period of time or enters the start state in the stress state under the influence of the inducement; then use the interval segmentation method in the enhanced VPR algorithm model to distribute the vehicles in the normal state to each interval; when the inducement is not invalid, after the vehicle arrives at the scheduled cluster boarding point, the passengers get on the vehicle, the vehicle enters the execution state, and feedback is given to step S2;

[0010] S4. Allocate the reservation requests and use the enhanced VPR algorithm model to plan the reservation route;

[0011] S5. After the vehicle arrives at the reservation site according to the planned route, the reservation task is completed and the passengers get off the vehicle. When the vehicle completes all current reservation tasks, it switches from the execution state to the normal state. Then, the interval segmentation method is used to allocate the vehicles entering the normal state to each interval.

[0012] As a further solution of the present invention: in the step S1, the method for constructing the reservation pick-up point clustering model is as follows:

[0013] First, the data space is divided into grids using the grid clustering method, and then the data is mapped to the grid cells according to certain rules, and then the density of each cell is calculated;

[0014] The construction method of the scheduled drop-off point clustering model is as follows:

[0015] First, the grid clustering method is used to cluster the remote alighting points to obtain the number of clusters, and then the enhanced partition clustering method is used to cluster all the alighting points again.

[0016] As a further solution of the present invention: in the step S2, the fish school algorithm model is used to solve the problem of vehicle switching between normal state and stress state to achieve vehicle deployment; wherein, the vehicle in normal state means a state in which the vehicle is traveling on a fixed route within a specified range; the normal state consists of a driving state and a stationary state, and in the normal state, the vehicle keeps switching repeatedly between the driving state and the stationary state.

[0017] As a further solution of the present invention: in the step S2, the vehicle in the stress state indicates that the vehicle is affected by the inducement and deviates from the fixed route, wherein the stress state consists of a start state, an execution state and a waiting state; whether the inducement is effective on the vehicle is determined by the inducement variable and the corresponding threshold.

[0018] As a further solution of the present invention: in the step S3, the vehicles in the waiting state enter the normal state according to the waiting time and the corresponding threshold, and the vehicle allocation adopts the interval allocation method, and the specific steps are as follows:

[0019] Firstly, the fixed route under normal conditions is divided into several initial intervals according to the number of vehicles in the current normal state, and then the vehicles that are about to enter the normal state are allocated to each interval using the dynamic programming method.

[0020] As a further solution of the present invention: in the S4 step, the enhanced VPR algorithm model includes a heuristic algorithm and an interval allocation method. When allocating reservation requests, it is ensured that the remaining passenger capacity of the vehicle is greater than or equal to the number of reservation requests, and a two-stage strategy of grouping first and then determining the route is adopted. When a reservation request is received, multiple reservation requests are grouped using the grouping strategy. When the vehicle enters the execution state, a group of reservation requests is obtained, and then a heuristic algorithm is used to calculate a route with the shortest driving route as the purpose.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] The present invention realizes the state switching between the normal state and the stress state of the vehicle by utilizing the fish school algorithm, and realizes the reservation route planning based on the enhanced VRP algorithm model, so that the idle reservation vehicle capacity is put into the regular route, and the empty regular vehicle also accepts the reservation order, which can effectively improve the time and space utilization rate of the unmanned vehicle, improve the service efficiency of public transportation, reduce the idle phenomenon, improve the resource utilization rate, and reduce the operating cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A flowchart of a method for mixing reservation and conventional public transportation based on driverless transportation is shown;

[0024] Figure 2 A schematic diagram of receiving a reservation request in a method for mixing reservation and conventional public transportation based on driverless transportation;

[0025] Figure 3 A schematic diagram of grouping reservation requests in a method for mixing reservations with conventional buses based on driverless buses;

[0026] Figure 4 A schematic diagram of reservation route planning in a mixed method of reservation and conventional public transportation based on driverless transportation;

[0027] Figure 5 The schematic diagram of the framework of a system based on the mixed method of unmanned reservation and conventional bus. DETAILED DESCRIPTION

[0028] See also Figure 1 In an embodiment of the present invention, a method for mixing reservation and conventional public transportation based on unmanned driving includes the following steps:

[0029] S1. Using the reservation clustering algorithm model, the discrete incentives are spatially clustered according to the pick-up point and the drop-off point, and the reservation pick-up point clustering model and the reservation drop-off point clustering model are obtained respectively, wherein the incentive is the passenger's reservation request for the vehicle;

[0030] S2. Using the fish school algorithm model to obtain a vehicle that is effective for the inducement, so that the vehicle switches from a normal state to a start-up state in a stress state, and automatically drives to the scheduled boarding point;

[0031] S3. On the way to the scheduled boarding point, determine whether the inducement is invalid; when the inducement is invalid, the vehicle switches from the start state to the waiting state. The vehicle in the waiting state stops on the spot and automatically enters the normal state after waiting for a period of time or enters the start state in the stress state under the influence of the inducement; then use the interval segmentation method in the enhanced VPR algorithm model to distribute the vehicles in the normal state to each interval; when the inducement is not invalid, after the vehicle arrives at the scheduled cluster boarding point, the passengers get on the vehicle, the vehicle enters the execution state, and feedback is given to step S2;

[0032] S4. Allocate the reservation requests and use the enhanced VPR algorithm model to plan the reservation route;

[0033] S5. After the vehicle arrives at the reservation site according to the planned route, the reservation task is completed and the passengers get off the vehicle. When the vehicle completes all current reservation tasks, it switches from the execution state to the normal state. Then, the interval segmentation method is used to allocate the vehicles entering the normal state to each interval.

[0034] Preferably, in step S1, the method for constructing the reservation pick-up point clustering model is as follows:

[0035] First, the data space is divided into grids using the grid clustering method, and then the data is mapped to the grid cells according to certain rules, and then the density of each cell is calculated. The larger the cell density, the larger the induced variable vb;

[0036] The construction method of the scheduled drop-off point clustering model is as follows:

[0037] First, the grid clustering method is used to perform cluster analysis on the far-end alighting points to obtain the number of clusters vk, and then the enhanced partition clustering method k-mean++ is used to perform cluster analysis on all alighting points again; wherein, the enhanced partition clustering method k-mean++ selects K cluster centers according to the following steps: assuming that n initial cluster centers have been selected, 0<n<K, then when selecting the n+1th cluster center, the farther the point is from the current n cluster centers, the higher the probability of being selected as the n+1th cluster center; when selecting the first cluster center, that is, when n=1, the same random method is used, where the parameter K=vk.

[0038] Preferably, in step S2, the fish school algorithm model is used to solve the problem of switching between the normal state and the stress state of the vehicle to achieve vehicle deployment; wherein the vehicle in the normal state means that the vehicle is in a state of traveling on a fixed route within a specified range; the normal state consists of a driving state and a stationary state, and in the normal state, the vehicle is repeatedly switching between the driving state and the stationary state;

[0039] The switching from the stationary state to the driving state is determined by the current startup variable Vm, and the variables related to it are the stationary time Vsn, the front and rear headways Vsd1 and Vsd2; and the average headway Evsd; the calculation formula of the average headway Evsd is as follows:

[0040]

[0041] The switch from driving state to stationary state is determined by the current stop variable vs, and the variable related to it is the idle driving time vsh; the calculation formula of the start variable Vm is as follows:

[0042]

[0043] In the above formula, the driving state coefficients Wm, Wsn, Wsd and the fixed value a are calculated by the machine learning algorithm; it can be obtained that: the larger the starting variable is, the closer the vehicle is to the starting state.

[0044] Preferably, in step S2, the vehicle in the stress state indicates that the vehicle is affected by the inducement and deviates from the fixed route, wherein the stress state is composed of the start state, the execution state and the waiting state; whether the inducement is effective for the vehicle is determined by the inducement variable vb and the corresponding threshold value ythres1, and the variables related thereto are the number of unit reservations vnumt, the distance between the scheduled boarding point and the vehicle vdis, the vehicle idle driving time vsh, and the waiting time vwait and the corresponding threshold value ythres2; when vb>ythres1, and the vehicle is not in the execution state, it indicates that the inducement is effective for the vehicle, the vehicle switches to the start state, and the vehicle in the start state automatically drives to the scheduled boarding point; when vb<ythres1, it indicates that the inducement is not effective for the vehicle, if the current state of the vehicle is the start state, it switches to the waiting state, and if the current state of the vehicle is not the start state, it maintains the original state. The calculation formula of the inducement variable vb is as follows:

[0045]

[0046] In the above formula, Wnumt is the attraction coefficient. The larger the coefficient, the easier it is for the inducement to take effect. vdis is the constraint coefficient. The related variables vdis and vb are inversely correlated. The farther away from the constraint range, the faster the inducement variable will decrease. The derivative can obtain the constraint range:

[0047]

[0048] Let the derivative dy = 1, that is To restrict the scope.

[0049] Preferably, in step S3, whether a vehicle in the waiting state enters the normal state is determined by the waiting time vwait and the corresponding threshold ythres2. When vwait>ythres2, the vehicle enters the normal state. At this time, the vehicle will first travel from the current point to the station under the normal state. Since the station is calculated simply by the fish school algorithm model, the vehicle will first arrive at the nearest station. If there are multiple vehicles entering the normal state near the station or the distance from the front of the normal vehicle is too close, it is easy to cause serious uneven distribution of line vehicles. In order to avoid the above situation during state conversion, the vehicle allocation adopts the interval allocation method, and the specific steps are as follows:

[0050] First, the fixed route under normal conditions is divided into several initial intervals according to the number of vehicles in the current normal state, and then the vehicles that are about to enter the normal state are allocated to each interval using the dynamic programming method;

[0051] For each vehicle, if it is not assigned to a zone, the capacity remains unchanged; if it is assigned to a zone, the zone capacity decreases; the state is defined as follows:

[0052] The current vehicle number i and the interval capacity c are recorded as [i,c]; the subproblem corresponding to the state [i,c] is: the maximum value of the first i vehicles in the interval with capacity c, recorded as dp[i,c]; the problem to be solved is dp[n,cap], so a size of (n+1)×(cap+1) is required;

[0053] After making the decision for vehicle i, the remaining sub-problems for the first i-1 vehicle decisions can be divided into the following two cases:

[0054] When the vehicle i is not assigned, the interval capacity remains unchanged and the state changes to [i-1,c];

[0055] When the assigned vehicle is i, the interval capacity decreases by wgt[i-1], the transport capacity increases by val[i-1], and the state changes to [i-1,c-wgt[i-1]];

[0056] Then we can get the optimal substructure: the maximum value dp[i,c] is the solution with the greater value between not allocating vehicle i and allocating vehicle i; thus we can get the state transition equation: dp[i,c]=max(dp[i-1,c],dp[i-1,c-wgt[i-1]]+val[i-1]);

[0057] If the current vehicle service interval capacity wgt[i-1] exceeds the remaining interval capacity c, then choose not to allocate a vehicle;

[0058] When the vehicle arrives at the interval, it executes the normal task, wherein the vehicle state during this time period belongs to the normal driving state; after the vehicle reaches the reservation cluster boarding point, the vehicle enters the execution state, and after the vehicle reaches the reservation cluster boarding point, the same type of reservation request vnum is allocated according to the size of the reservation time difference. At this time, the reservation number of the reservation point is changed to vnumt(i-1)=vnumt(i)-vnum.

[0059] Preferably, in step S4, the enhanced VPR algorithm model includes a heuristic algorithm and an interval allocation method. Due to the maximum passenger capacity of the vehicle and the requirements of the riding experience, when allocating reservation requests, it is ensured that the remaining passenger capacity of the vehicle is greater than or equal to the number of reservation requests, that is, vcontent ≥ vnum; and a two-stage strategy of first grouping and then routing is adopted. When a reservation request is received, Figure 2 As shown, the grouping strategy is used to group multiple reservation requests, such as Figure 3 As shown in , when the vehicle enters the execution state, it obtains a set of reservation requests, and then uses the heuristic algorithm to calculate a route with the shortest driving route as the purpose, such as Figure 4 As shown; thus, the passenger needs are met and the path is shortest;

[0060] In addition, in the execution state, the original route may be adjusted by the influence of the incentive variable vb; however, the factors for the effectiveness of the vehicle incentive in the execution state are different from those in the normal state, and the current passenger experience needs to be taken into consideration. Therefore, in the execution state, the vehicle's scheduled boarding point and the vehicle distance vdis are relatively small, and the reservation type corresponding to the incentive must be consistent with it. Once the incentive takes effect, it enjoys the highest priority, and the incentive no longer affects other vehicles.

[0061] The grouping strategy is added to solve the problems caused by the single original VRP model. Because of its characteristic of the shortest distance as a single cost function, it always gives priority to the appointment orders with the longest distance. There are two problems here. First, it cannot solve the problem of long detours, which directly affects the user's service experience; second, long-distance appointment needs will be treated as low priority. For this reason, the grouping strategy needs to calculate the detour coefficient matrix Mp (n*n matrix, n is the number of appointment orders), each item in Mp is the detour coefficient Vpij, and the calculation of the detour coefficient requires the direct matrix Md (1*n matrix, n is the number of appointment orders), each item in Md is the direct distance Vdi and the drop-off point distance matrix Mdis (n*n matrix, n is the number of appointment orders), each item in Mdis is the distance between two points Vdisij.

[0062]

[0063] Matrix expression:

[0064]

[0065] Then, perform a simplification process on the matrix, where Vpij (i = j) = 0, Vpij = Max(Vpij, Vpji), and Vpji = 0

[0066]

[0067] Then, select Gp groups with the smallest detour coefficients from Mp, where

[0068]

[0069] After completing the Gp grouping, use the heuristic algorithm for each group to implement the planning of the reservation route. Since the VRP is an NP-hard problem and it is difficult to solve it with an exact algorithm, the heuristic algorithm is the main method for solving the vehicle transportation problem. Over the years, many scholars have studied the vehicle transportation problem and proposed various heuristic methods. Assume a boarding point (depot), with a total of Gp groups, a vehicle capacity of Vcontent, Vnum passengers, and each passenger has a demand of D. The vehicle departs from the boarding point to provide delivery services to the passengers and finally returns to normal. It is required that all passengers are delivered, each passenger is delivered once, and the limit of Vcontent cannot be violated. The goal is to minimize the total distance of all vehicle routes. Since the number of reservation orders in each group after grouping is not large, the solution method of the heuristic algorithm uses a local search method with tolerance to implement the reservation route planning. The process is as follows: The algorithm starts from a candidate route solution and continuously searches in its tolerance neighborhood until there is no better solution in the neighborhood. For an optimization problem:

[0070] (1) min f(x), x ∈ Rn

[0071] where f(x) is the objective function. The search can be understood as a process of moving from one solution to another. Let s(x) represent a solution obtained through the move, and S(x) be the set (neighborhood) of all possible solutions starting from the current solution. Then, the steps of the local search algorithm are described as follows:

[0072] 1. Initialize a feasible solution x, a tolerance R, and r = R.

[0073] 2. Select a solution s(x) after an r move within the neighborhood of the current solution such that f(s(x)) < f(x) and s(x) ∈ S(x). If there is no such solution, if r >= 0, execute step 3; if r < 0, execute step 4. If the solution exists, then r = R, let x = s(x), and repeat step 2.

[0074] 3. r = r - 1, let x = s(x), and repeat step 2.

[0075] 4. Then x is the optimal solution and the algorithm stops.

[0076] When our optimization goal is to maximize the objective function f(x), x is the final solution. The present invention also discloses a system for implementing a mixed method based on unmanned driving reservation and conventional public transportation, referring to Figure 5 The system includes a vehicle automatic dispatching subsystem, a reservation route planning subsystem, a reservation demand management subsystem, a basic data subsystem and a monitoring subsystem; among which,

[0077] The vehicle automatic dispatching subsystem includes an automatic task management module and an automatic vehicle distance adjustment module, among which:

[0078] The automatic task management module is used for task generation and switching management of normal and emergency states of vehicles. On the one hand, it realizes the allocation and management of operational tasks for normal vehicles and directs normal vehicles to perform tasks along normal routes. On the other hand, when receiving an appointment request, the system can automatically allocate vehicles with effective inducements, generate emergency start tasks, and allocate appointment requests when reaching the appointment point to perform the planned route tasks. Finally, after completing the appointment task, the vehicle is switched to normal to perform normal tasks.

[0079] The automatic vehicle distance adjustment module is used to control the vehicle distance under normal conditions to prevent inefficient operation caused by excessive uneven vehicle distance.

[0080] The reservation route planning subsystem includes a reservation route generation module. When a reservation request is received, the reservation route generation module is used to first group the reservation requests, then generate the optimal route based on the boarding and alighting point locations in the grouped passenger reservation requirements and predict the current passenger's waiting time, and then feed the information back to the passenger end, and finally plan the vehicle's return route after completing the reservation task.

[0081] The reservation demand management subsystem includes a boarding point management module, which is used for passenger reservation management. The system provides a mini-program-side reservation application. Passengers can initiate requests within the effective range of the reservation on the mini-program side. At that time, the passengers need to provide the boarding and alighting locations and the number of passengers. After the system receives the passenger demand information, it first clusters the boarding and alighting locations, and then uploads them together with other demand information to the vehicle automatic dispatch subsystem and the reservation route planning subsystem. The former is responsible for dispatching vehicles to the destination, and the latter is responsible for planning the subsequent delivery route. The system then feeds back the route information to the passenger side, and the passenger side can also query the real-time status of the vehicle and the estimated waiting time.

[0082] The basic data subsystem includes a normal line management module and a normal vehicle allocation management module, which are used to manage static services in the effective area.

[0083] The monitoring subsystem includes a location monitoring module, a video media monitoring module, and a route monitoring module.

[0084] The above are only preferred specific implementation modes of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical solutions and inventive concepts of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for mixing reservation and conventional public transportation based on unmanned driving, characterized in that: The following steps are involved: S1. Using the reservation clustering algorithm model, the discrete incentives are spatially clustered according to the pick-up point and the drop-off point, and the reservation pick-up point clustering model and the reservation drop-off point clustering model are obtained respectively, wherein the incentive is the passenger's reservation request for the vehicle; S2. Using the fish school algorithm model to obtain a vehicle that is effective for the inducement, so that the vehicle switches from a normal state to a start-up state in a stress state, and automatically drives to the scheduled boarding point; S3. On the way to the scheduled boarding point, determine whether the inducement is invalid; when the inducement is invalid, the vehicle switches from the start state to the waiting state. The vehicle in the waiting state stops on the spot and automatically enters the normal state after waiting for a period of time or enters the start state in the stress state under the influence of the inducement; then use the interval segmentation method in the enhanced VPR algorithm model to distribute the vehicles in the normal state to each interval; when the inducement is not invalid, after the vehicle arrives at the scheduled cluster boarding point, the passengers get on the vehicle, the vehicle enters the execution state, and feedback is given to step S2; S4. Allocate the reservation requests and use the enhanced VPR algorithm model to plan the reservation route; S5. After the vehicle arrives at the reservation site according to the planned route, the reservation task is completed and the passengers get off the vehicle. When the vehicle completes all current reservation tasks, it switches from the execution state to the normal state. Then, the interval segmentation method is used to allocate the vehicles entering the normal state to each interval.

2. A method for mixing reservation and regular public transportation based on unmanned driving according to claim 1, characterized in that: In the step S1, the method for constructing the reservation pick-up point clustering model is as follows: First, the data space is divided into grids using the grid clustering method, and then the data is mapped to the grid cells according to certain rules, and then the density of each cell is calculated; The construction method of the scheduled drop-off point clustering model is as follows: First, the grid clustering method is used to cluster the remote alighting points to obtain the number of clusters, and then the enhanced partition clustering method is used to cluster all the alighting points again.

3. The method for mixing reservation and regular public transportation based on unmanned driving according to claim 1 is characterized in that: In the step S2, the fish school algorithm model is used to solve the problem of vehicle switching between normal state and stress state to achieve vehicle deployment; wherein, the vehicle in normal state means that the vehicle is traveling on a fixed route within a specified range; the normal state consists of a driving state and a stationary state, and in the normal state, the vehicle keeps switching repeatedly between the driving state and the stationary state.

4. The method for mixing reservation and regular public transportation based on unmanned driving according to claim 1 is characterized in that: In step S2, the vehicle in the stress state means that the vehicle is affected by the inducement and deviates from the fixed route, wherein the stress state consists of a start state, an execution state and a waiting state; whether the inducement is effective on the vehicle is determined by the inducement variable and the corresponding threshold.

5. The method for mixing reservation and regular public transportation based on unmanned driving according to claim 1 is characterized in that: In the step S3, the vehicles in the waiting state enter the normal state according to the waiting time and the corresponding threshold, and the vehicle allocation adopts the interval allocation method. The specific steps are as follows: Firstly, the fixed route under normal conditions is divided into several initial intervals according to the number of vehicles in the current normal state, and then the vehicles that are about to enter the normal state are allocated to each interval using the dynamic programming method.

6. The method for mixing reservation and regular public transportation based on unmanned driving according to claim 1 is characterized in that: In the S4 step, the enhanced VPR algorithm model includes a heuristic algorithm and an interval allocation method. When allocating reservation requests, it ensures that the remaining passenger capacity of the vehicle is greater than or equal to the number of reservation requests, and adopts a two-stage strategy of grouping first and then determining the route. When a reservation request is received, multiple reservation requests are grouped using the grouping strategy. When the vehicle enters the execution state, a group of reservation requests is obtained, and then a heuristic algorithm is used to calculate a route with the shortest driving route as the purpose.

Images