Traffic risk assessment method, electronic equipment, storage medium and computer program product for unmanned delivery vehicle deployment test

Abstract

The present invention discloses a method for assessing traffic risks in testing the deployment of unmanned delivery vehicles, including: step one, extracting static elements and dynamic elements of road collection data, and constructing a road feature library; step two, defining road network parameters and traffic flow parameters in each time period, deploying unmanned delivery vehicles, setting safety officer ratios, and conducting simulation experiments; extracting simulation result data, and calculating accident indexes and congestion indexes in each time period; step three, constructing a data set, and using a neural network to construct a road-level risk assessment model; step four, based on the road-level risk assessment model, determining the deployment amount of unmanned delivery vehicles and safety officer ratio limit values ​​in each time period. The present invention also discloses an electronic device, a storage medium, and a computer program product. The present invention can rely on real road scenes, and by simulating the perception of real unmanned delivery vehicles on the road traffic environment, determine the human-vehicle ratio, deployment time, and deployment amount limit values ​​after the batch deployment of unmanned delivery vehicles.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned delivery vehicle deployment testing. More specifically, the present invention relates to a traffic risk assessment method, electronic device, storage medium and computer program product for unmanned delivery vehicle deployment testing. Background Art

[0002] Unmanned delivery vehicles are new smart logistics products that have emerged in the development of the autonomous driving industry. They are officially "on duty" on more and more public roads, and can meet operational requirements and provide good services in specific areas. However, due to the short development time of its technology, there are still technical deficiencies. If it is to be fully promoted, it is necessary to further predict the unknown risks after the mass deployment of unmanned delivery vehicles, study the control rules of the ratio of people and vehicles, the time of deployment, and the amount of deployment after the mass deployment of unmanned delivery vehicles, as well as the impact on road traffic. This is the main technical problem to be solved by this application. Summary of the invention

[0003] The present invention provides a traffic risk assessment method, electronic device, storage medium and computer program product for unmanned delivery vehicle deployment testing, which can rely on real road scenes, simulate the perception of real unmanned delivery vehicles on the road traffic environment, and determine the human-vehicle ratio, deployment time and deployment quantity limit after the batch deployment of unmanned delivery vehicles, thereby improving service effects and minimizing the impact on road traffic congestion and accidents.

[0004] In order to achieve these purposes and other advantages according to the present invention, a method for evaluating traffic risk in an unmanned delivery vehicle deployment test is provided, comprising:

[0005] Step 1: extract static elements and dynamic elements of road collection data and build a road feature library;

[0006] Step 2: Define the road network parameters and traffic flow parameters for each time period, deploy unmanned delivery vehicles, set the safety officer ratio, conduct simulation experiments, and statistically control the behavior of the unmanned delivery vehicles in the simulation experiments based on the test results of the unmanned delivery vehicles in various dangerous scenarios, including deceleration, emergency stop, collision, and calculate the traffic congestion index TPI;

[0007] Extract the simulation result data and calculate the accident index Y1 and congestion index Y2 in each time period:

[0008]

[0009]

[0010]

[0011] Among them, τ in formula (1) jis the probability of a dangerous scenario occurring in the jth experiment, d is the number of dangerous scenarios, c is udv is the number of unmanned delivery vehicles deployed in the jth experiment, k i is the number of dangerous scenarios that occurred in the i-th unmanned delivery vehicle in the j-th experiment; n in formula (2) is the number of experimental repetitions;

[0012] Step 3: The time period of a road experiment, road network parameters, traffic flow parameters, number of unmanned delivery vehicles, safety officer ratio, accident index Y1 and congestion index Y2 constitute a piece of data;

[0013] Construct a data set, divide the data into training set, test set, and validation set, and use a neural network to build a road-level risk assessment model. The input layer of the neural network includes time period, road network parameters, traffic flow parameters, number of unmanned delivery vehicles, and ratio of safety personnel. The output layer includes the accident index Y1 and congestion index Y2 of the road.

[0014] Step 4: Set the accident index Y1 threshold and congestion index Y2 threshold of the road, and determine the limit values ​​of the deployment volume of unmanned delivery vehicles and the number of safety personnel in each time period based on the road-level risk assessment model.

[0015] Preferably, it also includes:

[0016] Step 5: Set a unit within a certain range, define the road network parameters and traffic flow parameters of the unit, deploy unmanned delivery vehicles, set the safety officer ratio, conduct simulation experiments, and select the roads that are most frequently passed by unmanned delivery vehicles as thermal roads;

[0017] Taking the thermal road as the object, a simulation experiment is carried out, the simulation result data is extracted, and the accident index Y1 and congestion index Y2 are calculated to represent the accident index Y1 and congestion index Y2 of the unit to which it belongs;

[0018] Step 6: Repeat step 5 to collect multiple unit data. The road network parameters, traffic flow parameters, number of unmanned delivery vehicles, safety officer ratio, accident index Y1 and congestion index Y2 of one unit constitute one piece of data.

[0019] Construct a data set, divide the data into training set, test set, and validation set, and use a neural network to build a regional risk assessment model. The input layer of the neural network includes the unit's road network parameters, traffic flow parameters, number of unmanned delivery vehicles, and safety officer ratios. The output layer includes the unit's accident index Y1 and congestion index Y2.

[0020] Step 7: Divide the region into multiple uniform units, calculate the accident index Y1 and congestion index Y2 of each unit based on the regional risk assessment model, and calculate the risk factor Y of the unit;

[0021]

[0022] The congestion and smoothness levels are divided in the interval Y=(0,1). The congestion and smoothness level of the unit is determined according to the unit's risk factor Y. The smoothness rate of the area is determined according to the proportion of smooth units. The smoothness rate threshold is set, and based on the regional risk assessment model, the limit values ​​of the ratio of unmanned delivery vehicles and the number of safety personnel in the area are determined.

[0023] Preferably, the shape of the unit is a regular hexagon, and the area is divided into a plurality of uniform delivery units using a honeycomb network structure.

[0024] Preferably, when a fully connected neural network is used to construct a road-level risk analysis model, the number of hidden layers is 2, the number of nodes is 128, the weight decay coefficient is 0.2, the loss function is the minimum mean square error, and the minimum error threshold is 0.05.

[0025] Preferably, when a fully connected neural network is used to construct a regional risk analysis model, the number of hidden layers is 2, the number of nodes is 128, the weight decay coefficient is 0.2, the loss function is the minimum mean square error, and the minimum error threshold is 0.05.

[0026] Preferably, when a neural network is used to construct a road-level risk analysis model, the road network parameters mainly include road length, proportion of road sections with mixed motor and non-motor vehicle traffic, number of intersections, intersection shape, number of traffic lights, road type, road speed limit, number of two-way motor vehicle lanes, presence or absence of isolation, width of motor vehicle lanes, width of mixed motor and non-motor vehicle lanes, and width of non-motor vehicle lanes; and the traffic flow parameters mainly include the total number of traffic participants, as well as their types, paths, and movement parameters.

[0027] Preferably, when a neural network is used to construct a regional risk analysis model, the road network parameters mainly include the regional area, the length of regional roads, the number of intersections, the number of traffic lights, and the distribution and concentration of road intersections; the traffic flow parameters mainly include the total number of traffic participants, as well as the types, paths, and movement parameters of traffic participants.

[0028] An electronic device comprises: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor executes the described method.

[0029] A storage medium stores a computer program, which, when executed by a processor, implements the method described.

[0030] A computer program product comprises a computer program, which, when executed by a processor, implements the method described.

[0031] The present invention has at least the following beneficial effects:

[0032] First, the present invention can build a road feature library based on real road scenes. The statistical probability of the test results of the unmanned delivery vehicle in various dangerous scenes controls the behavior of the unmanned delivery vehicle in the simulation experiment, simulating the perception of the road traffic environment by the real unmanned delivery vehicle, which is authentic, objective and feasible.

[0033] Second, the present invention extracts simulation result data, constructs a data set, and uses a neural network to construct a road-level risk assessment model to determine the human-vehicle ratio, delivery time, and delivery quantity limit values ​​of unmanned delivery vehicles in various time periods on the road, so as to minimize the impact on road traffic congestion and accidents.

[0034] Third, the present invention sets units within a certain range, conducts simulation experiments with thermal roads as representative units, extracts simulation result data, constructs data sets, and uses neural networks to construct regional risk assessment models to determine the human-vehicle ratio and delivery volume limit values ​​of the unmanned delivery vehicles in the region, so as to minimize the impact of congestion and accidents on regional traffic.

[0035] Other advantages, objectives and features of the present invention will be embodied in part through the following description, and in part will be understood by those skilled in the art through study and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 A road-level unmanned delivery vehicle risk assessment flow chart of a technical solution of the present invention;

[0037] Figure 2 A regional level unmanned delivery vehicle risk assessment flow chart of a technical solution of the present invention;

[0038] Figure 3 A visualization style of an unmanned delivery vehicle in a simulation experiment according to a technical solution of the present invention;

[0039] Figure 4 A model is constructed for the behavior performance of an unmanned delivery vehicle in a simulation experiment according to a technical solution of the present invention;

[0040] Figure 5 An example of an input layer of a road-level risk assessment model constructed by a neural network of a technical solution of the present invention;

[0041] Figure 6 An example of an output layer of a road-level risk assessment model constructed by a neural network of a technical solution of the present invention;

[0042] Figure 7 A schematic diagram of a regional division unit when constructing a regional risk assessment model according to a technical solution of the present invention;

[0043] Figure 8 A schematic diagram of thermal road selection when constructing a road area-level risk assessment model according to a technical solution of the present invention;

[0044] Fig. 9 A schematic diagram of the regional smoothness rate when constructing a road regional level risk assessment model according to a technical solution of the present invention;

[0045] Fig.10 An example of constructing an input layer of a regional risk assessment model for a neural network of a technical solution of the invention;

[0046] Fig.11 An example of an output layer of a regional-level risk assessment model constructed by a neural network of a technical solution of the present invention. DETAILED DESCRIPTION

[0047] The present invention is further described in detail below in conjunction with the accompanying drawings so that those skilled in the art can implement the invention with reference to the description.

[0048] It should be understood that the terms such as "having", "including" and "comprising" used herein do not exclude the existence or addition of one or more other elements or their combinations. It should be noted that the experimental methods described in the following embodiments are conventional methods unless otherwise specified and cannot be understood as limitations of the present invention.

[0049] Through the extensive collection and research of information on policies, standards, technical requirements at all levels for unmanned delivery vehicles, combined with the closed-field test performance of unmanned delivery vehicles and the traffic operation characteristics of non-motorized vehicle lanes and mixed motor and non-motorized vehicle lanes, a comprehensive assessment is made of the impact of unmanned delivery vehicles on road traffic after being deployed on typical roads, typical routes, and typical areas. The traffic risks of unmanned delivery vehicle test demonstration roads are scientifically assessed, and on this basis, a traffic risk assessment method for unmanned delivery vehicle deployment tests has been invented.

[0050] like Figure 1 As shown, the present invention provides a method for evaluating traffic risks in a test of unmanned delivery vehicle deployment, comprising:

[0051] Step 1: extract static elements and dynamic elements of road collection data and build a road feature library;

[0052] In the actual research process, in order to ensure road safety, it is impossible to complete various tasks by running real unmanned delivery vehicles on open roads in the city. Therefore, the existing simulation technology is used to construct and reproduce real scenes, conduct simulation experiments, and realize real road scenes. In this application, various road data of 32 roads are collected, covering non-motorized vehicle lanes and mixed lanes in the city, basically covering the typical road conditions where unmanned delivery vehicles are intended to operate. According to the road data collection, dangerous scenes are collected, sorted, divided, and summarized, and a dangerous scene library that is easy to cause the failure of the operation of unmanned delivery vehicles is established, which includes dangerous scenes of unmanned delivery vehicles. For the road data, the dynamic and static elements of the road are extracted respectively, and the road network structure in the simulation experiment is defined according to the static elements, and the traffic flow in the simulation experiment is defined according to the dynamic elements. The dynamic elements include the type and proportion of traffic participants and traffic flow, and simulation experiments are carried out.

[0053] Step 2: Define the road network parameters and traffic flow parameters for each time period, deploy unmanned delivery vehicles, set the safety officer ratio, conduct simulation experiments, and statistically control the behavior of the unmanned delivery vehicles in the simulation experiments based on the test results of the unmanned delivery vehicles in various dangerous scenarios, including deceleration, emergency stop, collision, and calculate the traffic congestion index TPI;

[0054] Extract the simulation result data and calculate the accident index Y1 and congestion index Y2 in each time period:

[0055]

[0056]

[0057]

[0058] Among them, τ in formula (1) j is the probability of a dangerous scenario occurring in the jth experiment, d is the number of dangerous scenarios, c is udv is the number of unmanned delivery vehicles deployed in the jth experiment, k i is the number of dangerous scenarios that occurred in the i-th unmanned delivery vehicle in the j-th experiment; n in formula (2) is the number of experimental repetitions;

[0059] In order to improve the authenticity and accuracy of the simulation test, the key parameters of dangerous scenarios are output for the dangerous scenario library that may easily cause the operation of unmanned delivery vehicles to fail, including dense non-motor vehicle traffic in the same direction, pedestrians crossing behind parked vehicles on the roadside, motor vehicles crossing, and motor vehicles starting and cutting in on the roadside, so as to construct dangerous scenarios in the simulation experiment.

[0060] The road network structure in the simulation experiment is defined through the road feature library (the road-level road network file and the regional-level road network file can be specified, and the road-level road network file is specified in this technical solution), and the traffic flow parameters of each time period are configured (including but not limited to ① the total number of traffic participants; ② the proportion of each type of traffic participants, including: small passenger cars, pedestrians, bicycles, unmanned delivery vehicles, electric bicycles; ③ total duration (0-3600s); ④ small passenger cars support driving risk ratio configuration: normal, conservative, brave; ⑤ unmanned delivery vehicles use fixed paths, and other traffic flow participants use random paths).

[0061] In one example, when a neural network is used to construct a road-level risk analysis model, the road network parameters mainly include road length, proportion of road sections with mixed motor and non-motor vehicle traffic, number of intersections, intersection shape, number of traffic lights, road type, road speed limit, number of two-way motor vehicle lanes, presence or absence of isolation, width of motor vehicle lanes, width of mixed motor and non-motor vehicle lanes, and width of non-motor vehicle lanes. Traffic flow parameters mainly include the total number of traffic participants, as well as their types, paths, and movement parameters.

[0062] Deployment of unmanned delivery vehicles, such as Figure 3 As shown in the figure, the basic model of the unmanned delivery vehicle is an electric vehicle. On this basis, the UDV (unmanned delivery vehicle) equipment is expanded to make it an unmanned delivery vehicle with the ability to detect scenes in real time and control the behavior of scene processing results. The green car is an unmanned delivery vehicle, the white car is a normal bus, the orange car is a conservative bus, and the red car is an aggressive bus. The figure also shows bicycles and pedestrians, such as Figure 4 As shown in the figure, under the test results of each dangerous scenario, the unmanned delivery vehicle will have specific behaviors, which can be summarized into three major behaviors: deceleration, emergency stop, and collision (emergency stop). When conducting simulation experiments, assuming that a certain scenario occurs, the behavior of the unmanned delivery vehicle can be controlled according to the probability value of the test result corresponding to the scenario, and the behavior of the unmanned delivery vehicle can be indirectly defined by the probability transfer method of the scenario result behavior. Unmanned delivery vehicles may have different results at different times in the same dangerous scenario. Based on the actual vehicle data of unmanned delivery vehicles in the past three years, an analysis was conducted. Table 1 shows some examples. In each scenario description, the test results correspond to three major behaviors: deceleration, emergency stop, and collision (emergency stop). The probability value is used to define the relevant parameters of the unmanned delivery vehicle in the simulation experiment, further ensuring the accuracy of the calibration of the unmanned delivery vehicle in the simulation experiment.

[0063] Table 1

[0064]

[0065]

[0066] When calculating the accident index Y1 and the congestion index Y2 in each time period, the accident index Y1 and the congestion index Y2 are normalized.

[0067] Step 3: The time period of a road experiment, road network parameters, traffic flow parameters, number of unmanned delivery vehicles, safety officer ratio, accident index Y1 and congestion index Y2 constitute a piece of data;

[0068] Construct a data set, divide the data into training set, test set, and validation set, and use a neural network to build a road-level risk assessment model. The input layer of the neural network includes time period, road network parameters, traffic flow parameters, number of unmanned delivery vehicles, and ratio of safety personnel. The output layer includes the accident index Y1 and congestion index Y2 of the road.

[0069] After completing the simulation experiment, the simulation result data is trained through the neural network to establish the road-level risk assessment model for unmanned delivery vehicles. The road network parameters and traffic flow parameters of the input layer of the neural network are as follows: Figure 5 When a fully connected neural network is used to construct a road-level risk analysis model, the number of hidden layers is 2, the number of nodes is 128, the weight decay coefficient is 0.2, the loss function is the minimum mean square error, and the minimum error threshold is 0.05.

[0070] Step 4: Set the accident index Y1 threshold and congestion index Y2 threshold of the road, and determine the limit values ​​of the deployment volume of unmanned delivery vehicles and the number of safety personnel in each time period based on the road-level risk assessment model.

[0071] The deployment time, deployment quantity, and safety officer ratio of unmanned delivery vehicles are derived through reverse deduction. According to the pre-defined safety officer's ability to handle vehicle disengagement, the safety officer ratio is determined. Through simulation experiments, the accident index Y1 threshold and congestion index Y2 of the road are calculated. Under the principle of ensuring that the road is basically unobstructed, the deployment quantity of unmanned delivery vehicles and the safety officer ratio limit values ​​for each time period are given. According to the needs, the actual deployment time, deployment quantity, and safety officer ratio are formulated within the limit value range, such as Figure 6 As shown, taking the thresholds Y1≤0.01 and Y2≤0.2 as an example, an optional solution for deploying unmanned delivery vehicles on this road is: the ratio of people to vehicles is 1:5, the number of deployments is 50, and the deployment time is 00:00-7:00.

[0072] In the above technical solution, simulation result data are extracted through simulation experiments, and a road-level risk assessment model is constructed using a neural network to provide scientific suggestions on the deployment time, deployment volume, and ratio of safety personnel for road-level unmanned delivery vehicles.

[0073] In another technical solution, Figure 2 As shown, it also includes:

[0074] Step 5: Set a unit within a certain range, define the road network parameters and traffic flow parameters of the unit, deploy unmanned delivery vehicles, set the safety officer ratio, conduct simulation experiments, and select the roads that are most frequently passed by unmanned delivery vehicles as thermal roads;

[0075] Taking the thermal road as the object, a simulation experiment is carried out, the simulation result data is extracted, and the accident index Y1 and congestion index Y2 are calculated to represent the accident index Y1 and congestion index Y2 of the unit to which it belongs;

[0076] like Figure 7 As shown in the figure, the shape of the unit is a regular hexagon, and the area is divided into multiple uniform distribution units using a honeycomb network structure. Figure 8 As shown in the figure, the unit is used as the smallest structure to reproduce the real traffic flow in the area in the simulation experiment. Taking the delivery point as the center, the ratio of people and vehicles and the number of unmanned delivery vehicles under certain rules are input to conduct simulation experiments, calculate the road section with the most vehicles passing by at the same time, and then derive the road with the highest congestion index, which is based on the thermal road data and represents the unit.

[0077] Step 6: Repeat step 5 to collect multiple unit data. The road network parameters, traffic flow parameters, number of unmanned delivery vehicles, safety officer ratio, accident index Y1 and congestion index Y2 of one unit constitute one piece of data.

[0078] Construct a data set, divide the data into training set, test set, and validation set, and use a neural network to build a regional risk assessment model. The input layer of the neural network includes the unit's road network parameters, traffic flow parameters, number of unmanned delivery vehicles, and safety officer ratios. The output layer includes the unit's accident index Y1 and congestion index Y2.

[0079] After completing the simulation experiment, the simulation result data is trained through the neural network to establish a regional risk assessment model for unmanned delivery vehicles. The road network parameters and traffic flow parameters of the input layer of the neural network are as follows: Fig.10 When the fully connected neural network is used to construct the regional risk analysis model, the number of hidden layers is 2, the number of nodes is 128, the weight decay coefficient is 0.2, the loss function is the minimum mean square error, and the minimum error threshold is 0.05.

[0080] In one example, when a neural network is used to construct a regional risk analysis model, the road network parameters mainly include the regional area, the length of regional roads, the number of intersections, the number of traffic lights, and the distribution and concentration of road intersections. The traffic flow parameters mainly include the total number of traffic participants, as well as the type, path, and movement parameters of traffic participants.

[0081] Step 7: Divide the region into multiple uniform units, calculate the accident index Y1 and congestion index Y2 of each unit based on the regional risk assessment model, and calculate the risk factor Y of the unit;

[0082]

[0083] The congestion and smoothness levels are divided in the interval Y=(0,1). The congestion and smoothness level of the unit is determined according to the unit's risk factor Y. The smoothness rate of the area is determined according to the proportion of smooth units. The smoothness rate threshold is set, and based on the regional risk assessment model, the limit values ​​of the ratio of unmanned delivery vehicles and the number of safety personnel in the area are determined.

[0084] When calculating the deployment rules of unmanned delivery vehicles in a region, the region is divided into several units, the deployment rules of each unit are calculated separately, and the traffic conditions of each unit in the region are compared under the same deployment rules. Fig. 9 As shown, green indicates that the road is unobstructed or basically unobstructed after the distribution unit has deployed the unmanned delivery vehicle in accordance with the deployment rules. Yellow indicates light congestion, orange indicates moderate congestion, and red indicates moderate congestion.

[0085] The deployment volume of unmanned delivery vehicles and the ratio of safety officers are determined by reverse deduction. Based on the pre-defined safety officer's ability to handle vehicle disengagement, the ratio of safety officers is determined. Through simulation experiments, the risk factor Y of the unit in the area is calculated. Under the principle of ensuring the basic smooth flow of the area, the limit value of the deployment volume of unmanned delivery vehicles and the ratio of safety officers are given. According to the needs, the actual deployment volume and the ratio of safety officers are formulated within the limit value, such as Fig.11 As shown, taking the threshold value Y≤0.5 as an example, an optional solution for unmanned delivery vehicles in this area is: the ratio of people to vehicles is 1:5, and the delivery quantity is 300.

[0086] In the above technical solution, the unit is set first, then the thermal road is selected, and then the thermal road simulation experiment is carried out to obtain the single simulation result data. A neural network is used to build a regional risk assessment model, and scientific suggestions are made on the deployment volume and the ratio of safety personnel of regional unmanned delivery vehicles.

[0087] The present invention also provides an electronic device, comprising: at least one processor, and a memory connected to the at least one processor in communication, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor executes the method. The electronic device can be any terminal device including a mobile phone, a laptop computer, a desktop computer, a tablet computer, a PDA (Personal Digital Assistant), a POS (Point of Sales), a vehicle-mounted computer, etc.

[0088] The present invention also provides a storage medium on which a computer program is stored, and when the program is executed by a processor, the method described is implemented. Through the description of the above implementation methods, the technicians in the relevant field can clearly understand that the present invention can be implemented by means of software plus necessary general hardware, and of course it can also be implemented by special hardware including special integrated circuits, special CPUs, special memories, special components, etc. In general, all functions completed by computer programs can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be various, such as analog circuits, digital circuits or special circuits. However, for the present invention, software program implementation is a better implementation method in more cases. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product is stored in a readable storage medium, such as a computer floppy disk, a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a disk or an optical disk, etc., including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0089] The present invention also provides a computer program product, which includes a computer program, which, when executed by a processor, implements the method described. The computer program (which may also be referred to or described as a program, software, software application, module, software module, script or code) can be written in any form of programming language, including a compiled language or an interpreted language or a declarative language or a procedural language, and the computer program can be expanded in any form, including as an independent program or as a module, component, subroutine or other unit suitable for use in a computing environment. The computer program may but does not necessarily correspond to a file in a file system. The program may be stored in a part of a file that stores other programs or data, for example, one or more scripts stored in the following: in a markup language document; in a single file dedicated to a related program; or in multiple collaborative files, for example, files storing one or more modules, subroutines or code portions. The computer program can be expanded to execute on one or more computers, the computers being located in one place, or distributed to multiple locations and interconnected through a communication network. The processing and logic flow described in this specification can be performed by one or more programmable computers, which execute one or more computer programs by operating input data and generating output to run functions.

[0090] The number of devices and processing scales described here are used to simplify the description of the present invention. Applications, modifications and variations of the present invention will be obvious to those skilled in the art.

[0091] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and the implementation modes, and they can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and the illustrations shown and described herein.

Claims

1. Traffic risk assessment method for unmanned delivery vehicle deployment test, characterized by: include: Step 1: extract static elements and dynamic elements of road collection data and build a road feature library; Step 2: Define the road network parameters and traffic flow parameters for each time period, deploy unmanned delivery vehicles, set the safety officer ratio, where the safety officer ratio is the deployed human-vehicle ratio, conduct simulation experiments, and control the behavior of the unmanned delivery vehicle in the simulation experiment based on the probability values ​​of the unmanned delivery vehicle's behavior performance statistically calculated based on the test results of the unmanned delivery vehicle in various dangerous scenarios. The behavior performance includes deceleration, emergency stop, and collision, and calculate the traffic congestion index TPI; Extract the simulation result data and calculate the accident index Y1 and congestion index Y2 in each time period: Among them, τ in formula (1) j is the probability of a dangerous scenario occurring in the jth experiment, d is the number of dangerous scenarios, c is udv is the number of unmanned delivery vehicles deployed in the jth experiment, k i is the number of dangerous scenarios that occurred in the i-th unmanned delivery vehicle in the j-th experiment; n in formula (2) is the number of experimental repetitions; Step 3: The time period of a road experiment, road network parameters, traffic flow parameters, number of unmanned delivery vehicles, safety officer ratio, accident index Y1 and congestion index Y2 constitute a piece of data; Construct a data set, divide the data into training set, test set, and validation set, and use a neural network to build a road-level risk assessment model. The input layer of the neural network includes time period, road network parameters, traffic flow parameters, number of unmanned delivery vehicles, and ratio of safety personnel. The output layer includes the accident index Y1 and congestion index Y2 of the road. Step 4: Set the accident index Y1 threshold and congestion index Y2 threshold of the road, and determine the limit values ​​of the number of unmanned delivery vehicles and safety officer ratios in each time period based on the road-level risk assessment model.

2. The method for evaluating traffic risks during the deployment test of an unmanned delivery vehicle according to claim 1, characterized in that: Also includes: Step 5: Set a unit within a certain range, define the road network parameters and traffic flow parameters of the unit, deploy unmanned delivery vehicles, set the safety officer ratio, conduct simulation experiments, and select the roads that are most frequently passed by unmanned delivery vehicles as thermal roads; Taking the thermal road as the object, a simulation experiment is carried out, the simulation result data is extracted, and the accident index Y1 and congestion index Y2 are calculated to represent the accident index Y1 and congestion index Y2 of the unit to which it belongs; Step 6: Repeat step 5 to collect multiple unit data. The road network parameters, traffic flow parameters, number of unmanned delivery vehicles, safety officer ratio, accident index Y1 and congestion index Y2 of one unit constitute one piece of data. Construct a data set, divide the data into training set, test set, and validation set, and use a neural network to build a regional risk assessment model. The input layer of the neural network includes the unit's road network parameters, traffic flow parameters, number of unmanned delivery vehicles, and safety officer ratios. The output layer includes the unit's accident index Y1 and congestion index Y2. Step 7: Divide the region into multiple uniform units, calculate the accident index Y1 and congestion index Y2 of each unit based on the regional risk assessment model, and calculate the risk factor Y of each unit; The congestion and smoothness levels are divided in the interval Y=(0,1). The congestion and smoothness levels of each unit are determined according to the risk factor Y of each unit. The smoothness rate of the area is determined according to the proportion of smooth units. The smoothness rate threshold is set, and based on the regional risk assessment model, the number of unmanned delivery vehicles and the limit value of the safety officer ratio in the area are determined.

3. The method for evaluating traffic risks during the deployment test of an unmanned delivery vehicle according to claim 2, characterized in that: The shape of the unit is a regular hexagon, which divides the area into multiple uniform distribution units using a honeycomb network structure.

4. The method for evaluating traffic risks during the deployment test of an unmanned delivery vehicle according to claim 1, characterized in that: When a fully connected neural network is used to construct a road-level risk analysis model, the number of hidden layers is 2, the number of nodes is 128, the weight decay coefficient is 0.2, the loss function is the minimum mean square error, and the minimum error threshold is 0.

05.

5. The method for evaluating traffic risks during the deployment test of an unmanned delivery vehicle according to claim 2, characterized in that: When a fully connected neural network is used to construct a regional risk analysis model, the number of hidden layers is 2, the number of nodes is 128, the weight decay coefficient is 0.2, the loss function is the minimum mean square error, and the minimum error threshold is 0.

05.

6. The method for evaluating traffic risks during the deployment test of an unmanned delivery vehicle according to claim 1, characterized in that: When using neural networks to construct a road-level risk analysis model, the road network parameters include road length, proportion of road sections with mixed motor and non-motor vehicle traffic, number of intersections, intersection shape, number of traffic lights, road type, road speed limit, number of two-way motor vehicle lanes, presence or absence of isolation, width of motor vehicle lanes, width of mixed motor and non-motor vehicle lanes, and width of non-motor vehicle lanes. Traffic flow parameters include the total number of traffic participants, as well as their types, paths, and movement parameters.

7. The method for evaluating traffic risks during the deployment test of an unmanned delivery vehicle according to claim 2, characterized in that: When using neural networks to construct a regional risk analysis model, the road network parameters include regional area, regional road length, number of intersections, number of traffic lights, and the distribution and concentration of road intersections; the traffic flow parameters include the total number of traffic participants, as well as the type, path, and movement parameters of traffic participants.

8. An electronic device, characterized in that include: At least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor executes the method according to any one of claims 1 to 7.

9. A storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method described in any one of claims 1 to 7 is implemented.

10. A computer program product comprising a computer program, characterized in that When the program is executed by a processor, the method described in any one of claims 1 to 7 is implemented.

Images