Method for evaluating the visibility safety of intelligent driving vehicles in severe weather based on virtual simulation

Through the virtual simulation evaluation model, the attenuation effect and line-of-sight safety risks of lidar in bad weather are calculated, which solves the problem of line-of-sight safety evaluation of intelligent driving vehicles in bad weather, provides safety risk assessment and early warning to ensure that they are in bad weather. safe operation.

CN115081300BActive Publication Date: 2024-11-08SOUTHEAST UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210618342.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-06-01
Publication Date
2024-11-08
Estimated Expiration
2042-06-01

AI Technical Summary

Technical Problem

There are few researches on the safety evaluation of the visual range of intelligent driving vehicles under severe weather conditions. It is difficult for the existing technology to efficiently, safely and accurately evaluate the safety of their visual range of sight, which affects their driving safety on the road.

Method used

Using a virtual simulation evaluation method, by establishing a visual range safety risk assessment model, obtaining the driving working conditions of intelligent driving vehicles, calculating the equivalent atmospheric decay coefficient of lidar in bad weather, and building intelligent driving vehicles under bad weather conditions In the working condition scenario, the supply driving sight line and the demand driving sight line are calculated, and the reliability random model is used to calculate the vision safety risk probability, and compared it with the safety risk threshold to conduct a vision safety assessment.

Benefits of technology

A quantitative assessment of the safety risks of intelligent driving vehicles in severe weather conditions has been achieved, and early warning design optimization is provided to help ensure the safe operation of intelligent driving vehicles under different road lines.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115081300B_ABST
    Figure CN115081300B_ABST
Patent Text Reader

Abstract

The present invention discloses a method for evaluating the sight distance safety of intelligent driving vehicles in bad weather based on virtual simulation, including: obtaining the driving condition information of the intelligent driving vehicle; calculating the equivalent atmospheric attenuation coefficient function according to the transmission attenuation effect; calculating the equivalent atmospheric transmittance function in the virtual simulation laser radar equation according to the equivalent attenuation coefficient function; calculating the required driving sight distance of intelligent driving vehicles of different driving automation levels according to the intelligent driving vehicle information; establishing the reliability function function of the sight distance failure of the intelligent driving vehicle under bad weather conditions according to the supplied driving sight distance and the required driving sight distance, and calculating the sight distance safety risk probability using a reliability random model; comparing the sight distance safety risk probability with the sight distance safety risk threshold, and evaluating the sight distance safety. The present invention is helpful to quantitatively evaluate the sight distance safety risk of intelligent driving vehicles in bad weather with different road lines, and provides a theoretical reference for the design optimization field of intelligent driving vehicle system warning.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of road safety evaluation, and in particular to a method for evaluating the visibility safety of intelligent driving vehicles in severe weather based on virtual simulation. Background Art

[0002] Intelligent driving vehicle technology is an emerging technology launched by the automotive industry to meet passenger comfort, alleviate traffic congestion and fuel emissions. This technology is characterized by phased development. During its development, it is inevitable that there will be certain conflicts with the current design concept for manned driving on existing roads. Whether existing roads can ensure the safe landing of intelligent driving vehicles and transition to popularization has become a fascinating research topic in the design of smart road traffic systems.

[0003] Intelligent driving vehicles perceive the surrounding environment through sensor fusion technology. LiDAR has gradually become the research focus of intelligent driving vehicle perception technology due to its high resolution and strong ranging capability. However, LiDAR is affected by severe weather environments such as rainy, snowy or foggy days, and its working performance is greatly affected in severe weather environments, which in turn affects the driving safety of intelligent driving vehicles on existing roads.

[0004] Sight safety is an important control factor in road design concepts, and the performance of intelligent driving vehicle perception systems in severe weather environments is extremely important. However, the current research results on the performance of lidar in severe weather are mainly the farthest straight-line distance of laser obstacles in severe weather. In road design, sight distance is the length of the vehicle's driving path, which can be calculated based on the road geometry design parameters. There are few studies on the sight safety of intelligent driving vehicles on different road lines in severe weather. Therefore, how to efficiently, safely and accurately evaluate the sight safety of intelligent driving vehicles in severe weather has become an important issue that must be solved to ensure the smooth development of intelligent driving vehicles. Summary of the invention

[0005] Purpose of the invention: In view of the above problems, the purpose of the present invention is to provide a method for evaluating the visibility safety of intelligent driving vehicles in severe weather based on virtual simulation. By establishing an assessment model for visibility safety risks, roads where there are risks to the visibility safety of intelligent driving vehicles in severe weather can be identified and warned.

[0006] Technical solution: A method for evaluating the sight distance safety of an intelligent driving vehicle in bad weather based on virtual simulation of the present invention comprises the following steps:

[0007] Step 1, obtaining driving condition information of the intelligent driving vehicle, including intelligent driving vehicle information, road design information carrying the intelligent driving vehicle, environmental information of the area where the road is located, and obstacle information;

[0008] Step 2, calculating the transmission attenuation effect of the laser radar according to the environmental information of the road area, and calculating the equivalent atmospheric attenuation coefficient function of the laser radar under severe weather conditions according to the transmission attenuation effect;

[0009] Step 3, calculating the equivalent atmospheric transmittance function in the virtual simulation lidar equation according to the equivalent attenuation coefficient function, and obtaining the supply driving sight distance of the intelligent driving vehicle in bad weather;

[0010] Step 4, calculating the required driving sight distances of intelligent driving vehicles of different driving automation levels according to the intelligent driving vehicle information;

[0011] Step 5: Establish a reliability function of the sight distance failure of the intelligent driving vehicle under severe weather conditions according to the supply sight distance and the demand sight distance, and calculate the sight distance safety risk probability using a reliability random model;

[0012] Step 6: Compare the line-of-sight safety risk probability with the line-of-sight safety risk threshold to evaluate line-of-sight safety.

[0013] Furthermore, the transmission attenuation effect in step 2 includes the absorption and scattering effect of atmospheric molecules on the laser radar, the absorption and scattering effect of atmospheric aerosols on the laser radar, and the attenuation effect of the virtual noise point cloud on the laser radar;

[0014] The equivalent atmospheric attenuation coefficient function includes the functional relationship between the cumulative absorption and scattering effects of atmospheric molecules and atmospheric aerosols during the laser radar transmission process, and the attenuation effect of false noise point clouds in the laser radar identifying the surrounding scene information.

[0015] Furthermore, the calculation process of the absorption and scattering effect of the atmospheric molecules on the laser radar includes:

[0016] The standard atmospheric model is used to calculate the absorption coefficient of atmospheric molecules to laser, and the absorption effect of atmospheric molecules on laser is calculated based on the absorption coefficient. The scattering coefficient of atmospheric molecules to laser is calculated using the scattering model, and the scattering effect of atmospheric molecules on laser is calculated based on the scattering coefficient. The absorption effect and scattering effect are added together to obtain the transmission attenuation of the laser radar affected by atmospheric molecules in bad weather.

[0017] Furthermore, the calculation process of the absorption and scattering effect of the atmospheric aerosol on the lidar includes:

[0018] Step 21, based on the raindrop characteristic information of the environmental information of the road area, using the raindrop distribution function, the raindrop size distribution model and the scattering model, calculate the absorption and scattering effects of the atmospheric aerosol molecules on the laser radar under rainy weather conditions, and calculate the extinction coefficient of the laser radar under rainy weather conditions through the absorption and scattering effects;

[0019] Step 22, according to the snowflake characteristic information of the environmental information of the road area, using the snowflake distribution model and the scattering model, calculate the absorption and scattering effects of the atmospheric aerosol molecules on the laser radar under snowy weather conditions, and calculate the extinction coefficient of the laser radar under snowy weather conditions through the absorption and scattering effects;

[0020] Step 23, based on the fog particle characteristic information of the road area environmental information, use the empirical model to calculate the absorption and scattering effects of the fog particles on the laser radar.

[0021] Furthermore, the calculation process of the attenuation effect of the virtual noise point cloud on the laser radar includes:

[0022] According to the characteristic information of raindrops, snowflakes and fog particles, the particle sizes of atmospheric molecules and atmospheric aerosols in Monte Carlo simulation of severe weather conditions, the particle generation locations and the theoretical calculation energy of particle reflected laser are calculated. According to the technical parameters of lidar, the probability of false noise point clouds generated by atmospheric molecules and atmospheric aerosols in Monte Carlo simulation of severe weather conditions is calculated, and the attenuation effect of false noise point clouds generated by atmospheric molecules and atmospheric aerosols reflecting lasers in severe weather conditions is obtained.

[0023] Furthermore, the equivalent atmospheric transmittance function T(R) is expressed as:

[0024]

[0025] Where α(r,λ) is the equivalent atmospheric attenuation coefficient function, r is the unit length for calculating the attenuation effect during the laser transmission process of the lidar, λ is the wavelength of the lidar, and R is the distance caused by the absorption and scattering of lasers by atmospheric molecules under severe weather conditions or the absorption and scattering of lidars by atmospheric aerosols under severe weather conditions;

[0026] According to the driving condition information of the intelligent driving vehicle and the equivalent atmospheric transmittance function, a driving condition scenario of the intelligent driving vehicle under severe weather conditions is constructed in the autonomous driving simulation software, including a road scenario, an intelligent driving vehicle model based on a lidar, a weather model and an obstacle model. Obstacles are placed continuously at a certain interval on the driving path of the intelligent driving vehicle, and the final position of the obstacle farthest from the position of the intelligent driving vehicle is recorded. The path length between the intelligent driving vehicle and the obstacle is set as the driving sight distance of the intelligent driving vehicle in severe weather, and the driving sight distance is used as the supply driving sight distance ASD of the intelligent driving vehicle in severe weather.

[0027] The placement position of the farthest obstacle is determined based on the point cloud quantity threshold that meets the on-board laser radar laser receiver acceptance power threshold and the visual recognition algorithm requirements of the intelligent driving vehicle.

[0028] Furthermore, according to the driving automation level information, the required parking sight distance expression for vehicles of each level is:

[0029] RSD=S1+S2+S3

[0030] Where S1 is the parking sight distance in the perception and reaction stage of the autonomous driving system or the parking sight distance in the perception and reaction stage of the driver; S2 is the parking sight distance in the braking stage preset by the conditional autonomous driving system to obtain the minimum driving risk; S3 is the parking sight distance in the driver takeover stage or the parking sight distance in the braking stage of the autonomous driving system;

[0031] The driving sight distance of intelligent driving vehicles with different driving automation levels is taken as the required driving sight distance RSD of the intelligent driving vehicle in severe weather.

[0032] Further, the step 5 comprises:

[0033] Step 51, the reliability function expression is:

[0034] Z SD =ASD-RSD

[0035] According to the reliability function, an evaluation model for the safety risk of the intelligent driving vehicle in bad weather visibility is established, and the expression is:

[0036] P r =P(Z SD <0)

[0037] Where P r represents the safety risk probability of the intelligent driving vehicle in bad weather visibility, Z SD <0 means that the intelligent driving vehicle has a visual range failure in bad weather, P(Z SD <0) indicates the probability of sight failure of intelligent driving vehicles in bad weather;

[0038] Step 52, according to the road driving sight distance safety risk assessment target requirements, determine the random variables and deterministic variables of the reliability function, and calculate the random variable distribution, mean value and standard deviation of the random variables and the deterministic variables;

[0039] Step 53, according to the random variable distribution, mean and standard deviation, the reliability index is calculated using the reliability random simulation model, and the evaluation model is calculated to obtain the safety risk probability P of the intelligent driving vehicle in bad weather visibility. r .

[0040] Further, the step 6 comprises:

[0041] The safety risk probability threshold of the visibility distance of the intelligent driving vehicle in bad weather is set in advance as P0, and the safety risk probability of the visibility distance of the intelligent driving vehicle in bad weather P is set in advance as P0. rCompared with the safety risk probability threshold P0 of the intelligent driving vehicle in bad weather visibility, if P r <P0, it means that the intelligent driving vehicle has safe sight distance in bad weather; if P r =P0, it means that the smart driving vehicle is in a critical driving risk state in bad weather. r >P0, it means that there is a risk of visibility in bad weather for the smart driving vehicle.

[0042] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages:

[0043] 1. The present invention starts from the perspective of the impact of road conditions on the safety of sight distance of intelligent driving vehicles in bad weather, builds a scene that can map the driving conditions of intelligent driving vehicles in real bad weather through automatic driving virtual simulation software, and establishes an equivalent lidar model under bad weather conditions based on the reflection and scattering effects of atmospheric molecules and atmospheric aerosols on lasers, as well as the generation of false noise point clouds;

[0044] 2. Consider the sight failure mode of intelligent driving vehicles in bad weather conditions in combination with reliability theory, and establish a driving risk assessment model for intelligent driving vehicles with sight failure under bad weather conditions. This will help to quantitatively assess the sight safety risk of intelligent driving vehicles in bad weather on different road lines, and provide theoretical reference for the design optimization of intelligent driving vehicle system warning.

[0045] 3. The method disclosed in the present invention is used to obtain the driving risk probability of the intelligent driving vehicle, which can be used to check for safety hazards in the actual operation of the intelligent driving vehicle on different road lines under severe weather conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a flow chart of a method for evaluating the visibility safety of an intelligent driving vehicle in severe weather based on virtual simulation according to the present invention. DETAILED DESCRIPTION

[0047] In order to make the objectives, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0048] like Figure 1 The method for evaluating the visibility safety of an intelligent driving vehicle in bad weather based on virtual simulation described in this embodiment includes the following steps:

[0049] Step 1: Obtain driving condition information of the intelligent driving vehicle, including intelligent driving vehicle information, road design information carrying the intelligent driving vehicle, environmental information of the area where the road is located, and obstacle information.

[0050] The information of smart driving vehicles includes driving automation level information, driving speed and on-board laser radar technical parameter information, among which the on-board laser radar technical parameter information includes resolution, field of view angle, installation elevation angle, installation position, perception distance, laser radar measurement distance technical information, emission laser wavelength information, signal-to-noise ratio and laser receiver acceptance power threshold. The above information can be obtained by referring to the relevant parameter data of existing mass-produced smart driving vehicle products and the laser radar manufacturer's product manual.

[0051] The design information of roads carrying intelligent driving vehicles includes geometric parameter information of horizontal, vertical and horizontal sections, and horizontal, vertical and horizontal combination information. The above information can be obtained by referring to the design data of the road section to be evaluated.

[0052] The environmental information of the road includes weather information, roadside structure height information and roadside structure boundary point coordinates, where the weather information should include temperature, humidity, precipitation information, visibility information, raindrop characteristic information, snowflake characteristic information and fog particle characteristic information. The above information can be obtained by consulting the meteorological and geographical data of the area where the road to be evaluated is located.

[0053] Obstacle information includes obstacle material information, reflectivity information, color information, surface texture information, shape information and surface humidity information. The above information can be obtained by referring to the actual scene data of the road section to be evaluated.

[0054] The road environment information is used to classify the driving conditions of the intelligent driving vehicle according to the weather information. Rainy days are classified into trace rainfall (scattered light rain), light rain, moderate rain, heavy rain, rainstorm, heavy rainstorm, and extremely heavy rainstorm based on the precipitation level; snowy days are classified into trace snowfall (scattered light snow), light snow, moderate snow, heavy snow, snowstorm, heavy snowstorm, and extremely heavy snowstorm; foggy days are classified into light fog, light fog, moderate fog, heavy fog, dense fog, and heavy fog according to visibility. The severe weather referred to in this embodiment includes all the above-mentioned levels of rainy days, snowy days, and foggy days.

[0055] Step 2: Calculate the transmission attenuation effect of the laser radar based on the environmental information of the road area, and calculate the equivalent atmospheric attenuation coefficient function of the laser radar under severe weather conditions based on the transmission attenuation effect.

[0056] Transmission attenuation effects include the absorption and scattering effects of atmospheric molecules on lidar, the absorption and scattering effects of atmospheric aerosols on lidar, and the attenuation effect of virtual noise point clouds on lidar.

[0057] The calculation process of the absorption and scattering effect of atmospheric molecules on lidar includes:

[0058] According to the wavelength of the laser radar in the intelligent driving vehicle and the atmospheric absorption spectrum, the standard atmospheric model is used to calculate the absorption coefficient of the atmospheric molecules to the laser, and the absorption effect of the atmospheric molecules on the laser is calculated based on the absorption coefficient; the scattering model is used to calculate the scattering coefficient of the atmospheric molecules to the laser, and the scattering effect of the atmospheric molecules on the laser is calculated based on the scattering coefficient; the absorption effect and the scattering effect are added together to obtain the transmission attenuation of the laser radar affected by the atmospheric molecules in bad weather.

[0059] The calculation process of the absorption and scattering effect of atmospheric aerosol on lidar includes:

[0060] Step 21, based on the raindrop characteristic information of the environmental information of the road area, the tail speed of the raindrop when falling, the axial ratio of the raindrop ellipsoid, the reflectivity and differential reflectivity of the raindrop, the raindrop shape parameters, the slope of the raindrop distribution, the laser absorption cross section of the raindrop, the scattering angle of the raindrop to the laser, the scattering cross section of the raindrop to the laser and the drag coefficient of the raindrop when falling should be considered, and the absorption and scattering effects of the atmospheric aerosol molecules on the laser radar under rainy weather conditions are calculated by using the raindrop distribution function, the raindrop size distribution model and the scattering model, and the extinction coefficient of the laser radar under rainy weather conditions is calculated through the absorption and scattering effects;

[0061] Step 22, based on the snowflake characteristic information of the road area environmental information, the mass density of the snowflake, the water content of the snowflake, the snowflake particle size spectrum, the snowflake falling speed, the air density, the resistance coefficient of the snowflake falling, the absorption cross section of the snowflake to the laser, the scattering angle of the snowflake to the laser, the scattering cross section of the snowflake to the laser, the snowflake shape parameters and the slope of the snowflake distribution should be considered, and the absorption and scattering effects of the atmospheric aerosol molecules on the laser radar under snowy weather conditions are calculated using the snowflake distribution model and the scattering model, and the extinction coefficient of the laser radar under snowy weather conditions is calculated through the absorption and scattering effects;

[0062] Step 23, based on the fog particle characteristic information of the road area environmental information, the radius of the fog particles and the fog formation mechanism should be considered, and the absorption and scattering effects of the fog particles on the lidar should be calculated using an empirical model.

[0063] The calculation process of the attenuation effect of the virtual noise point cloud on the lidar includes:

[0064] According to the characteristic information of raindrops, snowflakes and fog particles, the size of atmospheric molecules and atmospheric aerosols in Monte Carlo simulation of severe weather conditions, the location of particle generation and the theoretical calculation energy of particle reflection laser are calculated. According to the technical parameters of LiDAR, the probability of false noise point clouds generated by atmospheric molecules and atmospheric aerosols in Monte Carlo simulation of severe weather conditions is calculated, and the attenuation effect of false noise point clouds generated by atmospheric molecules and atmospheric aerosols reflecting lasers in severe weather conditions is obtained. The specific process is as follows:

[0065] Step 24, based on the raindrop characteristic information, the snowflake characteristic information and the fog particle characteristic information, determine the atmospheric molecule and atmospheric aerosol particle size, particle generation location, and particle reflected laser theoretical calculation energy of Monte Carlo simulation of severe weather conditions;

[0066] Step 25, based on the laser radar technical parameters, calculate the probability of generating false noise point clouds by the atmospheric molecules and atmospheric aerosols of the Monte Carlo simulation of severe weather conditions, wherein the probability of generating false noise point clouds should ensure that the atmospheric molecules and atmospheric aerosols of the Monte Carlo simulation of severe weather conditions should be within the laser radar recognition range, and the theoretical calculation energy of the particle reflected laser should ensure that it exceeds the signal-to-noise ratio in the laser radar technical parameters;

[0067] Step 26, based on the Monte Carlo simulation, the probability of atmospheric molecules and atmospheric aerosols generating false noise point clouds under severe weather conditions is equivalently replaced by the attenuation effect of atmospheric molecules reflecting lasers and atmospheric aerosols reflecting lasers under severe weather conditions of the lidar to generate false noise point clouds.

[0068] Based on the classification results of the driving conditions of the intelligent driving vehicle in bad weather, the transmission attenuation of the laser radar affected by the atmospheric molecules in bad weather is obtained by considering the absorption and scattering effects of the atmospheric molecules in bad weather on the laser, the extinction coefficient function or attenuation coefficient function is obtained by considering the absorption and scattering effects of the atmospheric aerosols in bad weather on the laser radar, and the attenuation effect of the atmospheric molecules in bad weather reflecting the laser and the atmospheric aerosol reflecting the laser to produce false noise point clouds is obtained by considering the effect of the atmospheric molecules in bad weather reflecting the laser and the atmospheric aerosol reflecting the laser to produce false noise point clouds. The equivalent atmospheric attenuation coefficient function of the laser radar in bad weather is calculated. The equivalent atmospheric attenuation coefficient function includes the functional relationship between the cumulative absorption and scattering effects of atmospheric molecules and atmospheric aerosols in the laser radar transmission process, and the attenuation effect of the false noise point cloud in the laser radar identifying the surrounding scene information.

[0069] Step 3, calculating the equivalent atmospheric transmittance function in the virtual simulation lidar equation according to the equivalent attenuation coefficient function, and obtaining the supply driving sight distance of the intelligent driving vehicle in bad weather;

[0070] The expression of the equivalent atmospheric transmittance function T(R) is:

[0071]

[0072] Where α(r,λ) is the equivalent atmospheric attenuation coefficient function, r is the unit length for calculating the attenuation effect during the laser transmission process of the lidar, λ is the wavelength of the lidar, and R is the distance caused by the absorption and scattering of lasers by atmospheric molecules under severe weather conditions or the absorption and scattering of lidars by atmospheric aerosols under severe weather conditions;

[0073] According to the driving condition information of the intelligent driving vehicle and the equivalent atmospheric transmittance function, the driving condition scenario of the intelligent driving vehicle under severe weather conditions is built in the autonomous driving virtual simulation software, including the road scenario, the intelligent driving vehicle model based on LiDAR, the weather model and the obstacle model. Obstacles are placed continuously at a certain interval on the driving path of the intelligent driving vehicle, and the final position of the obstacle farthest from the position of the intelligent driving vehicle is recorded. The path length between the intelligent driving vehicle and the obstacle is set as the driving sight distance of the intelligent driving vehicle in severe weather, and the driving sight distance is used as the supply driving sight distance ASD of the intelligent driving vehicle in severe weather. The autonomous driving virtual simulation software is run and the LiDAR point cloud information is recorded, which is compared with the LiDAR point cloud information of the intelligent driving vehicle in severe weather in the real world to verify the effectiveness of the LiDAR equivalent atmospheric transmittance function and the driving condition scenario of the intelligent driving vehicle under severe weather conditions.

[0074] The placement of the farthest obstacle is determined based on the point cloud quantity threshold that the intelligent driving vehicle needs to meet the on-board lidar laser receiver acceptance power threshold and the visual recognition algorithm requirements.

[0075] Step 4: Calculate the required driving sight distances of intelligent driving vehicles with different driving automation levels based on the intelligent driving vehicle information.

[0076] According to the driving automation level information, the required parking sight distance expression for vehicles of each level is:

[0077] RSD=S1+S2+S3

[0078] Wherein, S1 is the parking sight distance in the perception and reaction stage of the automatic driving system or the parking sight distance in the perception and reaction stage of the driver; S2 is the parking sight distance in the braking stage preset by the conditional automatic driving system to obtain the minimum driving risk; S3 is the parking sight distance in the driver takeover stage or the parking sight distance in the braking stage of the automatic driving system.

[0079] The driving sight distance of intelligent driving vehicles with different driving automation levels is taken as the required driving sight distance RSD of the intelligent driving vehicle in severe weather.

[0080] Step 5: Establish a reliability function of the sight distance failure of the intelligent driving vehicle under severe weather conditions based on the supply sight distance and the demand sight distance, and calculate the sight distance safety risk probability using the reliability random model:

[0081] Step 51, the reliability function expression is:

[0082] Z SD =ASD-RSD

[0083] According to the reliability function, an evaluation model for the safety risk of sight distance in bad weather for intelligent driving vehicles is established, and the expression is:

[0084] P r =P(Z SD <0)

[0085] Where P r represents the safety risk probability of the intelligent driving vehicle in bad weather visibility, Z SD <0 means that the intelligent driving vehicle has a visual range failure in bad weather, P(Z SD <0) indicates the probability of sight failure of intelligent driving vehicles in bad weather;

[0086] Step 52, according to the road driving sight distance safety risk assessment target requirements, determine the random variables and deterministic variables of the reliability function, and calculate the random variable distribution, mean value and standard deviation of the random variables and the deterministic variables;

[0087] Step 53, according to the random variable distribution, mean and standard deviation, the reliability index is calculated using the reliability random simulation model, and the evaluation model is calculated to obtain the safety risk probability P of the intelligent driving vehicle in bad weather visibility. r .

[0088] Step 6: Compare the line-of-sight safety risk probability with the line-of-sight safety risk threshold to evaluate line-of-sight safety.

[0089] The safety risk probability threshold of the visibility distance of the intelligent driving vehicle in bad weather is set in advance as P0, and the safety risk probability of the visibility distance of the intelligent driving vehicle in bad weather P is set in advance as P0. r Compared with the safety risk probability threshold P0 of the intelligent driving vehicle in bad weather visibility, if P r <P0, it means that the intelligent driving vehicle has safe sight distance in bad weather; if P r =P0, it means that the smart driving vehicle is in a critical driving risk state in bad weather. r >P0, it means that there is a risk of visibility in bad weather for the smart driving vehicle.

[0090] In order to further verify the effect of using the method of the present invention to evaluate the visibility safety of intelligent driving vehicles in severe weather environments, the present application is further explained using the following examples.

[0091] In this example, a straight mountain road with a design speed of 40 km / h is selected as the road to be evaluated, the bad weather type is rainy, the intelligent driving vehicle is a conditional automatic driving vehicle, and the obstacle is a traditional manned car. According to the classification standard, the road condition of the road to be evaluated is light rain, and the rainfall is 5 mm / h.

[0092] In this embodiment, a 905nm laser radar is used for the intelligent driving vehicle to be evaluated. The absorption coefficient of atmospheric molecules to laser is very small, and the attenuation effect of atmospheric molecules on laser can be ignored. According to the Rayleigh scattering model, the larger the wavelength of the laser, the smaller the attenuation coefficient. Therefore, the calculation of the attenuation of this laser radar in the atmospheric molecules is ignored. According to the rainy weather characteristics of the selected area, the Marshall-Palmer raindrop distribution model is selected. The specific model is as follows:

[0093] N(D)=N0e -ΛD (m -3 mm -1 )

[0094] Where D is the diameter of the raindrop, in mm; N0 = 0.08 cm -4 =8000m -3 mm -1 , Λ=41F rain -0.21 cm -1 =4.10F rain -0.21 mm -1 41, F rain Indicates rainfall rate in mm / h;

[0095] The Marshall-Palmer raindrop distribution model is used to calculate the absorption and scattering effects of atmospheric aerosol molecules on lidar under rainy weather conditions, and then the extinction coefficient of lidar under rainy conditions is calculated. The extinction coefficient of the evaluated area is obtained to be about 3.8 dB / km.

[0096] When a smart driving vehicle is driving, the conditions for identifying raindrop noise points should satisfy the requirement that the raindrop particles are within the recognition range of the on-board laser radar, and the raindrop particles that need to be scanned can reflect enough energy. The Monte Carlo virtual simulation method can be used to randomly simulate the size of raindrop particles, the position distribution of raindrop particles, and the probability of calculating noise points. By setting an appropriate number of iterations, the probability P of the scanned raindrop particles can be obtained. return In this example, the number of iterations is set to N = 1.0 × 10 7 , for rainy weather return Taking the mean, we can get P returnIt is about 0.273, and the probability of raindrop noise points that the raindrop particles are within the recognition range of the on-board laser radar and the raindrop particles that need to be scanned can reflect back enough energy is 0.001. After the environmental conditions are determined, the equivalent atmospheric attenuation coefficient function of the road to be evaluated in light rain weather is calculated to be 1.1126 / km. When building a driving scene of an intelligent driving vehicle in severe weather conditions in the autonomous driving simulation software, this example uses Prescan for construction because it has a more realistic physics-based sensor model and can provide feedback on changes in obstacle recognition rate in different weather environments. The simulation results show that the final position of the intelligent driving vehicle's driving path where the intelligent driving vehicle meets the threshold of the laser receiver receiving power of the on-board laser radar and the threshold of the number of point clouds required by the visual recognition algorithm, that is, the path length between the intelligent driving vehicle and the obstacle is 71.345m.

[0097] In this embodiment, the control of the L3 autonomous driving vehicle is transferred from the system to the driver. In this example, the required parking sight distance for the L3 level is:

[0098]

[0099] In the formula, RSD L3 is the required stopping sight distance (m) for L3 autonomous driving that varies with the position point on the driving path, t p_L3 is the perception-braking reaction time of the autonomous driving system (s), t T The response time for the L3 driver to take over (s); A dp Preset braking deceleration for L3 autonomous driving system (m / s 2 ).

[0100] According to the road driving sight distance safety risk assessment target requirements, the random variables and deterministic variables of the reliability function of the intelligent driving vehicle's sight distance failure in bad weather are determined. In this process, the L3 level automatic driving system preset braking deceleration, perception reaction time, automatic driving system perception-braking reaction time, automatic driving system braking deceleration, etc. in the driving automation system of the automatic driving vehicle are deterministic variables, which have been determined before the car leaves the factory, and only the determined value needs to be determined. The driving speed, driver's braking deceleration, L3 level driver takeover reaction time, etc. are random variables. These variables are not exactly the same for different drivers based on their driving experience and driving habits. It is necessary to determine the mean and standard deviation of the above random variables, as well as the distribution form, and use them in the reliability analysis method.

[0101] The reliability index is solved by using the reliability random simulation method. In this example, the first-order second-moment reliability random simulation method is used to solve the established evaluation model of the intelligent driving vehicle's bad weather visibility safety risk, and the intelligent driving vehicle's bad weather visibility safety risk probability P of the intelligent driving vehicle's driving condition is obtained. r About 0.068.

[0102] Set the intelligent driving vehicle sight distance safety risk threshold P0 = 0.5; the intelligent driving vehicle bad weather sight distance safety risk probability P r Compared with the safety risk probability threshold P0 of the intelligent driving vehicle in bad weather, P r <P0, indicating that the smart driving vehicle has safe visibility in bad weather.

Claims

1. A method for evaluating the visibility safety of intelligent driving vehicles in bad weather based on virtual simulation, characterized in that: The following steps are involved: Step 1, obtaining driving condition information of the intelligent driving vehicle, including intelligent driving vehicle information, road design information carrying the intelligent driving vehicle, environmental information of the area where the road is located, and obstacle information; Step 2, calculating the transmission attenuation effect of the laser radar according to the environmental information of the road area, and calculating the equivalent atmospheric attenuation coefficient function of the laser radar under severe weather conditions according to the transmission attenuation effect; Step 3, calculating the equivalent atmospheric transmittance function in the virtual simulation lidar equation according to the equivalent attenuation coefficient function, and obtaining the supply driving sight distance of the intelligent driving vehicle in bad weather; Step 4, calculating the required driving sight distances of intelligent driving vehicles of different driving automation levels according to the intelligent driving vehicle information; Step 5: Establish a reliability function of the sight distance failure of the intelligent driving vehicle under severe weather conditions according to the supply sight distance and the demand sight distance, and calculate the sight distance safety risk probability using a reliability random model; Step 6, comparing the line-of-sight safety risk probability with the line-of-sight safety risk threshold to evaluate line-of-sight safety; The transmission attenuation effect in step 2 includes the absorption and scattering effect of atmospheric molecules on the laser radar, the absorption and scattering effect of atmospheric aerosols on the laser radar, and the attenuation effect of virtual noise point cloud on the laser radar; The equivalent atmospheric attenuation coefficient function includes the functional relationship between the cumulative absorption and scattering effects of atmospheric molecules and atmospheric aerosols during the laser radar transmission process, and the attenuation effect of the false noise point cloud in the laser radar identifying the surrounding scene information; The expression of the equivalent atmospheric transmittance function T(R) is: Where α(r,λ) is the equivalent atmospheric attenuation coefficient function, r is the unit length for calculating the attenuation effect of the laser radar during laser transmission, λ is the laser radar wavelength, and R is the distance caused by the absorption and scattering of laser by atmospheric molecules under severe weather conditions or the absorption and scattering of laser by atmospheric aerosol under severe weather conditions; According to the driving condition information of the intelligent driving vehicle and the equivalent atmospheric transmittance function, a driving condition scenario of the intelligent driving vehicle under severe weather conditions is constructed in the autonomous driving simulation software, including a road scenario, an intelligent driving vehicle model based on a lidar, a weather model and an obstacle model. Obstacles are placed continuously at a certain interval on the driving path of the intelligent driving vehicle, and the final position of the obstacle farthest from the position of the intelligent driving vehicle is recorded. The path length between the intelligent driving vehicle and the obstacle is set as the driving sight distance of the intelligent driving vehicle in severe weather, and the driving sight distance is used as the supply driving sight distance ASD of the intelligent driving vehicle in severe weather. The placement position of the farthest obstacle is determined based on the point cloud quantity threshold that meets the on-board laser radar laser receiver acceptance power threshold and the visual recognition algorithm requirements of the intelligent driving vehicle.

2. The method according to claim 1, characterized in that The calculation process of the absorption and scattering effect of the atmospheric molecules on the laser radar includes: The standard atmospheric model is used to calculate the absorption coefficient of atmospheric molecules to laser, and the absorption effect of atmospheric molecules on laser is calculated based on the absorption coefficient. The scattering coefficient of atmospheric molecules to laser is calculated using the scattering model, and the scattering effect of atmospheric molecules on laser is calculated based on the scattering coefficient. The absorption effect and scattering effect are added together to obtain the transmission attenuation of the laser radar affected by atmospheric molecules in bad weather.

3. The method according to claim 2, characterized in that The calculation process of the absorption and scattering effect of atmospheric aerosol on lidar includes: Step 21, based on the raindrop characteristic information of the environmental information of the road area, using the raindrop distribution function, the raindrop size distribution model and the scattering model, calculate the absorption and scattering effects of the atmospheric aerosol molecules on the laser radar under rainy weather conditions, and calculate the extinction coefficient of the laser radar under rainy weather conditions through the absorption and scattering effects; Step 22, according to the snowflake characteristic information of the environmental information of the road area, using the snowflake distribution model and the scattering model, calculate the absorption and scattering effects of the atmospheric aerosol molecules on the laser radar under snowy weather conditions, and calculate the extinction coefficient of the laser radar under snowy weather conditions through the absorption and scattering effects; Step 23, based on the fog particle characteristic information of the road area environmental information, use the empirical model to calculate the absorption and scattering effects of the fog particles on the laser radar.

4. The method according to claim 3, characterized in that The calculation process of the attenuation effect of the virtual noise point cloud on the laser radar includes: According to the characteristic information of raindrops, snowflakes and fog particles, the particle sizes of atmospheric molecules and atmospheric aerosols in Monte Carlo simulation of severe weather conditions, the particle generation locations and the theoretical calculation energy of particle reflected laser are calculated. According to the technical parameters of lidar, the probability of false noise point clouds generated by atmospheric molecules and atmospheric aerosols in Monte Carlo simulation of severe weather conditions is calculated, and the attenuation effect of false noise point clouds generated by atmospheric molecules and atmospheric aerosols reflecting lasers in severe weather conditions is obtained.

5. The method according to claim 4, characterized in that According to the driving automation level information, the required parking sight distance expression for vehicles of each level is: RSD=S1+S2+S3 Where S1 is the parking sight distance in the perception and reaction stage of the autonomous driving system or the parking sight distance in the perception and reaction stage of the driver; S2 is the parking sight distance in the braking stage preset by the conditional autonomous driving system to obtain the minimum driving risk; S3 is the parking sight distance in the driver takeover stage or the parking sight distance in the braking stage of the autonomous driving system; The driving sight distance of intelligent driving vehicles with different driving automation levels is taken as the required driving sight distance RSD of the intelligent driving vehicle in severe weather.

6. The method according to claim 5, characterized in that The step 5 comprises: Step 51, the reliability function expression is: Z SD =ASD-RSD According to the reliability function, an evaluation model for the safety risk of the intelligent driving vehicle in bad weather visibility is established, and the expression is: P r =P(Z SD <0) Where P r represents the safety risk probability of the intelligent driving vehicle in bad weather visibility, Z SD <0 means that the intelligent driving vehicle has a visual range failure in bad weather, P(Z SD <0) indicates the probability of sight failure of intelligent driving vehicles in bad weather; Step 52, according to the road driving sight distance safety risk assessment target requirements, determine the random variables and deterministic variables of the reliability function, and calculate the random variable distribution, mean value and standard deviation of the random variables and the deterministic variables; Step 53, according to the random variable distribution, mean and standard deviation, the reliability index is calculated using the reliability random simulation model, and the evaluation model is calculated to obtain the safety risk probability P of the intelligent driving vehicle in bad weather visibility. r .

7. The method according to claim 6, characterized in that The step 6 comprises: The safety risk probability threshold of the visibility distance of the intelligent driving vehicle in bad weather is set in advance as P0, and the safety risk probability of the visibility distance of the intelligent driving vehicle in bad weather P is set in advance as P0. r Compared with the safety risk probability threshold P0 of the intelligent driving vehicle in bad weather visibility, if P r <P0, it means that the intelligent driving vehicle has safe sight distance in bad weather; if P r =P0, it means that the smart driving vehicle is in a critical driving risk state in bad weather. r >P0, it means that there is a risk of visibility in bad weather for the smart driving vehicle.