CN110008577B - Vehicle automatic lane changing function evaluation method based on worst global risk degree search - Google Patents

Abstract

The invention discloses a vehicle automatic lane changing function evaluation method based on worst global risk search_danAnd for output, reproducing the most dangerous lane changing scene corresponding to the worst global risk degree through the actual field, verifying the to-be-tested automatic driving lane changing function, and if the to-be-tested automatic driving lane changing function can meet the operation requirement under the most dangerous lane changing scene, determining that the to-be-tested automatic driving lane changing function passes the evaluation. The invention can reduce the simulation and field test quantity and improve the safety of the automatic driving automatic lane changing function in the real vehicle running process, thereby playing a certain role in promoting the development of the automatic driving technology and industrialization.

Description

Vehicle automatic lane changing function evaluation method based on worst global risk degree search

Technical Field

The invention relates to the technical field of intelligent transportation, in particular to an evaluation method for an automatic lane changing function of an automatic driving vehicle based on the worst global risk.

Background

One of the main problems faced by the development of autonomous vehicles to date is how to land, which not only needs to further improve the autonomous technology and solve the problem of how to commercialize autonomous driving, but also what is more important is how to ensure the safety of the autonomous vehicle after the vehicle has traveled the way. The automatic lane changing function is one of the most basic functions of an automatic driving vehicle, but compared with a traditional vehicle active safety system, the automatic lane changing function has longer duration and greatly increases the number of related scene variable parameters, and meanwhile, when the vehicle speed is higher, once the automatic lane changing function is operated in error, greater danger is brought. Therefore, the test evaluation of the automatic lane changing function is not only more complex, but also more strict. If the automatic lane changing function of the automatic driving is tested according to the traditional vehicle active safety testing method, in order to achieve higher safety, a site test with huge number is required, and the time and money cost is too huge to be practical. However, theoretical research on test and evaluation of the automatic lane changing function of the automatic driving vehicle is still lacked in the existing research, and especially how to ensure the safety of the automatic lane changing function while reducing the test quantity is lacked.

Disclosure of Invention

Aiming at the prior art, the invention provides a vehicle automatic lane changing function evaluation method based on worst global risk degree search.

The method comprises the steps of firstly determining a function boundary of an automatic driving lane changing function to be detected, and simultaneously determining the configuration of a vehicle sensor with the automatic driving lane changing function to be detected, vehicle parameters having influence on the automatic lane changing function and the like. And on the basis, a simulation platform of the to-be-tested automatic driving vehicle model, the scene model and the to-be-tested automatic driving algorithm black box model is set up. Furthermore, a targeted site experiment can be performed on the vehicle comprising the sensor, so that parameter adjustment of the vehicle simulation platform is realized, and the accuracy of the simulation platform is ensured to be in a certain range.

And determining the boundaries of each parameter in the scene by analyzing the automatic lane changing function, the ODD (Operational Design Domain) corresponding to the automatic lane changing function and the application scene. And selecting initial parameters according to lane change scene analysis, and then converting the problem of searching the most dangerous scene into an optimization problem to solve. The input of the optimization problem is a lane changing scene parameter and a vehicle initial parameter, and the optimization target is a global risk degree S_danGlobal risk S_danThe related calculation method is embedded into the simulation platform, and after the simulation platform for determining the scene parameters and the vehicle initial parameters is operated each time, the corresponding global risk degree S is obtained_dan. According to the optimization algorithm, the worst global risk MS can be reached_danIs searched for from the scene parameters and the vehicle initial parametersTo obtain the worst global risk MS_danAnd (5) corresponding vehicle lane changing scenes. The scene is the worst scene which can be met by the corresponding automatic lane changing function in the ODD range, so that the worst scene can cover other scenes to a certain extent in the passing difficulty, the field test amount is reduced, the performance of the corresponding automatic lane changing function under extreme conditions is verified, and the safety of the automatic driving vehicle on the road is improved.

The technical scheme adopted by the invention is as follows: a vehicle automatic lane changing function evaluation method based on worst global risk degree search comprises the following steps:

s1: building a simulation platform, loading a vehicle model and a scene model, determining key parameters such as vehicle dynamics parameters, appearance parameters and sensor parameters, and environment parameters such as road parameters, weather parameters, traffic indication parameters, the number of interfering vehicles and the speed of the interfering vehicles at each moment;

furthermore, the modeling parameters can be adjusted through field experiments, and the accuracy of the simulation platform is ensured.

S2: the boundaries of the environmental parameters are determined by analyzing the automatic lane change function, the functional boundaries and the lane change scenario.

S3: selecting an initial lane changing scene, further carrying out global search on the most dangerous lane changing scene, obtaining the most dangerous scene by means of solving an optimization problem, wherein the input of the optimization problem is an environmental parameter, a lane changing function and a boundary, and the output is a global risk degree S_danWherein the worst global risk MS is defined_danThe corresponding scene is the most dangerous lane changing scene.

S4: the most dangerous lane changing scene is reproduced through an actual field, the to-be-tested automatic driving lane changing function is verified, if the to-be-tested automatic driving lane changing function can meet the operation requirement under the most dangerous lane changing scene, the to-be-tested automatic driving lane changing function can be determined to meet the requirement probably under other scenes, and the to-be-tested automatic driving lane changing function can be determined to pass the evaluation.

Further, solving global risk S_danThe method comprises the following steps:

1) defining a global risk S_dan-vmWherein the maximum global risk MS_dan-vmThe maximum collision risk degree of the vehicle to be detected and each interference vehicle at each moment is determined, and the maximum collision risk degree is expressed as follows:

MS_dan-vm＝max{sup(S_dan-1(t)),sup(S_dan-2(t)),...,sup(S_dan-i(t))} (1)

wherein S is_dan-i(t) is the danger degree of the vehicle to be measured and the ith interfering vehicle at the time t, and S is obtained from the sup table_dan-i(t) upper bound value, S_dan-iThe expression (t) is as follows:

in the formula (x)_i(t),y_i(t))、V_i(t) is the position coordinate and speed of the ith vehicle in the interfering vehicle at time t, (x)_v(t),y_v(t))、V_v(t) is the position coordinate and speed of the vehicle to be measured at the moment t, W is the width of the vehicle, W_LFor road width, C_VIs a constant coefficient set to ensure (- | x)_v(t)-x_i(t)|+(V_i(t)-V_v(t))+C_V) And (- | x)_v(t)-x_i(t)|-(V_i(t)-V_v(t))+C_V) Not negative, C_VTaking a sufficiently large value;

2) defining a lane departure global risk S_dan-lmWherein the maximum lane departure global risk MS_dan-lmThe maximum deviation risk degree of the vehicle to be detected relative to the lane boundary at each moment is determined, and the maximum deviation risk degree is expressed as follows:

MS_dan-lm＝max{sup(S_dan-llane(t)),sup(S_dan-rlane(t))} (4)

in the formula, S_dan-llane(t) and S_dan-rlane(t) lane departure risk degrees at t moments of the vehicle to be tested and any one of the left and right boundaries of the relevant lane, S_dan-llane(t) and S_dan-rlaneThe expression (t) is as follows:

in the formula, y_lY-coordinate of left/right boundary of lane, C_lIs a constant coefficient set to ensure (- | x)_v(t)-x_i(t)|+V_v(t)+C_l) Not negative, C_lTaking a sufficiently large value;

3) defining a global risk S_danOf which the worst global risk MS_danIs represented as follows:

MS_dan＝max{MS_dan-vm,MS_dan-lm} (6)。

by adopting the technical scheme, the safety of the automatic driving lane changing function in the real vehicle running process can be improved while the simulation and field test quantity is reduced, so that the development of the automatic driving technology and industrialization is promoted to a certain extent.

Drawings

The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:

FIG. 1 illustrates a lane-change scenario for an automatic lane-change function under test of the present invention;

FIG. 2 illustrates the establishment of a lane-change scene coordinate system and vehicle initial position parameters of the present invention;

FIG. 3 illustrates an automated search platform and search flow diagram of the present invention based on worst global risk.

Detailed Description

In the following description, numerous specific details are set forth in order to provide a more thorough understanding of the present invention. It will be apparent, however, to one skilled in the art, that the present invention may be practiced without one or more of these specific details. Embodiments of the present invention will be described below with reference to the drawings.

The method of the present invention will be further described and verified by testing, evaluating and analyzing the dual lane change function of an autonomous vehicle.

The description and lane-changing scenario of the dual lane-changing function is shown in fig. 1. In the figure, VUT denotes the autonomous vehicle to be tested, C₁₁、C₁₂、C₂₁、C₂₂The vehicles are called as interference vehicles because the vehicles play a certain interference role in the lane changing process of the detected vehicles, which represent other vehicles in the original lane and the target lane of the detected vehicles.

The goal of the VUT is to switch from the expressway to C₁₁、C₁₂Low lane position in between. Other factors relevant to the lane-changing function are vehicle profile parameters including vehicle width W, vehicle length L, lane width W_LHere, it is assumed that the contour parameters of 5 vehicles are consistent. Meanwhile, the speeds of 5 vehicles are V respectively_T、V_C11、V_C12、V_C21And V_C22，V_TThe vehicle speed is the vehicle speed to be measured, and the other vehicle speeds are the vehicle speeds corresponding to the vehicle numbers.

In order to further represent the relative positional relationship of the vehicle, a coordinate system fixed on the road as shown in fig. 2 is established. In the figure (x)_i,y_i) As the position coordinates of the i car, (x 0)_i,y0_i) The initial position of the vehicle i at the time t is 0 is denoted as (0,0) in the initial position of VUT at the time t is 0.

According to the lane change duration under the scene, the speed change parameter of each vehicle is defined to be within 5 seconds, and the variable parameter is the acceleration a of the vehicle per second_iAnd t is 0, the initial position of the vehicle. Since the degree of risk of the vehicle is generally proportional to the vehicle speed, the initial speed of the vehicle is determined as the upper limit value of the vehicle speed of the corresponding lane. The dynamic environment parameters of the interfering vehicle i are, as defined above:

[x0_i v0_i a_i|t＝0s a_i|t＝1s a_i|t＝2s a_i|t＝3s a_i|t＝4s a_i|t＝5s]

therefore, in the lane changing scene, the four interfering vehicles contain 28 dynamic environment parameters in total. In order to quickly search the scene with the worst global risk degree, a simulation platform comprising a scene model, a vehicle dynamics model and a black box of an automatic lane change algorithm to be tested is built. And meanwhile, establishing a global risk degree calculation module according to the formulas (1) to (6) to integrate the global risk degree calculation module into a simulation platform. Under each group of environmental parameters, after the simulation platform runs, the global risk corresponding to the group of parameters can be calculated through the global risk calculation module. Taking the 28 environmental parameters as optimization variables, and obtaining the global risk S_danThe maximum (the maximum corresponding to the worst scene) is an optimization target, and an optimization tool is combined to build a worst global risk degree automatic search platform together with the simulation platform, as shown in fig. 3.

The method comprises the steps of obtaining the most dangerous scene aiming at the automatic lane changing function to be tested through the automatic search platform, building the same parameter scene in a field test, carrying out a field test of the automatic lane changing vehicle to be tested under the scene, obtaining the worst or relatively poor scene which is possibly generated in the actual vehicle running of the automatic lane changing function, judging that the running of the automatic lane changing function in other scenes can meet the requirement probably if the worst scene can meet the running requirement of the automatic lane changing function, and considering that the automatic lane changing function passes the evaluation.

The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention.

Claims (1)

1. A vehicle automatic lane changing function evaluation method based on worst global risk degree search comprises the following steps:

s1: building a simulation platform, and building a vehicle model and a scene model;

s2: determining the boundary of the environmental parameter by analyzing the functional boundary of the automatic lane changing function to be tested and the lane changing scene;

s3: selecting an initial lane change scene, determining initial environment parameters, performing global search on the most dangerous lane change scene, and searching by the worst global risk MS_danAs an output;

s4: reproducing the most dangerous lane changing scene through the actual field, verifying the automatic driving lane changing function to be tested, and if the automatic driving lane changing function to be tested can meet the operation requirement under the most dangerous lane changing scene, determining that the automatic driving lane changing function to be tested passes the evaluation;

worst global risk MS_danWith maximum vehicle-to-vehicle global risk MS_dan-vmAnd maximum lane departure global risk MS_dan-lmThe maximum value of the two is taken as a selected value;

maximum vehicle-to-vehicle global risk MS_dan-vmThe solving method comprises the following steps:

defining a global risk S_dan-vmMaximum global vehicle-to-vehicle risk MS_dan-vmThe maximum collision risk degree of the vehicle to be detected and each interference vehicle at each moment is determined, and the maximum collision risk degree is expressed as follows:

MS_dan-vm＝max{sup(S_dan-1(t)),sup(S_dan-2(t)),...,sup(S_dan-i(t))} (1)

in the formula, S_dan-i(t) is the danger degree of the vehicle to be measured and the ith interfering vehicle at the time t, and S is obtained from the sup table_dan-i(t) upper bound value, S_dan-iThe expression (t) is as follows:

in the formula (x)_i(t),y_i(t))、V_i(t) is the position coordinate and speed of the ith vehicle in the interfering vehicle at time t, (x)_v(t),y_v(t))、V_v(t) is the position coordinate and speed of the vehicle to be measured at the moment t, W is the width of the vehicle, W_LFor road width, C_VIs a constant coefficient, ensures (| x)_v(t)-x_i(t)|+(V_i(t)-V_v(t))+C_V) And (- | x)_v(t)-x_i(t)|-(V_i(t)-V_v(t))+C_V) Not a negative value;

maximum global risk of lane departure MS_dan-lmThe solving method comprises the following steps:

defining global risk of lane departure S_dan-lmMaximum global risk of lane departure MS_dan-lmThe maximum deviation risk degree of the vehicle to be detected relative to the lane boundary at each moment is determined, and the maximum deviation risk degree is expressed as follows:

MS_dan-lm＝max{sup(S_dan-llane(t)),sup(S_dan-rlane(t))} (4)

in the formula, S_dan-llane(t) and S_dan-rlane(t) lane departure risk degrees at t moments of the vehicle to be tested and any one of the left and right boundaries of the relevant lane, S_dan-llane(t) and S_dan-rlaneThe expression (t) is as follows:

y_ly coordinate, C, corresponding to left/right boundary of the l-th lane_lIs a constant coefficient, ensures (| x)_v(t)-x_i(t)|+V_v(t)+C_l) It is not a negative value.

Images