Unmanned driving safe car following distance calculation method and safe car following method

Abstract

The invention discloses an unmanned driving safe vehicle following distance calculation method. The method comprises the following steps: step 1, data acquisition; 2, data preprocessing; 3, data fusion: fusing the information of the distance d and the speed v of the vehicle after data preprocessing in the step 2 with the humidity h, the temperature t and the rainfall intensity r sensed by the environment sensing sensor group to obtain comprehensive data; 4, determining to select a first algorithm model or a second algorithm model according to the numerical range of the comprehensive data; and 5, determining a safe car following distance. According to the invention, by integrating multi-source sensors such as the vehicle-mounted radar sensor and the environment sensing sensor group, all-around real-time monitoring of surrounding vehicles and environment conditions is realized. According to the characteristic, the scheme can accurately sense and dynamically adjust the vehicle following distance in a complex and changeable traffic environment, especially under the condition of severe weather, so that traffic accidents such as rear-end collision and the like are effectively prevented, and the safety performance of the unmanned vehicle is greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of unmanned driving technology, and in particular to the problem of how to calculate and maintain a safe following distance in real time and accurately for an unmanned vehicle in a complex and changeable traffic environment, especially under different weather conditions. Background Art

[0002] With the rapid development of driverless technology, the safety performance of driverless vehicles has become the focus of attention. When driving a driverless vehicle, maintaining a safe distance from the vehicle in front is the key to avoiding rear-end collisions and ensuring driving safety. However, traditional methods for calculating the following distance are often based on fixed algorithm models and are difficult to adapt to complex and changing traffic environments and weather conditions. Especially in severe weather conditions such as rainy and foggy days, the road surface is slippery and visibility is reduced. Traditional methods for calculating the following distance often cannot provide sufficient safety guarantees. Summary of the invention

[0003] In view of the problems existing in the background technology, the present invention provides a method for calculating the safe following distance of an unmanned driving vehicle and a safe following method.

[0004] Technical solution:

[0005] A method for calculating the safe following distance of an unmanned driving vehicle comprises the following steps:

[0006] Step 1: Data collection: the vehicle-mounted radar sensor detects the distance and speed information of surrounding vehicles in real time and transmits it to the data processing unit at a fixed frequency; the environmental perception sensor group detects the humidity, temperature and rainfall intensity of the current environment respectively, and transmits these data to the data processing unit at a fixed frequency;

[0007] Step 2: Data preprocessing;

[0008] Step 3: Data fusion: The distance and speed information of the vehicle after data preprocessing in step 2 is fused with the humidity, temperature, and rainfall intensity sensed by the environmental perception sensor group to obtain comprehensive data;

[0009] Step 4: Determine whether to use the first algorithm model or the second algorithm model to determine the safe following distance based on the numerical range of the comprehensive data;

[0010] Step 5: Determine safe following distance:

[0011] The safe following distance d is obtained by the first algorithm model safe :

[0012]

[0013] In the formula, v currentrepresents the current driving speed, μ is the friction coefficient, g is the acceleration of gravity, k1, k2 and k3 are weight coefficients, d 0 is the current measured distance;

[0014] The safe following distance d is obtained by the second algorithm model safe :

[0015]

[0016] In the formula, v current represents the current driving speed, k4, k5 and k6 are weight coefficients; a and b are constants obtained by fitting experimental data, and r 0 is the current measured rainfall intensity, and is used to describe the exponential model of how the friction coefficient changes with rainfall intensity.

[0017] Preferably, the data preprocessing in step 2 includes:

[0018] Data cleaning: Use threshold method to remove noise and outliers;

[0019] Data compression: Use data compression algorithms to reduce the amount of data to increase processing speed

[0020] Data calibration: Adjust the data of each sensor through the calibration algorithm to ensure the consistency and accuracy of the data.

[0021] Preferably, in step 3, the comprehensive data is obtained by fusion through the following formula:

[0022] Comprehensive data = ω d ·d+ω v ·v+ω h ·h+ω t ·t+ω r ·r

[0023] Among them, ω d ,ω v ,ω g ,ω t , and ω r are the weights of distance, speed, humidity, temperature and rainfall intensity, respectively, and d, v, h, t and r represent the standardized data of distance, speed, humidity, temperature and rainfall intensity, respectively.

[0024] Preferably, the standardization is performed by the following formula:

[0025]

[0026] Among them, m represents the standardized data, m 0 Indicates the measured value of the data, m max and m min Respectively represent the maximum and minimum values ​​in the historical data.

[0027] Preferably, the constants and weight coefficients of the algorithm model are optimized and updated through a machine learning algorithm; during driving, the system will continuously collect historical data and current environmental change information, and use these data to train and adjust the algorithm model to improve the accuracy and adaptability of the calculation.

[0028] Preferably, the machine learning algorithm includes neural network and reinforcement learning.

[0029] A method for calculating the safe following distance of an unmanned vehicle is disclosed. After obtaining the safe following distance, a control decision unit calculates the safe following distance d. safe Converted into specific control instructions and output to the actuator of the unmanned vehicle;

[0030] The actuator adjusts the vehicle's speed and following distance in real time according to the received control instructions to maintain a safe distance from the vehicle in front.

[0031] Preferably, the control instructions include brake pressure and throttle opening.

[0032] Preferably, the actuator includes a brake system and a throttle control system.

[0033] Beneficial Effects of the Invention

[0034] The present invention has shown remarkable beneficial effects in significantly improving the safety performance, adaptability and intelligence level of unmanned vehicles.

[0035] From the perspective of safety, the solution integrates multi-source sensors such as vehicle-mounted radar sensors and environmental perception sensor groups to achieve all-round real-time monitoring of surrounding vehicles and environmental conditions. This feature enables the solution to accurately sense and dynamically adjust the following distance in complex and changing traffic environments, especially in bad weather conditions (such as rainy days and foggy days), thereby effectively preventing traffic accidents such as rear-end collisions, greatly improving the safety performance of unmanned vehicles.

[0036] Secondly, the solution is highly adaptable. Through the application of dynamic friction coefficient model and real-time weather adaptation algorithm, the solution can intelligently adjust the following distance according to different weather conditions (such as road wetness, visibility, etc.), ensuring that a safe distance can be maintained in different environments. In addition, the solution's adaptive learning mechanism also allows it to continuously optimize the algorithm model based on historical data and current environmental changes, further improving its adaptability and accuracy.

[0037] Furthermore, the solution's intelligence level has been significantly improved. Through multi-sensor fusion technology and complex algorithm models, the solution can process and analyze large amounts of data in real time, make decisions quickly and output control instructions, allowing unmanned vehicles to respond to various driving conditions more intelligently. This level of intelligence not only improves the driving safety of unmanned vehicles, but also makes it possible for them to be used in a wider range of scenarios.

[0038] In addition, this solution will also help promote the further development of driverless technology. By continuously optimizing and improving the solution algorithms and models, the safety performance and intelligence level of driverless vehicles can be further improved, accelerating the popularization and application of driverless technology. At the same time, this solution can also provide strong support for the construction of intelligent transportation solutions and promote the intelligent and green development of urban transportation.

[0039] In summary, the unmanned driving safe following distance calculation scheme and method of the present invention have significant beneficial effects, which not only improve the safety performance and adaptability of unmanned vehicles, but also promote the further development of unmanned driving technology and the construction of intelligent transportation solutions. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 The figure is a flow chart of the method of the present invention.

[0041] Figure 2 It is a system module diagram of the present invention. DETAILED DESCRIPTION

[0042] The present invention will be further described below in conjunction with embodiments, but the protection scope of the present invention is not limited thereto:

[0043] Combination Figure 1 , a method for calculating the safe following distance of an unmanned driving vehicle, in a specific example, comprises the following steps:

[0044] Step 1: Data Collection

[0045] The vehicle-mounted radar sensor detects the distance (unit: meter) and speed (unit: meter / second) information of surrounding vehicles in real time and transmits it to the data processing unit at a fixed frequency (e.g., 10 times per second).

[0046] The environmental sensing sensor group (humidity sensor, temperature sensor, raindrop sensor, etc.) detects the humidity (unit: %), temperature (unit: Celsius) and rainfall intensity (unit: mm / min) of the current environment, and transmits these data to the data processing unit at a fixed frequency.

[0047] Step 2: Data Preprocessing

[0048] Data cleaning: Use thresholding to remove noise and outliers. For example, set a reasonable range for distance and speed, and data outside the range will be considered as outliers and removed.

[0049] Data compression: Use data compression algorithms (such as Huffman coding) to reduce the amount of data to increase processing speed while keeping the data accuracy within an acceptable range.

[0050] Data calibration: Adjust the data of each sensor through calibration algorithms (such as least squares method) to ensure data consistency and accuracy.

[0051] Step 3: Data Fusion

[0052] The vehicle distance and speed information detected by the on-board radar sensor is integrated with the weather conditions (such as humidity, temperature, and rainfall intensity) perceived by the environmental perception sensor group.

[0053] The fusion algorithm includes weighted fusion to form a more comprehensive and accurate traffic environment perception result (such as the effective braking distance under the influence of comprehensive weather).

[0054] Specifically, data fusion assigns different weights to data from different sources and then calculates the weighted sum to obtain a comprehensive result.

[0055] For the vehicle distance and speed information detected by the on-board radar sensor, and the weather conditions (humidity, temperature, rainfall intensity) perceived by the environmental perception sensor group, a weight can be assigned to each parameter, and then the weighted sum can be calculated to form a more comprehensive and accurate traffic environment perception result.

[0056] In a preferred embodiment, the weighted fusion formula can be expressed as:

[0057] Comprehensive data = ω d ·d+ω v ·v+ω h ·h+ω t ·t+ω r ·r

[0058] Among them, ω d ,ω v ,ω h ,ω t , and ω r are the weights for distance, speed, humidity, temperature and rainfall intensity, and their values ​​are based on the importance and reliability of each parameter. The sum of these weights should equal 1.

[0059] In practical applications, the weights can be adjusted flexibly. For example, if the rainfall intensity has a great impact on the braking distance, then ω r May be assigned a higher value.

[0060] After weighted fusion, a comprehensive perception result can be obtained, which will be used in subsequent calculation and decision-making processes.

[0061] Step 4: If the result of the comprehensive data normalization calculation is less than 0.5, the first algorithm model is used; otherwise, the second algorithm model is used to determine the safe following distance;

[0062] The safe following distance d is obtained by the first algorithm model safe :

[0063]

[0064] In the formula, v current represents the current driving speed, μ is the friction coefficient, g is the acceleration of gravity, k1, k2 and k3 are weight coefficients, d 0 is the current measured distance;

[0065] The second algorithm model takes into account the rainy day environment. Under rainy day conditions, the control decision unit will select the dynamic friction coefficient model (μ = f (r)), which is a mapping function from the rainfall intensity r to the tire-road friction coefficient μ.

[0066] In order to describe this model more specifically, we use a simplified nonlinear function form, the exponential model. Using the exponential model, the friction coefficient μ can be expressed as:

[0067] μ=a·e -br

[0068] Among them, a and b are constants. As the rainfall intensity r increases, the friction coefficient μ decays exponentially, reflecting the phenomenon of reduced friction caused by slippery road surfaces on rainy days.

[0069] Next, combined with the real-time weather adaptation algorithm, the following distance d is dynamically adjusted according to the current rainfall intensity r safe The calculation formula is given by safe =G(V current, μ,d-eff), where G is a complex function that takes into account the current speed V current , friction coefficient μ and effective braking distance d-eff. The effective braking distance d-eff is related to the maximum deceleration and initial velocity of the vehicle, but in rainy conditions, the change in friction coefficient also needs to be considered. Therefore, d-eff is expressed as:

[0070]

[0071] Here, g is the acceleration due to gravity.

[0072] Substitute the above μ and d-eff into dsafe In the calculation formula, we get:

[0073]

[0074] Since the specific form of the G function is relatively complex and involves the interaction of multiple parameters, we simplify it in this solution. G is a weighted sum or product, that is:

[0075]

[0076] In the formula, v current represents the current driving speed, k4, k5 and k6 are weight coefficients; a and b are constants obtained by fitting experimental data, and r 0 is the current measured rainfall intensity, which is used to describe the exponential model of the friction coefficient changing with rainfall intensity. This formula comprehensively considers the influence of speed, friction coefficient and effective braking distance on the following distance as the second algorithm model.

[0077] The control decision unit calculates the safe following distance d that meets the current driving conditions through the above specific algorithm based on the selected algorithm model and related parameters (such as the current driving speed Vcurrent, the current measured distance, etc.). safe During the calculation process, it is necessary to obtain the data of rainfall intensity r in real time, calculate the μ value through the dynamic friction coefficient model, and then substitute it into the calculation formula of the following distance.

[0078] Case 1 (using the first algorithm model):

[0079] (1) Scenario description:

[0080] Environmental conditions: sunny day (rainfall intensity r = 0 mm / min), temperature 50°C, humidity 40%;

[0081] Vehicle status: current speed v current =20m / s (about 72km / h), the vehicle-mounted radar detects the distance of the vehicle in front d=50m.

[0082] Road friction coefficient: μ=0.7 (standard reference value).

[0083] (2) Data standardization and integration

[0084] ①Data standardization

[0085] In order to eliminate the influence of different dimensions and dimensional units, the data is standardized. The 0-1 standardization method is used to map each parameter to the [0,1] interval.

[0086] Distance d standardization: Assuming that the maximum distance in the historical data is 100 meters and the minimum distance is 10 meters, then:

[0087]

[0088] Speed ​​v normalization: Assuming that the maximum speed in the historical data is 30 m / s and the minimum speed is 5 m / s, then:

[0089]

[0090] Humidity h standardization: Assuming that the maximum humidity in the historical data is 100% and the minimum humidity is 30%, then:

[0091]

[0092] Temperature t standardization: Assuming that the maximum temperature in the historical data is 35 degrees Celsius and the minimum temperature is 5 degrees Celsius, then:

[0093]

[0094] Standardization of rainfall intensity r: Assuming that the maximum rainfall intensity in historical data is 10 mm / min and the minimum rainfall intensity is 0 mm / min, then:

[0095]

[0096] ②Data fusion

[0097] According to the formula in the patent, the standardized data is weighted and fused. The weights are:

[0098] ω d =0.3;ω v =0.2;ω h =0.1;ω t =0.1;ω r =0.3; (the sum is 1).

[0099] Comprehensive data = 0.3 × 0.444 + 0.2 × 0.6 + 0.1 × 0.143 + 0.1 × 0.667 + 0.3 × 0 = 0.334

[0100] After weighted fusion, the result is less than 0.5, triggering the first algorithm model.

[0101] Model parameters:

[0102] The weight coefficient of the first algorithm is optimized through historical data: k 1 =0.4, k 2 =0.5, k 3 =0.1. g = 9.8 m / s 2

[0103] (3) Calculation process

[0104] According to the first algorithm formula:

[0105] Case 2 (using the second algorithm model):

[0106] (1) Scenario Description

[0107] Weather conditions: heavy rain, rainfall intensity r = 5 mm / min, temperature 20°C, humidity 80%.

[0108] Vehicle status: current speed v current =20m / s(72km / h), the radar detects the front vehicle distance d=50m.

[0109] Dynamic friction coefficient: fitting parameters a = 0.8, b = 0.1, calculated μ = 0.8e -0.5 ≈0.4852.

[0110] (2) Data standardization and integration

[0111] ①Data standardization

[0112] Distance d standardization: Assuming that the maximum distance in the historical data is 100 meters and the minimum distance is 10 meters, then:

[0113]

[0114] Speed ​​v normalization: Assuming that the maximum speed in the historical data is 30 m / s and the minimum speed is 5 m / s, then:

[0115]

[0116] Humidity h standardization: Assuming that the maximum humidity in the historical data is 100% and the minimum humidity is 80%, then:

[0117]

[0118] Temperature t standardization: Assuming that the maximum temperature in the historical data is 35 degrees Celsius and the minimum temperature is 5 degrees Celsius, then:

[0119]

[0120] Standardization of rainfall intensity r: Assuming that the maximum rainfall intensity in historical data is 10 mm / min and the minimum rainfall intensity is 0 mm / min, then:

[0121]

[0122] ②Data fusion

[0123] According to the formula in the patent, the standardized data is weighted and fused. The weights are:

[0124] ω d =0.3;ω v =0.2;ω h =0.1;ω t =0.1;ω r =0.3; (the sum is 1).

[0125] Comprehensive data = 0.3 × 0.444 + 0.2 × 0.6 + 0.1 × 0.714 + 0.1 × 0.5 + 0.3 × 0.5 = 0.524

[0126] After weighted fusion, the result is greater than 0.5, triggering the second algorithm model.

[0127] Model parameters:

[0128] The weight coefficient of the second algorithm is optimized through historical data: k 4 =0.3, k 5 =0.5, k 6 =0.2. g = 9.8 m / s 2

[0129] (3) Calculation process

[0130] According to the first algorithm formula:

[0131] The calculation process fully considers the relative speed and distance between vehicles, as well as the specific impact of weather conditions on braking performance (such as the increase in braking distance due to slippery roads). In this way, the system can dynamically adjust the following distance according to different weather conditions to ensure driving safety.

[0132] Step 6: Control command output

[0133] The control decision unit calculates the safe following distance d safe It is converted into specific control instructions (such as brake pressure, throttle opening, etc.) and output to the actuators of the unmanned vehicle (such as brake system, throttle control system).

[0134] The actuator adjusts the vehicle's speed and following distance in real time according to the received control instructions to maintain a safe distance from the vehicle in front.

[0135] Step 7: Adaptive Learning and Optimization

[0136] The system has an adaptive learning mechanism that optimizes and updates the algorithm model through machine learning algorithms (such as neural networks, reinforcement learning, etc.).

[0137] During driving, the system will continuously collect historical data and current environmental change information, and use this data to train and adjust the algorithm model to improve the accuracy and adaptability of the calculation.

[0138] The present invention also discloses an unmanned driving safe following method. Based on the unmanned driving safe following distance calculation method, after obtaining the safe following distance, the control decision unit calculates the safe following distance d safe Converted into specific control instructions and output to the actuator of the unmanned vehicle;

[0139] The actuator adjusts the vehicle's speed and following distance in real time according to the received control instructions to maintain a safe distance from the vehicle in front.

[0140] The present invention also proposes an unmanned driving safety following vehicle distance calculation system, combined with Figure 2 , mainly includes the following modules:

[0141] Vehicle-mounted radar sensor: used to detect the distance and speed information of surrounding vehicles in real time.

[0142] Environmental perception sensor group: including humidity sensors, temperature sensors, raindrop sensors, etc., used to perceive current weather conditions and environmental changes in real time.

[0143] Data processing unit: receives data from various sensors and performs pre-processing and fusion analysis.

[0144] Algorithm model library: stores a variety of algorithm models, including dynamic friction coefficient model, real-time weather adaptation algorithm, etc.

[0145] Control decision unit: Based on the output results of the data processing unit and combined with the algorithm model in the algorithm model library, it calculates the safe following distance and outputs control instructions to the actuator of the unmanned vehicle.

[0146] The specific embodiments described herein are merely examples of the spirit of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in similar ways, but they will not deviate from the spirit of the present invention or exceed the scope defined by the appended claims.

Claims

1. A method for calculating the safe following distance of an unmanned vehicle, characterized in that It includes the following steps: Step 1: Data collection: the vehicle-mounted radar sensor detects the distance and speed information of surrounding vehicles in real time and transmits it to the data processing unit at a fixed frequency; the environmental perception sensor group detects the humidity, temperature and rainfall intensity of the current environment respectively, and transmits these data to the data processing unit at a fixed frequency; Step 2: Data preprocessing; Step 3: Data fusion: The distance and speed information of the vehicle after data preprocessing in step 2 is fused with the humidity, temperature, and rainfall intensity sensed by the environmental perception sensor group to obtain comprehensive data; Step 4: Determine whether to use the first algorithm model or the second algorithm model to determine the safe following distance based on the numerical range of the comprehensive data; Step 5: Determine safe following distance: The safe following distance d is obtained by the first algorithm model safe : In the formula, v current represents the current driving speed, μ is the friction coefficient, g is the acceleration of gravity, k1, k2 and k3 are weight coefficients, and d0 is the current measured distance; The safe following distance d is obtained by the second algorithm model safe : In the formula, v current represents the current driving speed, k4, k5 and k6 are weight coefficients; a and b are constants obtained by fitting the experimental data, and r0 is the current measured rainfall intensity, which is used to describe the exponential model of the change of friction coefficient with rainfall intensity.

2. The method according to claim 1, characterized in that The data preprocessing in step 2 includes: Data cleaning: Use threshold method to remove noise and outliers; Data compression: Use data compression algorithms to reduce the amount of data to increase processing speed Data calibration: Adjust the data of each sensor through the calibration algorithm to ensure the consistency and accuracy of the data.

3. The method according to claim 1, characterized in that In step 3, the comprehensive data is obtained by fusion through the following formula: Comprehensive data = ω d ·d+ω v ·v+ω h ·h+ω t ·t+ω r ·r Among them, ω d ,ω v ,ω h ,ω t , and ω r are the weights of distance, speed, humidity, temperature and rainfall intensity, respectively, and d, v, h, t and r represent the standardized data of distance, speed, humidity, temperature and rainfall intensity, respectively.

4. The method according to claim 3, characterized in that The normalization is performed by the following formula: Among them, m represents the standardized data, m0 represents the measured value of the data, and m max and m min Respectively represent the maximum and minimum values ​​in the historical data.

5. The method according to claim 1, characterized in that The constants and weight coefficients of the algorithm model are optimized and updated through machine learning algorithms; during driving, the system will continuously collect historical data and current environmental change information, and use these data to train and adjust the algorithm model to improve the accuracy and adaptability of the calculation.

6. The method according to claim 5, characterized in that The machine learning algorithms include neural networks and reinforcement learning.

7. An unmanned driving safety following method, based on an unmanned driving safety following distance calculation method according to any one of claims 1 to 6, characterized in that After obtaining the safe following distance, the control decision unit calculates the safe following distance d safe Converted into specific control instructions and output to the actuator of the unmanned vehicle; The actuator adjusts the vehicle's speed and following distance in real time according to the received control instructions to maintain a safe distance from the vehicle in front.

8. The method according to claim 7, characterized in that The control instructions include brake pressure and throttle opening.

9. The method according to claim 7, characterized in that The actuator includes a brake system and a throttle control system.

Images