An automatic driving BEV safe driving space construction method

Abstract

The application relates to an automatic driving BEV safe driving space construction method, the method comprises the following steps: S1, acquiring environment target perception information of an automobile to an environment, identifying a drivable boundary of the automobile, and performing boundary type risk assessment on the drivable boundary according to the environment target information; S2, correcting a risk assessment result of the boundary type risk assessment according to perception uncertainty in the environment target perception information to obtain a corrected risk assessment result; and S3, delimiting an adaptive safety margin according to the perception uncertainty and the corrected risk assessment result, and constructing a safe driving space in which the automobile can safely drive according to the adaptive safety margin.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology for automobiles, and in particular to a method for constructing a safe driving space for an autonomous BEV. Background Art

[0002] As assisted driving functions have gradually been mass-produced, high-level autonomous driving has become a hot topic in research and engineering practice at this stage.

[0003] High-level autonomous vehicles rely on environmental perception systems to obtain information about surrounding vehicles, pedestrians, and other road conditions, enabling them to make autonomous driving decisions. Therefore, the accuracy of environmental perception results is crucial to autonomous driving safety. However, external environmental factors such as weather and lighting can affect the quality of sensor data, and perception algorithms cannot achieve perfect object detection. This can lead to autonomous driving accidents and pose a safety threat.

[0004] From the perspective of the chain of autonomous driving perception, prediction, decision-making, and control, the decision-making layer needs to have the ability to tolerate perception and prediction uncertainty, that is, to recognize the risks caused by perception and prediction uncertainty and make safe driving decisions.

[0005] Research has found that existing research focuses on driving decisions that tolerate predictive uncertainty. Some studies predict areas where surrounding vehicles, pedestrians, and other targets are likely to reach within a certain period of time, allowing the vehicle to avoid these areas to ensure safety. Other studies use reinforcement learning to make driving decisions, modeling the possible future states of surrounding targets using state transition probabilities. These methods all assume that environmental perception results are completely accurate while considering the uncertainty of the target's future behavior. However, when perception results are subject to significant uncertainty, these methods are insufficient to ensure driving safety. Summary of the Invention

[0006] In response to the above problems, the purpose of the present invention is to provide a method for constructing a safe driving space for autonomous driving BEVs, realize environmental risk assessment based on real-time perception results and their uncertainties, and then define an adaptive safety margin to construct a boundary-type safe driving space from the perspective of BEVs, realize the recognition of the environmental risks around the vehicle, and improve the safety of autonomous driving.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] In a first aspect, the present application provides a method for constructing a safe driving space for an autonomous BEV, the method comprising:

[0009] S1, obtains the vehicle's environmental target perception information of the environment, identifies the vehicle's drivable boundary, and performs boundary-type risk assessment on the drivable boundary based on the environmental target information;

[0010] S2, based on the perceived uncertainty in the environmental target perception information, the risk assessment result of the boundary risk assessment is corrected to obtain a corrected risk assessment result;

[0011] S3, defining an adaptive safety margin based on the perceived uncertainty and the corrected risk assessment result, and constructing a safe driving space in which the car can safely drive based on the adaptive safety margin.

[0012] In one implementation, in S1, the environmental target perception information includes the spatial position, semantic category, and motion state of dynamic and static environmental targets detected by the environmental perception sensor of the vehicle.

[0013] In one implementation, in S1, the boundary risk assessment is performed using the microelement method. The total risk of each microelement on the drivable boundary of the vehicle is a weighted superposition of the spatial risk and the motion risk, and the formula is:

[0014]

[0015] Where, For the i The total risk of micro-dollars, and are the weights of spatial risk and motion risk, and They are spatial risk and motion risk respectively.

[0016] In one implementation, in a polar coordinate system centered on the vehicle, the spatial risk calculation formula is:

[0017]

[0018] Where, Represents the semantic category of the microelement, is the risk weight of the semantic category; is the polar diameter corresponding to the infinitesimal element length; is the spatial risk power index, set to a negative value.

[0019] In one implementation, in a polar coordinate system centered on the vehicle, the motion risk calculation formula is:

[0020]

[0021] In the formula represents the risk weight of the semantic category, Represents the radial component of velocity , where the speed can be absolute or relative; and are the power exponents of velocity and polar radius respectively.

[0022] In one implementation, the perception uncertainty in the environmental target perception information includes: target existence probability of the detected environmental target, target semantic category uncertainty, and target motion speed uncertainty;

[0023] According to the above uncertainties, the total risk of each infinitesimal element on the drivable boundary of the car is corrected respectively.

[0024] In one implementation, the total risk of each microelement on the drivable boundary of the vehicle is corrected. The calculation process includes:

[0025]

[0026] Where, represents the risk of the boundary segment obtained by the differential element method, Represents the corrected value of the boundary segment risk after correction based on the probability of target existence. is the probability of target existence;

[0027]

[0028] Where, is the correction value of the boundary segment risk after considering the uncertainty of the target semantic category;

[0029]

[0030] Where, It is the correction value of the boundary segment risk after considering the uncertainty of the target motion speed.

[0031] In one implementation, in S3, the safety margin is related to the size of the boundary risk and the uncertainty of the target space information.

[0032] In one implementation, the method further includes:

[0033] S4, obtaining a prediction result of the safe driving space, combining the prediction result with the already constructed safe driving space, and forming a time-series safe driving space.

[0034] In a second aspect, a computer-readable storage medium is also provided, which stores a computer program. The computer program is executed by a processor to implement the method for constructing a safe driving space for an autonomous driving BEV as described in the first aspect.

[0035] By adopting the above technical solutions, the present invention has the following advantages: a method for constructing a safe driving space for autonomous BEVs based on perceptual uncertainty. First, for dynamic and static environmental targets such as vehicles, pedestrians, and road boundaries, the spatial position, semantic category, and motion state information are considered. Based on the boundary of the drivable area, the distribution of environmental risks along the boundary of the drivable area is analyzed without considering perceptual uncertainty. Second, the boundary risk is corrected based on the target's probability of existence, semantic category uncertainty, and motion state uncertainty. Third, based on the target's spatial position uncertainty and the corrected boundary risk, an adaptive safety margin is defined to construct a boundary-based safe driving space. Finally, combined with the prediction of the target's future state, a time-series safe driving space is formed, providing spatiotemporal constraints for autonomous driving decision-making and improving the safety of autonomous driving. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a flowchart of constructing a safe driving space for an autonomous driving BEV provided by an embodiment of the present invention;

[0037] Figure 2 is a schematic diagram of constructing the boundary of a drivable area in one embodiment of the present invention;

[0038] Figure 3 is a schematic diagram of boundary risk assessment in one embodiment of the present invention;

[0039] Figure 4 This is a schematic diagram of constructing a safe driving space based on perception uncertainty in one embodiment of the present invention. DETAILED DESCRIPTION

[0040] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field are within the scope of protection of the present invention.

[0041] In one embodiment of the present application, Figure 1 , provides a method for constructing a safe driving space for an autonomous driving BEV, the method comprising:

[0042] S1, obtains the vehicle's environmental target perception information of the environment, identifies the vehicle's drivable boundary, and performs boundary-type risk assessment on the drivable boundary based on the environmental target information;

[0043] Specifically, first, according to the spatial positions of various surrounding environmental targets, the spatial area they occupy can be delineated, and then the boundary of the drivable area centered on the vehicle can be delineated. Figure 2 As shown in the figure, the drivable area, centered on the ego vehicle, is comprised of the boundaries of the range detectable by sensors such as lidar and cameras, traffic regulations such as stop lines or double solid lines, and obstacles such as vehicles and pedestrians. From a BEV perspective, this boundary can be described using a polar coordinate system centered on the ego vehicle. The boundary of the drivable area not only reflects whether the surrounding space is passable but also further reflects information such as the type of objects that constitute the boundary and their motion status.

[0044] For each microelement on the boundary, the risk of the microelement can be evaluated based on its position relative to the vehicle, motion state, and semantic category, such as Figure 3 shown.

[0045] In the figure, consider The boundary element, in the polar coordinate system centered on the vehicle, the polar radius vector of the element is recorded as , the velocity vector is , radial velocity, that is, the projection of velocity in the radial direction, is recorded as , the unit normal vector of the infinitesimal element is recorded as .

[0046] Boundary elements That is, the line segment in the figure , in the figure AB and AC vertical, and AC With the polar vector Vertical. Boundary element Can be used with angle differential Mutual transformation, according to the general rules of polar coordinate system, is:

[0047] (1)

[0048] According to the angle relationship, we can further have:

[0049] (2)

[0050] and The angle between Mutually complementary, which can be calculated as the inner product of vectors. Therefore:

[0051] (3)

[0052] In the formula and They are and Based on this, it can be achieved and mutual transformation.

[0053] Based on the radius, speed and semantic category of the microelement, the boundary risk of the microelement can be modeled.

[0054] In boundary-based risk assessment, risks are categorized into spatial and motion risks. Spatial risk considers the spatial position information of various dynamic and static targets, while motion risk considers both the target's spatial position and motion state, and also considers the ego vehicle's motion state to model the risk associated with relative motion. Because different target categories generate varying degrees of risk, the target's semantic category serves as a parameter in the calculation of spatial and motion risks. This allows the risk of the traffic environment to be modeled and expressed at the boundaries of the drivable area.

[0055] (1) The trend that spatial risk should reflect is that the smaller the drivable area is, the smaller the boundary diameter in the polar coordinate system centered on the vehicle is, and the greater the risk is. The risk of a microelement can be modeled as follows:

[0056] (4)

[0057] Where, Represents the semantic category of the microelement, is the risk weight of the semantic category. Different semantic categories such as vehicles and pedestrians correspond to different weights, which can be set according to actual conditions. is the polar diameter corresponding to the infinitesimal element length. is the spatial risk power index. Since the spatial risk is negatively correlated with the size of the polar diameter, it should be a negative value and can be set according to actual conditions.

[0058] (2) The calculation of motion risk is similar to that of occupation risk. However, due to the existence of “absolute speed” relative to the ground and “relative speed” relative to the vehicle, motion risk can be calculated as follows:

[0059] (5)

[0060] Where, The motion risk caused by absolute speed, The motion risk caused by relative speed, and are their coefficients respectively, which can be set according to actual conditions.

[0061] The analysis method for the risks caused by absolute speed and relative speed is the same. Considering that the movement of the target toward the ego vehicle will cause risk to the ego vehicle, the motion risk should be positively correlated with the radial speed of the microelement. At the same speed, the closer the target is to the ego vehicle, the greater the risk. The motion risk should also be negatively correlated with the distance from the microelement to the ego vehicle, that is, the size of the radial radius. , the exercise risk can be calculated by the following formula:

[0062] (6)

[0063] In the formula represents the risk weight of the semantic category, Represents the radial component of velocity , where the speed can be absolute or relative. and They are the power exponents of velocity and polar radius respectively, which can be set according to actual conditions.

[0064] (3) Considering both spatial risk and motion risk, the total risk of the microelement is:

[0065] (7)

[0066] Where, For the i The total risk of micro-dollars, and are the weights of spatial risk and movement risk, respectively, which can be set according to actual conditions. The total risk of the traffic environment can be further calculated by summing the risks of each microelement.

[0067] S2, based on the perceived uncertainty in the environmental target perception information, the risk assessment result of the boundary risk assessment is corrected to obtain a corrected risk assessment result;

[0068] Specifically, the perception uncertainty information includes the probability of target existence, the uncertainty of target semantic category, the uncertainty of target position, and the uncertainty of target motion speed. The boundary risk can be corrected according to the perception uncertainty, such as Figure 4 (c) shown.

[0069] (1) Risk adjustment considering the probability of target existence

[0070] Due to the uncertainty of perception, the perceived target may not actually exist, but may be a false detection. This uncertainty in perception of whether the target exists can be expressed as the probability of the target existing. To describe. The probability that the perceived target actually exists is The probability that the perceived target does not actually exist.

[0071] When considering the probability of the target's existence, its risk will be probabilistic, that is, The probability of generating full risk; The probability that it does not generate risk. The boundary generated by the target is recorded as the boundary segment generated by the target, and it is set as the first When the perception uncertainty is not considered, the risk of the boundary segment is recorded as ,like Figure 4 (b) The risk of a boundary segment can be calculated by integrating the risk of the infinitesimal element on the boundary segment (which can be obtained in step 1).

[0072] When considering the probability of target existence, the risk of the boundary segment generated by the target can be modified as follows:

[0073] (8)

[0074] Where, represents the risk of the boundary segment without considering the existential uncertainty, Represents the revised value of the boundary segment risk after considering existential uncertainty.

[0075] (2) Risk correction considering the uncertainty of target semantic category

[0076] There is also uncertainty in the semantic category of the perceived target. For example, the most likely category of a target is a vehicle, but it could also be a pedestrian, a cone, or other targets. In this case, when calculating the risk using Equations (4) and (6), if only the target is considered as the most likely semantic category (i.e., vehicle), its risk cannot be correctly reflected, so correction is required.

[0077] The uncertainty of the target semantic category can be expressed as, its possible semantic categories are The probability of the first possible semantic category is recorded as , the probability of the second possible semantic category is recorded as , and so on.

[0078] When considering the uncertainty of the target semantic category, the risk of the boundary segment generated by the target can be modified as follows:

[0079] (9)

[0080] It can also be written as:

[0081] (10)

[0082] Where, is the correction value of the boundary segment risk after considering the uncertainty of the target semantic category, It is the risk value of the boundary segment corrected according to the probability of target existence in the previous step. is the risk weight of the semantic category, is the risk weight corresponding to the most likely semantic category, i.e., the risk weight used in the risk assessment in step 1 without considering the perceived uncertainty.

[0083] (3) Risk correction considering target motion speed uncertainty

[0084] The speed of a perceived object detected by the perception system is not completely accurate. Unlike the existence of an object or its semantic category, speed information is a continuous value rather than a discrete value. It can take values ​​within a certain range and is randomly distributed according to a certain probability density function.

[0085] When considering the uncertainty of the target's motion speed, the risk of the boundary segment generated by the target can be modified as follows:

[0086] (11)

[0087] Where, is the correction value of the boundary segment risk after considering the uncertainty of the target motion speed, It is the risk value of the boundary segment corrected in the previous step according to the target existence probability and semantic category uncertainty.

[0088] is the speed value taken without considering the uncertainty of motion speed, is the power exponent of speed used in calculating motion risk in formula (6), and are the upper and lower bounds of the possible speed range, The speed is The probability density of .

[0089] S3, defining an adaptive safety margin based on the perceived uncertainty and the corrected risk assessment result, and constructing a safe driving space in a safe form for the vehicle based on the adaptive safety margin.

[0090] The spatial position of a perceived target detected by the perception system is not completely accurate. Similar to the uncertainty of motion speed, spatial position uncertainty is also a continuous value, which is reflected in the fact that the radius may take values ​​within a certain range and is randomly distributed according to a certain probability density function.

[0091] Based on the uncertainty of target existence, target semantic category, and target motion speed to modify the boundary risk, we can further set a safety margin based on the uncertainty of the target's spatial position to construct a safe driving space. The safety margin is related to the size of the boundary risk and the uncertainty of the target space information.

[0092] In order to set the safety margin, first analyze the risk value of the area inside the boundary based on the boundary risk. The risk at the target is attenuated according to the probability distribution of the target location uncertainty. Figure 4 (d) shows that The polar radius of a boundary segment is , then the radial distance from the boundary The polar diameter at , the risk here can be expressed as follows:

[0093] (12)

[0094] Where, is the radial inward distance from the boundary segment, is the boundary segment risk corrected based on the perceived uncertainty in step 2, is the radial inward distance of the boundary segment The risk value at is the spatial position probability density at that location, is the spatial position probability density at the boundary segment.

[0095] Based on this, we can draw the risk value curve of the inner area of ​​the boundary along the extreme radius distribution, such as Figure 4 (b)

[0096] Given a risk threshold , requiring that the risk inside the safe driving space cannot exceed the risk threshold, the safety margin can be obtained as:

[0097] (13)

[0098] Where, is the safety margin, is the inverse distribution of the target position uncertainty distribution.

[0099] From the above formula, we can see that According to formula (12), the risk value at the safety boundary is equal to the risk threshold, while inside the safety boundary, the risk value is less than the risk threshold. After calculating the safety margin at each extreme angle along the entire drivable area boundary, the closed safe driving space boundary can be obtained:

[0100] (14)

[0101] Where, is the boundary point sequence of the boundary safe driving space, is the boundary point of the first safe driving space, expressed in polar coordinates. and are the polar radius and polar angle of the boundary point, and so on. is the number of boundary points of the safe driving space. Figure 4 As shown in (d), based on the drivable area, After the safety margin is reduced, the safe driving space boundary can be formed.

[0102] S4, obtaining a prediction result of the safe driving space, combining the prediction result with the already constructed safe driving space, and forming a time-series safe driving space.

[0103] Combined with prediction, the environmental state at several future time points can be examined to construct a safe driving space according to formula (14). However, the prediction of future environmental state will also produce uncertainty. For a certain target, its prediction uncertainty is reflected as the inaccuracy of the spatial position at future time points, which can be expressed by standard deviation.

[0104] Considering the prediction uncertainty, the safe driving space boundary should be further revised as follows:

[0105] (15)

[0106] Where, is the standard deviation of the predicted position of the first boundary point, and so on. Due to the combined effect of perception uncertainty and prediction uncertainty, a superimposed safety margin should be set.

[0107] Consider the future At each moment, a time series of safe driving space boundaries can be formed:

[0108] (16)

[0109] Where, is the time series of the boundary of the boundary-type safe driving space.

[0110] From the above-mentioned construction method of the boundary-type safe driving space, it can be seen that the risk values ​​in the boundary-type safe driving space are all less than the risk threshold, so it can be considered safe. The decision algorithm plans the path in the boundary-type safe driving space to meet the safety requirements. Therefore, within the future time range considered by the decision, the future The trajectory points at each moment should satisfy the following constraints:

[0111] (17)

[0112] Where, is the decision trajectory, which is The trajectory points are composed of For the Track points, Meaning The trajectory point must fall on The interior of the safe driving space boundary.

[0113] The risk constraints specified in the above formula are applicable to different types of decision-making algorithms. Decision-making algorithms based on different principles, such as analytical solution, sampling screening, and reinforcement learning, can all use the requirements given in the above formula to verify and verify their trajectory point sequences to meet safety requirements. Based on this, from the perspective of BEV, it is possible to cope with the uncertainty of perception results based on safe driving space and improve driving safety.

[0114] The present application also provides a computer-readable storage medium, which includes a stored program, wherein when the program is executed, the device where the storage medium is located is controlled to execute the above method. The specific implementation process will not be repeated here.

[0115] The present application also provides a computer device. This embodiment of the computer device includes a processor, a memory, and a computer program stored in the memory and executable by the processor. When the processor executes the computer program, the aforementioned method of the embodiment is implemented. To avoid repetition, a detailed description thereof is omitted here. Alternatively, when the processor executes the computer program, the functions of each model / unit in the apparatus of the embodiment are implemented. To avoid repetition, a detailed description thereof is omitted here.

[0116] A computer device may be a desktop computer, laptop, PDA, server, or cloud server, among other computing devices. A computer device may include, but is not limited to, a processor and memory. Those skilled in the art will appreciate that a computer device may include more or fewer components than shown, or a combination of certain components, or different components. For example, a computer device may also include input / output devices, network access devices, and buses.

[0117] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0118] Memory can be an internal storage unit of a computer device, such as a computer device's hard drive or memory. It can also be an external storage device, such as a plug-in hard drive, a Smart Media Card (SMC), a Secure Digital (SD) card, or a flash memory card. Furthermore, memory can include both internal and external storage units. Memory is used to store computer programs and other programs and data required by the computer device. Memory can also be used to temporarily store data that has been output or is about to be output.

[0119] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0120] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the above units is merely a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0121] The aforementioned integrated unit implemented as a software functional unit can be stored in a computer-readable storage medium. The software functional unit, stored in a storage medium, includes instructions for causing a computer device (which may be a personal computer, server, or network device, etc.) or a processor to execute portions of the aforementioned method steps in each embodiment of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a removable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0122] The above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for constructing a safe driving space for an autonomous BEV, characterized in that: The method comprises: S1, obtains the vehicle's environmental target perception information of the environment, identifies the vehicle's drivable boundary, and performs boundary-type risk assessment on the drivable boundary based on the environmental target information; S2, based on the perceived uncertainty in the environmental target perception information, the risk assessment result of the boundary risk assessment is corrected to obtain a corrected risk assessment result; S3, defining an adaptive safety margin based on the perceived uncertainty and the corrected risk assessment result, and constructing a safe driving space in which the vehicle can safely drive based on the adaptive safety margin; Perceptual uncertainty in environmental target perception information includes: the probability of existence of detected environmental targets, uncertainty in target semantic categories, and uncertainty in target motion speed; According to the above uncertainties, the total risk of each microelement on the drivable boundary of the car is modified respectively; The total risk of each infinitesimal element on the drivable boundary of the car is corrected. The calculation process includes: Where, represents the risk of the boundary segment obtained by the differential element method, Represents the corrected value of the boundary segment risk after correction based on the probability of target existence. is the probability of target existence; Where, is the correction value of the boundary segment risk after considering the uncertainty of the target semantic category; Where, It is the correction value of the boundary segment risk after considering the uncertainty of the target motion speed.

2. The method for constructing a safe driving space for an autonomous driving BEV according to claim 1, characterized in that: In S1, the environmental target perception information includes the spatial position, semantic category and motion state of the dynamic and static environmental targets detected by the environmental perception sensor of the automobile.

3. The method for constructing a safe driving space for an autonomous driving BEV according to claim 2, characterized in that: In S1, the boundary risk assessment is performed using the microelement method. The total risk of each microelement on the drivable boundary of the car is a weighted superposition of the spatial risk and the motion risk, and the formula is: Where, For the i The total risk of micro-dollars, and are the weights of spatial risk and motion risk, and They are spatial risk and motion risk respectively.

4. The method for constructing a safe driving space for an autonomous driving BEV according to claim 3, characterized in that: In the polar coordinate system centered on the vehicle, the spatial risk calculation formula is: Where, Represents the semantic category of the microelement, is the risk weight of the semantic category; is the polar diameter corresponding to the infinitesimal element length; is the spatial risk power index, set to a negative value.

5. The method for constructing a safe driving space for an autonomous driving BEV according to claim 3, characterized in that: In the polar coordinate system centered on the vehicle, the motion risk calculation formula is: In the formula represents the risk weight of the semantic category, Represents the radial component of velocity , where the speed can be absolute or relative; and are the power exponents of velocity and polar radius respectively.

6. The method for constructing a safe driving space for an autonomous driving BEV according to claim 1, characterized in that: In S3, the safety margin is related to the size of the boundary risk and the uncertainty of the target space information.

7. The method for constructing a safe driving space for an autonomous driving BEV according to claim 1, characterized in that: The method further comprises: S4, obtaining a prediction result of the safe driving space, combining the prediction result with the already constructed safe driving space, and forming a time-series safe driving space.

8. A computer-readable storage medium, characterized in that A computer program is stored, and the computer program is executed by a processor to implement the method for constructing a safe driving space for an autonomous driving BEV as described in any one of claims 1 to 7.

Images