A method for scheduling autonomous taxi positions using water ripples

Abstract

The present invention belongs to the field of self-driving taxis, and specifically discloses a method for watermark scheduling of self-driving taxis, including the following steps: establishing a mathematical model for the scheduling problem; calculating the adjacent area set of each area; calculating the idle rate and turnover rate of vehicles; performing watermark mediation diffusion based on the watermark algorithm to achieve global simultaneous mobilization; algorithm initialization; executing the watermark algorithm of the pth round; watermark iterative calculation; calculating scheduling data, and outputting the scheduling result. The present invention can solve the problem of unbalanced regional scheduling, and the watermark algorithm used brings global dynamic scheduling, has the advantages of fast scheduling speed, automatic scheduling, no need for manual intervention, etc., and can meet the requirements of automatic intelligent operation of self-driving taxis.

Description

Technical Field

[0001] The present invention relates to the field of self-driving taxis, and in particular to a method for scheduling the position of self-driving taxis with watermarks. Background Art

[0002] In the field of self-driving taxis, passengers will choose nearby self-driving taxis to take a taxi. Over time, self-driving taxis in hot spots will be dispatched to remote areas, while self-driving taxis in remote areas will not get dispatch opportunities for a long time, resulting in serious imbalance in regional transportation capacity, which gradually reduces the matching rate between taxis and passengers. For traditional manual taxis, if the driver cannot accept orders for a long time, they will automatically move to hot spots to achieve balance again. However, since self-driving taxis are unmanned, they need to be dispatched intelligently by the system.

[0003] Based on this, the present invention discloses a watermark scheduling method for the position of an autonomous taxi to solve the problem of unbalanced orders received by the autonomous taxi and achieve a dynamic balance, thereby ensuring the operational efficiency of the autonomous taxi. Summary of the invention

[0004] The purpose of the present invention is to provide a method for watermark scheduling of autonomous taxi locations to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above-mentioned purpose, the present invention provides the following technical solution: a method for scheduling watermarks at the position of an autonomous driving taxi, comprising the steps of:

[0006] S1: Establish a mathematical model for the scheduling problem;

[0007] S2: Calculate the neighboring area set of each area;

[0008] S3: Calculate vehicle idle rate and turnover rate;

[0009] S4: Water ripple mediation and diffusion based on water ripple algorithm to achieve global simultaneous adjustment;

[0010] S5: algorithm initialization;

[0011] S6: Execute the watermark algorithm for round p;

[0012] S7: iterative calculation of water ripples;

[0013] S8: Calculate the scheduling data and output the scheduling results.

[0014] Preferably, in S1, n autonomous driving parking areas are set, and autonomous driving taxis can be charged within these areas. The position coordinates of the areas can be denoted as (lon, lat). The number of available parking spaces in area x is denoted as cap(x), and the number of currently parked vehicles in this area is denoted as park(x), where the area number 1 ≤ x ≤ n. Then, the number of arriving vehicles in area x within t hours is denoted as arr(x), and the number of dispatched vehicles with orders within t hours is denoted as ori(x). The target definition method can dynamically adjust the supply-demand imbalance relationship between areas, such that: A: In a hot spot area within a certain period of time, if arr(x) < ori(x) persists, resulting in a continuous decrease in park(x) and a shortage of vehicles, after a certain period of time, other areas will ultimately be scheduled to increase park(x); B: In a cold spot area within a certain period of time, if arr(x) > ori(x) persists, resulting in a continuous increase in park(x) and vehicle congestion, after a certain period of time, other areas will ultimately be scheduled to decrease park(x); The solution result is R(x, y), which represents the number of dispatched vehicles from area x to area y.

[0015] Preferably, in S2, by defining the distance between area x and area y as distance(x, y) < a, then x and y are adjacent areas, and a is the adjacent area coefficient. The set of adjacent areas of area x is denoted as S(x). For each element y in S(x), 1 ≤ y ≤ n, y ≠ x, and distance(x, y) < a.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0017] The present invention can solve the problem of unbalanced regional scheduling, and the adopted water ripple algorithm brings global dynamic scheduling, with advantages such as fast scheduling speed, automatic scheduling, and no need for manual intervention, which can meet the requirements of automatic and intelligent operation of autonomous driving taxis. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flowchart of the scheduling method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0020] Please refer to Figure 1 , the present invention provides a technical solution: a water ripple scheduling method for the position of autonomous driving taxis, including the steps:

[0021] S1: Establish a mathematical model for the scheduling problem;

[0022] S3: Calculate the set of adjacent areas for each area;

[0023] S6: Calculate the vehicle idle rate and turnover rate;

[0024] S4: Perform ripple adjustment and diffusion based on the ripple algorithm to achieve global simultaneous mobilization;

[0025] S5: Algorithm initialization;

[0026] S15: Execute the ripple algorithm for the p-th round;

[0027] S18: Ripple iterative calculation, ripple iteration p = p + 1, if 1 ≤ p ≤ m - 1, go to S15;

[0028] S8: Calculate the scheduling data and output the scheduling result.

[0029] In this embodiment, in S1, there are n autonomous driving parking areas set, and autonomous driving taxis can be charged within this area. Then the position coordinates of the area can be denoted as (lon, lat); the number of parking spaces available in area x is denoted as cap(x), and the number of vehicles currently parked in this area is denoted as park(x), where the area number 1 ≤ x ≤ n; then the number of vehicles arriving at area x within t hours is denoted as arr(x), and the number of orders dispatched within t hours is denoted as ori(x); the target definition method can dynamically adjust the supply-demand imbalance relationship between areas, such that: A: In a hot area within a certain period of time, if arr(x) < ori(x) persists, resulting in a continuous decrease in park(x) and a shortage of vehicles, after a certain period of time, finally, other areas will be scheduled to increase park(x); B: In a cold area within a certain period of time, if arr(x) > ori(x) persists, resulting in a continuous increase in park(x) and vehicle congestion, after a certain period of time, finally, other areas will be scheduled to decrease park(x); The solution result is R(x, y), which represents the number of vehicles scheduled from area x to area y.

[0030] In this embodiment, in S2, by defining the distance between area x and y as distance(x, y) < a, then x and y are adjacent areas to each other, and a is the adjacent area coefficient; the set of adjacent areas of area x is denoted as S(x). For each element y in S(x), then y satisfies 1 ≤ y ≤ n, y ≠ x, and distance(x, y) < a.

[0031] In this embodiment, in S3, after t hours, the idle rate of the area is f(x) = (cap(x) - park(x)) / cap(x);

[0032] After t hours, the regional transfer rate is g(x) = (ori(x) - arr(x)) / (t * cap(x));

[0033] Sort regions 1...n in ascending order according to the idle rate, and divide them into m equal parts, denoted as set H 1 , H 2 ...H m ; H 1 is the set with the lowest idle rate, 3 ≤ m ≤ 7;

[0034] Then the average regional transfer rate

[0035] In this embodiment, in S4, the water ripple algorithm adopts multi-round solutions, m is the number of solution rounds, that is, in the first round, the region H with the lowest idle rate 1 is moved to other regions with adjacent idle rates; in the second round, H 2 is moved to other regions with adjacent idle rates other than H 1 ; in the third round, H 3 is moved to other regions with adjacent idle rates other than H 1 , H 2 ; and so on. In the last round, H m-1 is moved to the adjacent region of H m and diffuses through the water ripple to achieve global simultaneous mobilization.

[0036] In this embodiment, in S5, for the algorithm initialization, for x = 1...n, in(x) = 0, out(x) = 0, in(x) represents the number of vehicles to be scheduled to enter the calculation, out(x) represents the number of vehicles to be scheduled to go out of the calculation, and p = 1;

[0037] In this embodiment, the water ripple algorithm in the p-th round in S6 is as follows, where 1 ≤ p ≤ m - 1;

[0038] The first step: For each region x in H p , clear T(x), screen its neighbor set S(x), if y ∈ S(x), the condition for y to be selected is:

[0039] a)

[0040] b) f(x) + α < f(y), α is the idle control factor;

[0041] c) g(y) > 0;

[0042] If y is selected, store it in the set T(x), and T(x) is the optional movement set for x;

[0043] Step 2: For each movable area y in the set T(x), calculate the movable vehicles M(x,y) from x to y and the receivable vehicles A(y) of y;

[0044] M(x,y)=park(x)*β*g(x) / Where β is the scheduling factor. Out(x) is not considered here because x is called out for the first time according to the watermark algorithm, and out(x) = 0 at this time.

[0045] A(y)=cap(x)-park(x)-in(y);

[0046] M(x,y) and A(y) are both rounded down;

[0047] Step 3: Construct a bipartite graph for all x and all y in T(x). The bipartite graph of x and y is constructed as M(x,y) departure nodes (these nodes belong to x), A(y) arrival nodes (these nodes belong to y), and the edge value is 1, where 1 means that there is 1 car that can move from x to y.

[0048] Step 4: Solve the bipartite graph based on the Hungarian algorithm to maximize the number of edges matched. x may match multiple different y, and the same, multiple different x may match one y; p In x, count all y1, y2… that x matches, the number of nodes of y1, y2… that x matches is z1, z2…, then out(x, y1) = z1, out(x, y2) = z2…;

[0049] in(y1)+=z1; in(y2)+=z2….

[0050] In this embodiment, in S8, for x, 1≤x, y≤n, y≠x;

[0051] Calculate R(x,y)=out(x,y); R(x,y) is the result that should be scheduled within t hours of this cycle.

[0052] The water ripple scheduling method of the present invention solves the problem that subsequent matching of vehicles and passengers is difficult and cannot be restored due to the imbalance of hot and cold spots in the region. The method uses multiple rounds of water ripple diffusion to enable global dynamic scheduling. The middle area can bear the function of the channel and transfer its own vehicles to the area with more scarce capacity, while receiving the scheduling of the upper area with abundant capacity. Compared with the existing technical methods, the scheduling efficiency is significantly improved.

[0053] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for position water pattern scheduling of autonomous taxis, characterized in that, it includes the steps: S1: Establish a mathematical model for the scheduling problem; S2: Calculate the set of adjacent areas for each area; S3: Calculate the vehicle idle rate and turnover rate; In S3, after t hours, the area idle rate is f(x) = (cap(x) - park(x)) / cap(x); After t hours, the area turnover rate is g(x) = (ori(x) - arr(x)) / (t * cap(x)); Sort regions 1…n in ascending order according to the idle rate, divide them into m equal parts, and record them as set H 1 , H 2 …H m ;H 1 It is the set with the lowest idle rate, 3≤m≤7; The average regional turnover rate S4: Perform water pattern mediation diffusion based on the water pattern algorithm to achieve global simultaneous mobilization; The watermark algorithm in S4 adopts multiple rounds of solving, m is the number of solving rounds, that is, in the first round, the lowest idle rate area H is firstly 1 Move to other areas with similar idle rates; in the second round, move H 2 Xiang H 1 Other idle rates other than the adjacent areas are moved; in the third round, H 3 Xiang H 1 , H 2 Other idle rate neighboring areas other than H are moved; and so on, in the last round, H m-1 Xiang H m Move adjacent areas and spread through water ripples to achieve global simultaneous mobilization; S5: Algorithm initialization; In S5, for the algorithm initialization, for x = 1...n, in(x) = 0, out(x) = 0, in(x) represents the number of vehicles to be scheduled to enter the calculation, out(x) represents the number of vehicles to be scheduled to go out of the calculation, and p = 1; S6: Execute the water pattern algorithm for the p-th round; The water pattern algorithm for the p-th round in S6 is as follows, where 1 ≤ p ≤ m - 1; Step 1: H p For each region x in , clear T(x) and filter its neighbor set S(x). If y∈S(x), the condition for y to be selected is: a) b) f(x) + α < f(y), α is the idle control factor; c) g(y) > 0; If y is selected, it is stored in the set T(x), and T(x) is the optional movement set for x; Second step: For each movable area y in the set T(x), calculate the movable vehicles M(x, y) from x to y and the receivable vehicles A(y) of y; Where β is the scheduling factor. Out(x) is not considered here because x is called out for the first time according to the watermark algorithm, and out(x) = 0 at this time. A(y) = cap(x) - park(x) - in(y); Both M(x, y) and A(y) are rounded down; Third step: Construct a bipartite graph for all x and all y in T(x). The construction of the bipartite graph of x and y is specifically M(x, y) departure nodes (these nodes belong to x), A(y) arrival nodes (these nodes belong to y), and the edge value is 1. Here, 1 represents that there is 1 vehicle that can move from x to y; Step 4: Solve the bipartite graph based on the Hungarian algorithm to maximize the number of edges matched; x may match multiple different y, and the same, multiple different x may match one y; p In x, count all y1, y2… that x matches, the number of nodes of y1, y2… that x matches is z1, z2…, then out(x, y1) = z1, out(x, y2) = z2…; in(y1) += z1; in(y2) += z2...; S7: Water pattern iterative calculation; S8: Calculate the scheduling data and output the scheduling result; In S8, for x, 1 ≤ x, y ≤ n, y ≠ x; Calculate R(x, y) = out(x, y); R(x, y) is the scheduling result that should be scheduled within the current period of t hours.

2. A method for position water pattern scheduling of autonomous taxis according to claim 1, characterized in that, In S1, n automatic driving parking areas are set, and automatic driving taxis can be charged within these areas. The position coordinates of the areas can be denoted as (lon, lat); the available parking capacity of area number x is denoted as cap(x), and the number of currently parked vehicles in this area is denoted as park(x), where the area number satisfies 1 ≤ x ≤ n; the number of arriving vehicles in area x within t hours is denoted as arr(x), and the number of dispatched vehicles with orders within t hours is denoted as ori(x); the target definition method can dynamically adjust the supply-demand imbalance relationship between areas, such that: A: In a hot area within a certain period of time, if arr(x) < ori(x) persists, resulting in a continuous decrease in park(x) and a shortage of vehicles, after a certain period of time, finally, park(x) will increase by dispatching vehicles from other areas; B: In a cold area within a certain period of time, if arr(x) > ori(x) persists, resulting in a continuous increase in park(x) and vehicle congestion, after a certain period of time, finally, park(x) will decrease by dispatching vehicles from other areas; the solution result is R(x, y), which represents the number of dispatched vehicles from area x to area y.

3. An automatic driving taxi position water pattern scheduling method according to claim 1, characterized in that in S2, by defining that the distance distance(x, y) between area x and area y < a, then x and y are adjacent areas to each other, and a is the adjacent area coefficient; the set of adjacent areas of area x is denoted as S(x), and for each element y in S(x), y satisfies 1 ≤ y ≤ n, y ≠ x, and distance(x, y) < a.

Images