A trajectory optimization method for intelligent networked vehicles driving on urban bus lanes

Abstract

The application provides a kind of intelligent network connection vehicle (Connected Automated Vehicle, hereinafter referred to as CAV) trajectory optimization method for driving on city bus lane. In order to improve the mixed traffic flow state of CAV and manually driven vehicle, and in order to effectively improve the operation efficiency of CAV, the operation gap of bus lane is used to improve the operation speed of CAV, so as to realize the separation of CAV and manually driven vehicle. In the trajectory planning process, firstly, the time when CAV arrives is obtained, based on the traffic state at this time, after considering the influence of the empty capacity on the city bus lane and the signal, whether CAV can run on the bus lane is reasonably arranged, and then the trajectory of CAV running on the bus lane is solved and optimized. The method of the application is a feasible and quite economical planning method, which not only ensures the independent operation of CAV to a certain extent, but also utilizes the idle part of traffic resources.

Description

Technical Field

[0001] The present invention relates to the field of CAV trajectory planning, and in particular to a CAV trajectory planning method for traveling on a city bus lane. Background Art

[0002] The rapid development of CAVs in recent years has led to increasing academic attention on the issue of mixed traffic between intelligent connected vehicles and human-driven vehicles. Consequently, a growing number of optimization solutions have been proposed.

[0003] High-level autonomous connected vehicles can theoretically have a significant impact on the stability and safety of real-world traffic flows and help optimize the traffic environment, making them a promising development direction for future transportation systems. However, given the current level of CAV technology, their direct involvement in traffic flows primarily composed of ordinary vehicles will disrupt the stability of that flow. Furthermore, due to the low penetration of CAV technology in traffic flows, these disruptions will be highly random and prone to safety issues. Furthermore, if CAV technology levels do not meet specific market share requirements, overall road performance will not improve, but will instead decrease the overall operational efficiency of the road network.

[0004] To address the CAV problem, the idea of ​​establishing dedicated CAV lanes was first proposed. The core idea is to completely separate CAVs from conventional vehicles, creating a completely closed, stable driving environment for CAVs and ensuring their safe operation. However, the main problem is that building dedicated CAV lanes inevitably squeezes space for conventional vehicles. Furthermore, when CAVs have a low market share, these lanes cannot be fully utilized, resulting in unnecessary waste of road resources.

[0005] Considering the infrastructure construction costs of building CAV-dedicated lanes, if the existing transportation infrastructure can be utilized to build a relatively more independent and stable operating environment for CAVs, the construction costs can be greatly reduced. Moreover, if the underutilized resources in the existing road environment are utilized well, it will also help to improve the resource waste phenomenon in the transportation system.

[0006] Bus lanes offer a new approach to separating CAVs from human-driven vehicles. In permanent bus lanes, some road resources are not fully utilized. While they ensure public transportation on a single dedicated lane, they can sometimes reduce the overall traffic efficiency of the entire road. Summary of the Invention

[0007] The purpose of the present invention is to address the deficiencies of the existing technology and provide a CAV trajectory planning method for traveling on urban bus lanes. The method takes into account multiple factors and can optimize the CAV trajectory traveling on bus lanes.

[0008] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0009] A CAV trajectory planning method for traveling on a city bus lane comprises the following steps:

[0010] c1. Divide the space-time grid of trajectory planning and sort it;

[0011] c2. Obtain the CAV arrival time and estimated speed sequence for each time segment to be optimized;

[0012] c3. Obtain the current state of the bus lane, that is, obtain a 0\1 variable sequence indicating whether the studied section is passable at all time points within the optimized section;

[0013] c4. Determine whether the CAV can enter the bus lane based on the results of c2 and c3, and establish the total sequence of the CAV's initial arrival speed;

[0014] c5. Using the total sequence of CAV initial arrival speeds established in c4 as the initial speed constraint, model the CAV trajectory by comprehensively considering the existence of bus bays, bus trajectories, and traffic light conditions;

[0015] c6. Solve the CAV trajectory optimization model obtained in c5 sequentially in time and space. The solution result of the model of the previous grid is used as the input of the model of the next grid. After all grids are solved, all results are merged to obtain the optimized trajectory matrix.

[0016] In the above technical solution, further, the process of step c1 includes:

[0017] First, the space-time grid is divided based on a certain time scale and a certain space scale. The time interval between bus arrivals is used as the time benchmark, and the space-time segment length is used as the space benchmark. Then, the space-time segments in each grid are arranged in time sequence and spatial order according to the direction of CAV travel, with spatial position as the priority. In this way, the problem is transformed from an optimization problem of the entire optimization period and the entire lane into a series of linear optimization problems within a single space-time grid.

[0018] Furthermore, the process of step c2 includes:

[0019] c21. Within an optimized time segment, obtain the corresponding time information of all CAVs arriving at the bus lane entrance within this time segment and mark the priority of the CAVs according to their entry direction. The priority of straight-going CAVs is higher than that of turning CAVs. The straight-going CAVs and turning CAVs form their own arrival time series.

[0020] c22. Within an optimized time segment, obtain the estimated speed information of all CAVs arriving at the bus lane entrance within this time segment, and construct a corresponding speed sequence according to the time sequence.

[0021] Furthermore, the process of step c3 includes:

[0022] c31. Obtain the bus flow on the bus lane within the optimized time range and calculate the remaining lane capacity and the time window available for CAVs to pass;

[0023] c32. Obtain the timing information of the intersection lights on this road section within the optimization time range and calculate the time window for the CAV to pass;

[0024] c33. Take the intersection of the time windows calculated by c31 and c32, and finally form a 0\1 variable sequence indicating whether all time points in the optimization segment are passable. If it is allowed to pass, it is 1, and if it is not allowed to pass, it is 0.

[0025] Furthermore, the process of step c4 includes:

[0026] c41. Construct a blank sequence with the length of the optimization segment as the initial speed sequence of the optimization process;

[0027] c42. First, based on the priority tag, the time nodes in the arrival time sequence of the straight CAV are compared with the 0\1 variable sequence constructed in step c3 in order. If the corresponding value in the 0\1 variable sequence is 0 at the time point corresponding to the CAV's arrival, the CAV is removed from the bus lane. If the corresponding value in the 0\1 variable sequence is 1, it is changed to 0, and the corresponding initial arrival speed is filled in the corresponding position of the blank sequence constructed in c41. Finally, the entire 0\1 variable sequence and the total initial arrival speed sequence are updated.

[0028] c43. The CAV arrival time sequence of the turning arrival and the 0\1 variable sequence updated by c42 are updated again according to the judgment process of c42 to obtain the CAV initial arrival speed sequence.

[0029] Furthermore, the process of step c5 includes:

[0030] c51. Obtain the existence of bus bays in the studied time and space segment. If a bay exists, the CAV can overtake the bus at the bay; if not, the CAV cannot overtake the bus.

[0031] c52. Using the total sequence of CAV initial arrival speeds obtained in c4 as the initial speed constraint, define the CAV's dynamic performance as the physical constraint, and the bus trajectory and traffic lights as the edge constraints.

[0032] Based on the minimum CAV travel time, the objective function is constructed, specifically, the travel time of CAV vehicles in all sections is accumulated and the minimum is obtained.

[0033] Furthermore, the process of step c6 includes:

[0034] c61. Establish the feature sequences and input data sets corresponding to the sub-models in all spatiotemporal grids. The feature sequences include all fixed physical parameters covered by the spatiotemporal grid where the sub-models are located. The input data mainly refers to the CAV traffic flow added at the node and the CAV traffic flow leaving the lane at the node.

[0035] c62. Build a model that includes an objective function, a decision variable matrix, and a set of general constraints. The objective function of c5 is constructed in the model based on the decision variables: p = min(f(V)), where V is a matrix containing the CAV positions at all time points, f(V) is the total travel time calculated using a linear function, and p is the reconstructed objective function. All constraints are also reconstructed based on the decision variables, with the general form A(V) ≤ c, where A(V) is a linear transformation of V and c is the maximum value of constraint A(V).

[0036] c63, input the feature sequence and input data set corresponding to each sub-problem into the solution model according to the order in c1;

[0037] c64. Analyze the transfer of the results in time and space, and divide the results into two parts, one for input to two adjacent grids in time and space;

[0038] c65. Select the problem in the next grid, merge its feature sequence and initial input with the output of the corresponding problem in the adjacent grid space and time that has been solved, update the initial input of this problem, and repeat steps c63 and c64;

[0039] c66. Combine the outputs of all sub-problems to form the final optimization result.

[0040] The present invention has the following beneficial effects:

[0041] By comprehensively considering bus trajectories and the influence of urban traffic lights, the method of the present invention establishes a hybrid lane model on urban roads that allows for use of bus lanes and traffic lights. This method, taking into account multiple factors, optimizes the trajectory of CAVs using bus lanes. Furthermore, the method considers the influence of bus bays, where CAVs can overtake, thereby fully leveraging their relative advantages. This method can improve the operational efficiency of traffic flow. Furthermore, the present invention utilizes a solution method similar to greedy piecewise solution on a spatiotemporal grid, achieving high computational efficiency.

[0042] The application can provide a new traffic management policy for the traffic management system. In the current stage of intelligent network connected vehicle development, the method can better adapt to the improvement of the traffic management system in this stage, and can avoid the high economic cost and potential traffic resource waste risk caused by the establishment of a special lane for unmanned vehicles. Moreover, the method has good application prospects in the context of mixed driving of CAV and human-driven vehicles in the future. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 A schematic diagram of the technical route of the method of the application is shown in the figure.

[0044] Figure 2 A schematic diagram of the physical information of a section of bus lane in a specific example of the application is shown in the figure.

[0045] Figure 3 A schematic diagram of the trajectory of a CAV on a corresponding road section after planning by the method of the application is shown in the figure. Figure 2 A schematic diagram of the trajectory of a CAV on a corresponding road section after planning by the method of the application is shown in the figure. DETAILED DESCRIPTION

[0046] The technical solutions of the application will be further described in detail below with reference to the accompanying drawings and specific examples.

[0047] The CAV trajectory planning method of the application uses the city bus lane, as shown in the figure, which mainly includes the following steps: Figure 1

[0048] c1, dividing the space-time grid for trajectory planning and sorting;

[0049] First, the space-time grid based on a certain scale of time and a certain scale of space is divided, the time interval of the arrival of the bus is taken as the basis on the time, and the length of the single lane section is taken as the basis on the space, then the time-space section in each grid is arranged in the order of time from late to early and space in the forward direction of the CAV, and the space position is given priority. Thus, the trajectory planning problem becomes a series of linear optimization problems.

[0050] c2, obtaining the time of arrival and the speed sequence of the CAV for each optimization time section; specifically including:

[0051] c21, in an optimization time section, obtaining the time information of all CAVs arriving at the entrance of the bus lane in this time section, and marking the priority according to the CAV entering direction, straight entering priority over turning entering, and straight CAV and turning CAV forming their own arrival time sequence respectively;

[0052] ​c22. Within an optimized time segment, obtain the estimated speed information of all CAVs arriving at the bus lane entrance within this time segment, and construct a corresponding speed sequence according to the time sequence.

[0053] c3. Obtain the current state of the bus lane, that is, obtain a 0\1 variable sequence indicating whether the studied section is passable at all time points within the optimized section. This process includes:

[0054] c31. Obtain the bus flow on the bus lane within the optimized time range and calculate the remaining lane capacity and the time window available for CAVs to pass;

[0055] c32. Obtain the timing information of the intersection lights on this road section within the optimization time range and calculate the time window for the CAV to pass;

[0056] c33. Take the intersection of the time windows calculated by c31 and c32, and finally form a 0\1 variable sequence indicating whether all time points in the optimization segment are passable. If it is allowed to pass, it is 1, and if it is not allowed to pass, it is 0.

[0057] c4. Determine whether the CAV can enter the bus lane. The process includes:

[0058] c41. Construct a blank sequence with the length of the optimization segment as the initial speed sequence of the optimization process;

[0059] c42. First, based on the priority tag, the time nodes in the arrival time sequence of the straight CAV are compared with the 0\1 variable sequence constructed in step c3 in order. If the corresponding value in the 0\1 variable sequence is 0 at the time point corresponding to the CAV's arrival, the CAV is removed from the bus lane. If the corresponding value in the 0\1 variable sequence is 1, it is changed to 0, and the corresponding initial arrival speed is filled in the corresponding position of the blank sequence constructed in c41. Finally, the entire 0\1 variable sequence and the total initial arrival speed sequence are updated.

[0060] c43. The CAV arrival time sequence of the turning arrival and the 0\1 variable sequence updated by c42 are updated again according to the judgment process of c42 to obtain the CAV initial arrival speed sequence.

[0061] c5. Model the CAV trajectory, including:

[0062] c51. Obtain the existence of a bus bay in this time and space segment. If a bus bay exists, the CAV can overtake the bus at the bay; if not, the CAV cannot overtake the bus.

[0063] The initial arrival speed total sequence of c52, c42 is taken as the initial speed constraint, the dynamic performance of CAV is defined as the physical constraint, the bus trajectory and the signal light are taken as the edge constraint, and the objective function is constructed according to the minimum CAV travel time:

[0064]

[0065] Constraints:

[0066]

[0067]

[0068]

[0069]

[0070] π≥0,λ≥0,μ≥0

[0071] T=nC i n∈N*

[0072] Wherein, is the traffic on the optimization section at a specific time stage, is the travel time spent by vehicles in the traffic on the section per unit time; t, is the time pair of the arrival time to the departure time of the target vehicle on the section, T a is the travel time of all vehicles, and α represents the section where the traffic is located; all the travel times of the vehicles on all sections A b are added up, and the minimum is calculated, which constitutes the objective function of the model; in constraint (3.4), D is the traffic demand on the studied lane section, λ and μ are the CAV traffic entering and leaving at a specific OD or section, z α is the travel time of CAV on the road section, T is the time length of the space-time grid, which is equal to the bus departure interval, and C is the time length of the signal cycle.

[0073] c6, the model is established and solved, and the process includes:

[0074] c61, the characteristic sequence and the input data set corresponding to all problems are established. The input data mainly refer to the CAV traffic added at the node and the CAV traffic leaving the lane at the node;

[0075] c62. Establish a model that includes an objective function, a decision variable matrix, and a set of general constraints. The objective function of c5 is constructed in the model in the form of decision variables: p = min(f(V)), where V is a matrix containing the CAV positions at all time points, f(V) is the total travel time calculated by a linear function, and p is the reconstructed objective function; all constraints are also reconstructed based on the decision variables, in the general form A(V) ≤ c, where A(V) is a linear transformation of V and c is the maximum value of constraint A(V); c63. Input the feature sequence and input data set corresponding to each subproblem into the solution model in order;

[0076] c64. Analyze the transfer of the results in time and space, and divide the results into two parts, one for input to two adjacent grids in time and space;

[0077] c65. Select the problem in the next grid, merge its feature sequence and initial input with the output of the corresponding problem in the adjacent grid space and time that has been solved, update the initial input of this problem, and repeat steps c63 and c64;

[0078] c66. Combine the outputs of all sub-problems to form the final optimization result.

[0079] Application examples:

[0080] After searching and comparing multiple bus lane cases distributed in Hangzhou and surrounding cities, a part of the bus lanes in Hangzhou was finally selected as the object of the application example of the present invention and numerical verification was carried out under actual data. Figure 2 The physical information of this section of bus lane is indicated in the figure.

[0081] Based on the above example, the method of the present invention is used to establish a model and solve the problem to obtain the CAV trajectory on the corresponding road section, such as Figure 3 As shown in FIG, which shows the CAV trajectory, bus trajectory, and signal light information, it can be seen that the present invention successfully achieves the optimization of the CAV trajectory when traveling in the bus lane, and the optimization process is highly efficient.

Claims

1. A CAV trajectory optimization method for traveling on urban bus lanes, characterized in that: The method comprises the following steps: c1. Divide and sort the spatiotemporal grids for trajectory planning. First, divide the spatiotemporal grids based on a certain temporal scale and a certain spatial scale. In terms of time, the time interval between bus arrivals is used as the benchmark, and in terms of space, the length of a single lane segment is used as the benchmark. Then, the spatiotemporal segments in each grid are arranged in temporal order and in spatial order based on the direction of travel of the CAV, with spatial position as the priority. This transforms the optimization problem of the entire optimization period and the entire lane into a series of linear optimization problems within each spatiotemporal grid. c2. Obtain the CAV arrival time and estimated speed sequence for each time segment to be optimized; c3. Obtain the current state of the bus lane, that is, obtain a 0\1 variable sequence indicating whether the studied section is passable at all time points within the optimized section; including: c31. Obtain the bus flow on the bus lane within the optimized time range and calculate the remaining lane capacity and the time window available for CAVs to pass; c32. Obtain the timing information of the intersection lights on this road section within the optimization time range and calculate the time window for the CAV to pass; c33, take the intersection of the time windows calculated by c31 and c32, and finally form a 0\1 variable sequence indicating whether all time points in the optimization section are passable, with 1 for passable and 0 for non-passable; c4. Determine whether the CAV can enter the bus lane based on the results of c2 and c3, and establish the total sequence of the CAV's initial arrival speed; C5. Using the total sequence of CAV initial arrival speeds established in C4 as the initial speed constraint, model the CAV trajectory by comprehensively considering the existence of bus bays, bus trajectories, and traffic light conditions. This includes: c51. Obtain the existence of bus bays in the studied time and space segment. If a bay exists, the CAV can overtake the bus at the bay; if not, the CAV cannot overtake the bus. c52. Using the total sequence of CAV initial arrival speeds obtained in c4 as the initial speed constraint, define the CAV's dynamic performance as the physical constraint, and the bus trajectory and traffic lights as the edge constraints. Based on minimizing the CAV travel time, construct an objective function. Specifically, the travel time of CAV vehicles in all sections is accumulated and minimized. c6. Solve the CAV trajectory optimization model obtained in c5 sequentially in time and space. The solution result of the model of the previous grid is used as the input of the model of the next grid. After all grids are solved, all results are merged to obtain the optimized trajectory matrix.

2. The CAV trajectory optimization method according to claim 1, characterized in that: The process of step c2 includes: c21. Within an optimized time segment, obtain the corresponding time information of all CAVs arriving at the bus lane entrance within this time segment and mark the priority of the CAVs according to their entry direction. The priority of straight-going CAVs is higher than that of turning CAVs. The straight-going CAVs and turning CAVs form their own arrival time series. c22. Within an optimized time segment, obtain the estimated speed information of all CAVs arriving at the bus lane entrance within this time segment, and construct a corresponding speed sequence according to the time sequence.

3. The CAV trajectory optimization method according to claim 1, characterized in that: The process of step c4 includes: c41. Construct a blank sequence with the length of the optimization segment as the initial speed sequence of the optimization process; c42. First, based on the priority tags, the time nodes in the arrival time sequence of the straight CAV are compared with the 0\1 variable sequence constructed in step c3. If the corresponding value in the 0\1 variable sequence is 0 at the time point corresponding to the CAV's arrival, the CAV is removed from the bus lane. If the corresponding value in the 0\1 variable sequence is 1, it is changed to 0, and the corresponding initial arrival speed is filled in the corresponding position of the blank sequence constructed in c41. Finally, the entire 0\1 variable sequence and the total initial arrival speed sequence are updated. c43. The CAV arrival time sequence of the turning arrival and the 0\1 variable sequence updated by c42 are updated again according to the judgment process of c42 to obtain the CAV initial arrival speed sequence.

4. The CAV trajectory optimization method according to claim 1, characterized in that: The process of step c6 includes: c61. Establish the feature sequences and input data sets corresponding to the sub-models in all spatiotemporal grids. The feature sequences include all fixed physical parameters covered by the spatiotemporal grid where the sub-models are located. The input data mainly refers to the CAV traffic flow added at the node and the CAV traffic flow leaving the lane at the node. C62. Build a model, including the objective function, decision variable matrix, and general constraint set. The objective function of C5 is constructed in the model based on the decision variables: ,in V is the matrix containing the CAV positions at all time points, is the total travel time calculated as a linear function, p is the reconstructed objective function; all constraints are also reconstructed based on the decision variables, and the general form is ,in For V The linear transformation of c For constraints The maximum value of c63, input the feature sequence and input data set corresponding to each sub-problem into the solution model according to the order in c1; c64. Analyze the transfer of the results in time and space, and divide the results into two parts, one for input to two adjacent grids in time and space; c65. Select the problem in the next grid, merge its feature sequence and initial input with the output of the corresponding problem in the adjacent grid space and time that has been solved, update the initial input of this problem, and repeat steps c63 and c64; c66. Combine the outputs of all sub-problems to form the final optimization result.

Images