Vehicle risk field generation method, device and equipment and storage medium
By constructing a unified sensing coordinate system and fusing observation data from multiple sensing devices, a vehicle risk field is generated, which solves the problem of poor accuracy in vehicle risk perception and achieves higher accuracy in risk assessment and driving safety.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-21
- Publication Date
- 2026-03-10
AI Technical Summary
The different types of information and risk perception standards among different vehicles lead to poor accuracy in risk perception results.
By acquiring sensing data from multiple sensing devices, a unified sensing coordinate system is constructed, the observation vector of the object to be evaluated is determined, and the data is fused through dynamic confidence weights to calculate the final state, generate and broadcast the risk field.
It improves the accuracy of risk assessment and driving safety, enabling vehicles to avoid potential risks in advance.
Smart Images
Figure CN121640707A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of intelligent driving, in particular to a vehicle risk field generation method and device, equipment and a storage medium. BACKGROUND
[0002] With the rapid development of advanced driver assistance systems and autonomous driving technology, the dangerous perception ability of vehicles to the surrounding environment becomes crucial. Since the perception range of the vehicle itself is limited, the Internet of Vehicles technology is generated. Vehicles can communicate with each other through the Internet of Vehicles technology, broadcast their own position information, etc., to improve the dangerous perception range of the vehicle. However, the information types and risk perception standards between different vehicles are different, resulting in poor accuracy of the risk perception result. SUMMARY
[0003] Therefore, the purpose of the present application is to provide a vehicle risk field generation method, device, equipment and storage medium to solve the problem of poor risk perception accuracy of the existing vehicle risk perception method.
[0004] In order to achieve the above-mentioned purpose, the technical scheme adopted by the embodiments of the present application is as follows: In a first aspect, the embodiments of the present application provide a vehicle risk field generation method, comprising: obtaining perception data of each perception device for a perception area, and constructing a perception coordinate system corresponding to the perception area according to the perception data; determining an object to be evaluated in the perception coordinate system, and constructing an observation vector of the object to be evaluated according to observation data of each perception device on the object to be evaluated; obtaining a dynamic confidence weight of each observation data, fusing the dynamic confidence weight and the observation vector to obtain a final state of the object to be evaluated in the perception coordinate system; calculating a plurality of risk factors of the object to be evaluated in the perception coordinate system according to the final state, superimposing a plurality of risk factors to obtain a risk field corresponding to the object to be evaluated and broadcasting.
[0005] In an optional embodiment, the perception device includes a road end perception device and a vehicle, and the step of obtaining perception data of each perception device for a perception area comprises: respectively obtaining first perception data collected by the road end perception device, second perception data collected by the vehicle and historical perception data; respectively obtaining a first collection time of the first perception data, a second collection time of the second perception data and a third collection time of the historical perception data; Based on the first acquisition time, the second acquisition time, and the third acquisition time, the first sensing data, the second sensing data, and the historical sensing data are spatiotemporally aligned to obtain the sensing data of each sensing device for the sensing area.
[0006] In an optional implementation, the step of constructing a sensing coordinate system corresponding to the sensing area based on the sensing data includes: The sensing data collected by each of the sensing devices is spatially aligned and stitched together to obtain panoramic data of the sensing area; The coordinate system parameters are determined based on the panoramic data, and the perception coordinate system is constructed based on the coordinate system parameters.
[0007] In an optional implementation, the step of constructing the observation vector of the object to be evaluated based on the observation data of each of the sensing devices includes: From the sensing data of each of the sensing devices, the sensing data corresponding to the object to be evaluated is obtained as the observation data; Calculate the vector corresponding to each observation data in the sensing coordinate system, and fit the vector corresponding to each observation data to obtain the observation vector.
[0008] In an optional implementation, the step of obtaining the dynamic confidence weight for each of the observation data includes: Obtain the influence factors of the dynamic credibility weight, and assign a corresponding dynamic weight to each influence factor; The dynamic weight is calculated based on the property weight corresponding to the data property of each observed data; the property weight includes source type weight, data quality weight, and historical performance weight. The dynamic credibility weight is calculated based on the influence factor and the dynamic weight.
[0009] In an optional implementation, the step of calculating the dynamic weight based on the property weight corresponding to the data property of each observed data includes: Construct an adjacency matrix for every two observations and calculate the Mahalanobis distance between every two observations; The calculation result of the adjacency matrix is determined based on the Mahalanobis distance; Based on all the observation data corresponding to the object to be evaluated, calculate the sum of the calculation results of the adjacency matrix corresponding to the observation data; The cross-validation consistency weight of the observed data is determined based on the sum of the calculation results, and the dynamic weight is calculated based on the cross-validation consistency weight, the source type weight, the data quality weight, and the historical performance weight.
[0010] In an optional implementation, the step of fusing the dynamic confidence weights and the observation vectors to obtain the final state of the object to be evaluated in the perceptual coordinate system includes: Based on the dynamic confidence weight and the observation vector, the observation state corresponding to each observation vector is determined; Obtain the Gaussian distribution corresponding to the observed state, and calculate the state mean and covariance corresponding to the observed state based on the Gaussian distribution; The final state is calculated using Kalman filtering based on the state mean and covariance.
[0011] Secondly, embodiments of the present invention provide a vehicle risk field generation device, comprising: The data acquisition module is used to acquire the perception data of each sensing device for the sensing area, and to construct a sensing coordinate system corresponding to the sensing area based on the perception data; A vector construction module is used to determine the object to be evaluated in the sensing coordinate system and construct the observation vector of the object to be evaluated based on the observation data of each sensing device on the object to be evaluated. The state calculation module is used to obtain the dynamic confidence weight of each observation data, and fuse the dynamic confidence weight and the observation vector to obtain the final state of the object to be evaluated in the perception coordinate system. The risk field calculation module is used to calculate multiple risk factors of the object to be evaluated in the perception coordinate system based on the final state, superimpose the multiple risk factors to obtain the risk field corresponding to the object to be evaluated, and broadcast it.
[0012] Thirdly, embodiments of the present invention provide an electronic device, including a processor and a memory, wherein the memory stores machine-executable instructions that can be executed by the processor, and the processor can execute the machine-executable instructions to implement the vehicle risk field generation method described in the first aspect.
[0013] Fourthly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the vehicle risk field generation method as described in the first aspect.
[0014] This invention provides a method, apparatus, device, and storage medium for generating a vehicle risk field. By constructing a unified perception coordinate system from the perception data of multiple sensing devices, and then calculating the final state of each object to be evaluated in the perception coordinate system using its corresponding observation data and dynamic confidence weight, a risk assessment is performed on the final state, generating a risk field for the object to be evaluated and broadcasting it. This invention can perform risk assessment on all objects within the perception area of the sensing devices, generate the corresponding risk field, and broadcast it, enabling vehicles to take risk avoidance measures in advance, thus improving the accuracy of risk assessment and driving safety.
[0015] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description
[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0017] Figure 1 A block diagram of an electronic device provided by an embodiment of the present invention is shown; Figure 2 A flowchart illustrating a vehicle risk field generation method provided by an embodiment of the present invention is shown; Figure 3 A flowchart illustrating a method for calculating an observation vector according to an embodiment of the present invention is shown. Figure 4 The diagram shows a functional module schematic of a vehicle risk field generation device provided in an embodiment of the present invention.
[0018] Icons: 100 - Electronic device; 110 - Memory; 120 - Processor; 130 - Communication module; 400 - Vehicle risk field generation device; 401 - Data acquisition module; 402 - Vector construction module; 403 - State calculation module; 404 - Risk field calculation module. Detailed Implementation
[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.
[0020] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.
[0021] It should be noted that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0022] Please refer to Figure 1 This is a block diagram of an electronic device 100 provided in an embodiment. The electronic device 100 includes a memory 110, a processor 120, and a communication module 130. The memory 110, processor 120, and communication module 130 are electrically connected to each other directly or indirectly to realize data transmission or interaction. For example, these components can be electrically connected to each other through one or more communication buses or signal lines.
[0023] The memory 110 is used to store programs or data. The memory 110 may be, but is not limited to, random access memory (RAM), read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), etc.
[0024] The processor 120 is used to read / write data or programs stored in the memory 110 and to perform corresponding functions.
[0025] The communication module 130 is used to establish a communication connection between the electronic device 100 and other communication terminals through the network, and to send and receive data through the network.
[0026] It should be understood that, Figure 1 The structure shown is only a schematic diagram of the electronic device 100. The electronic device 100 may also include components that are larger than... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown. Figure 1 The components shown can be implemented using hardware, software, or a combination thereof.
[0027] Please refer to Figure 2 , Figure 2 This is a flowchart illustrating a method for generating a vehicle risk field provided in this embodiment. The method includes: S201. Obtain the perception data of each sensing device for the sensing area, and construct a sensing coordinate system corresponding to the sensing area based on the perception data. When collecting and monitoring road condition data, corresponding roadside sensing devices are set up for each road segment to acquire real-time data for the current road segment. These sensing devices can be high-definition cameras, millimeter-wave radar, lidar, etc., installed on the current road segment. Since vehicles also collect environmental information around them while driving, the sensing devices can also include vehicles traveling on the current road segment.
[0028] Roadside sensing devices and vehicles upload the sensing data they collect to a cloud platform or server. The server then constructs a sensing coordinate system for the current road segment based on the sensing data, environmental information, and location information.
[0029] S202. Determine the object to be evaluated in the sensing coordinate system, and construct the observation vector of the object to be evaluated based on the observation data of each sensing device on the object to be evaluated.
[0030] The object to be evaluated can be a vehicle traveling on the current road segment, an obstacle, or other object. This embodiment uses a target vehicle as the object to be evaluated. When the target vehicle enters the current road segment, the sensing devices on that segment observe the target vehicle and acquire observation data related to it. Since the position and angle of each sensing device when observing the target vehicle may be different, the observation data from each sensing device can be fused to generate an observation vector of the vehicle in the sensing coordinate system.
[0031] S203. Obtain the dynamic confidence weight of each observation data, and fuse the dynamic confidence weight and the observation vector to obtain the final state of the object to be evaluated in the perception coordinate system.
[0032] As a vehicle travels on a given road segment, it is affected by various factors, and these factors have varying degrees of impact on the vehicle. For example, different vehicles entering the same road segment from the same location may have different final states, such as their final position and speed, due to the influence of different factors.
[0033] Therefore, a dynamic confidence weight can be assigned to each observation data point. The dynamic confidence weight can represent the degree of influence of the observation data on the vehicle state. Then, based on the observation data and the dynamic confidence weight, the final state of the vehicle in the perception coordinate system can be determined.
[0034] S204. Calculate multiple risk factors of the object to be evaluated in the perception coordinate system based on the final state, superimpose the multiple risk factors to obtain the risk field corresponding to the object to be evaluated, and broadcast it.
[0035] Based on the vehicle's final state, such as its position and speed, potential risk factors can be determined, such as the risk of collision with other vehicles and other risks. Since each risk factor may affect other vehicles, it is necessary to superimpose all the risk factors contained in the vehicle to generate a risk field for the vehicle, and then broadcast it so that other vehicles on the current road segment can perceive the risks in advance and take avoidance measures.
[0036] This embodiment constructs a unified perception coordinate system from the perception data of multiple sensing devices. Then, for each object to be evaluated in the perception coordinate system, the final state of the object to be evaluated is calculated using its corresponding observation data and dynamic confidence weight. The final state is then used to perform a risk assessment, generate a risk field for the object to be evaluated, and broadcast it. This embodiment of the invention can perform risk assessment on all objects within the perception area of the sensing devices, generate a corresponding risk field, and broadcast it, enabling vehicles to take risk avoidance measures in advance, thereby improving the accuracy of risk assessment and driving safety.
[0037] In one embodiment, the sensing device includes a roadside sensing device and a vehicle, and the step of acquiring sensing data for the sensing area by each sensing device includes: The first sensing data collected by the roadside sensing device, the second sensing data collected by the vehicle, and the historical sensing data are acquired respectively. The first acquisition time of the first sensing data, the second acquisition time of the second sensing data, and the third acquisition time of the historical sensing data are respectively acquired. Based on the first acquisition time, the second acquisition time, and the third acquisition time, the first sensing data, the second sensing data, and the historical sensing data are spatiotemporally aligned to obtain the sensing data of each sensing device for the sensing area.
[0038] Because the environmental information of the same road segment is different at different times, such as lane lines, stop lines, traffic signs, surrounding buildings, trees, traffic congestion, construction, etc., the data collected by each sensing device needs to be aligned in time and space when acquiring the sensing data of the current road segment.
[0039] When each sensing device acquires sensing data for the current road segment, it can assign a corresponding timestamp to the sensing data, and then synchronize the timestamps of different sensing data through the NTP or PTP protocol.
[0040] Historical sensing data is primarily used to provide prior knowledge for risk field calculations, such as the high volume of left-turning vehicles during weekday morning rush hours and peak hours. The historical sensing data can be collected over the most recent week or two weeks to ensure reliability.
[0041] This embodiment obtains the data acquisition time corresponding to different sensing data, and then timestamps the different sensing data according to the data acquisition time to achieve time alignment of different sensing data, providing an accurate data foundation for subsequent risk field generation.
[0042] In one implementation, the step of constructing a sensing coordinate system corresponding to the sensing area based on the sensing data includes: The sensing data collected by each of the sensing devices is spatially aligned and stitched together to obtain panoramic data of the sensing area; The coordinate system parameters are determined based on the panoramic data, and the perception coordinate system is constructed based on the coordinate system parameters.
[0043] Since there may be vehicles and obstacles on the current road section, and the sensing range and accuracy of a single sensing device are limited, in order to restore the road scene, the sensing data collected by each sensing device can be converted into the same format, such as image format. Then, the images are converted and stitched together according to the position and angle of different sensing devices to obtain the panoramic data of the current road section.
[0044] Then, coordinate system parameters are determined based on the panoramic image. For example, the center point of the road segment is used as the origin of the coordinate system, and the corresponding coordinate axes are determined to generate a perception coordinate system. Then, the position information of different perception data is determined under the perception coordinate system to achieve spatial alignment of the perception data.
[0045] Please refer to Figure 3 In one embodiment, step S202 includes: step S2021-step S2022.
[0046] S2021. Obtain the sensing data corresponding to the object to be evaluated from the sensing data of each of the sensing devices as the observation data.
[0047] S2022. Calculate the vector corresponding to each observation data in the sensing coordinate system, and fit the vector corresponding to each observation data to obtain the observation vector.
[0048] Perception data can be real-time data collected by sensing devices on the current road segment. Perception data may include data for the entire road segment, while the object to be evaluated, i.e. the target vehicle, is only located at a certain position on the current road segment. Therefore, it is necessary to extract data related to the target vehicle from the perception data as observation data.
[0049] After constructing the perception coordinate system, the observation data of each perception device on the target vehicle can be represented by a vector. For example, the first coordinate of the perception device in the perception coordinate system and the second coordinate of the target vehicle in the perception coordinate system can be obtained, and then a vector can be generated based on the first and second coordinates.
[0050] Since there may be multiple sensing devices, the observation vector of the target vehicle in the sensing coordinate system can be obtained by fitting the vector corresponding to the observation data of each sensing device.
[0051] For any observed physical target O_i in the current road segment, such as a moving target vehicle, maintain a multi-source observation vector V(O_i)={obs_1,obs_2,...,obs_k}, where obs_j is the observation data of the target from the BSM of the j-th sensing device, such as camera CAM_01, radar RAD_02, and vehicle Car_A.
[0052] This embodiment performs vector fusion and other operations on the observation data of the object to be evaluated by each sensing device to obtain the final vector representation of the object to be evaluated in the sensing coordinate system, which provides a data foundation for the subsequent generation of the risk field.
[0053] In one implementation, the step of obtaining the dynamic confidence weight of each of the observation data includes: Obtain the influence factors of the dynamic credibility weight, and assign a corresponding dynamic weight to each influence factor; The dynamic weight is calculated based on the property weight corresponding to the data property of each observed data; the property weight includes source type weight, data quality weight, and historical performance weight. The dynamic credibility weight is calculated based on the influence factor and the dynamic weight.
[0054] For each observation data obs_j, calculate its dynamic confidence weight w_j.
[0055] w_j=α×W_type×β×W_quality×γ×W_consistency×δ×W_history.
[0056] Where α, β, γ, and δ are normalization coefficients.
[0057] W_type represents the source type weight, which is a priori weight based on the sensor's physical characteristics and data type. For example, it is generally believed that the positioning accuracy of LiDAR is higher than that of a camera, and the positioning accuracy of the vehicle's own GNSS broadcast via BSM is higher than that of other targets perceived by its camera. This is a configurable static lookup table.
[0058] W_quality represents the data quality weight, which is a quality indicator extracted directly from the data packet. For example, for a camera, it is the confidence score of target detection; for radar, it is the signal-to-noise ratio (SNR); and for GNSS, it is the differential state and PDOP value.
[0059] W_consistency represents the cross-validation consistency weight, and W_history represents the historical performance weight.
[0060] In one implementation, the step of calculating the dynamic weight based on the property weight corresponding to the data property of each of the observed data includes: Construct an adjacency matrix for every two observations and calculate the Mahalanobis distance between every two observations; The calculation result of the adjacency matrix is determined based on the Mahalanobis distance; Based on all the observation data corresponding to the object to be evaluated, calculate the sum of the calculation results of the adjacency matrix corresponding to the observation data; The cross-validation consistency weight of the observed data is determined based on the sum of the calculation results, and the dynamic weight is calculated based on the cross-validation consistency weight, the source type weight, the data quality weight, and the historical performance weight.
[0061] For the object to be evaluated \(O_i\), all its observed data \(obs_j\) are compared pairwise. If the Mahalanobis distance between two observed data, such as positions, is less than a certain threshold, they are considered to be consistent, and they "vote" for each other to increase the consistency score of the other.
[0062] The specific algorithm is as follows: 1. For any two observations \(obs_a\) and \(obs_b\), calculate their Mahalanobis distance \(D_M(obs_a, obs_b)\) of the state vectors, such as positions and velocities. 2. Construct an adjacency matrix \(M\). If \(D_M < threshold\), then \(M[a, b]=1\), otherwise \(M[a, b]=0\).
[0063] 3. The consistency score \(W_{consistency}(obs_a)\) of each observation is equal to the sum of the \(a\)-th row or column in the matrix \(M\), and then it is normalized. The higher the score, the more "allies" support it.
[0064] In one implementation, the step of fusing the dynamic credibility weight and the observation vector to obtain the final state of the object to be evaluated in the perception coordinate system includes: Determine the observation state corresponding to each observation vector according to the dynamic credibility weight and the observation vector; Obtain the Gaussian distribution corresponding to the observation state, and calculate the state mean and covariance corresponding to the observation state according to the Gaussian distribution; Calculate the final state through Kalman filtering according to the state mean and covariance.
[0065] After calculating the credibility weight \(w_j\) of each observed data \(obs_j\), a weighted Bayesian or Gaussian fusion method is used to estimate the final state \(S_{fused}\) of the object to be evaluated \(O_i\).
[0066] Assume that the state \(s_j\) of each observation vector follows a Gaussian distribution \(N(\mu_j,\Sigma_j)\), where \(\mu_j\) is the state mean (position, velocity), and \(\Sigma_j\) is the covariance matrix. The calculation formulas for the fused state mean \(\mu_{fused}\) and covariance \(U_{fused}\) are:
[0067] This process can be efficiently implemented through a variant of Kalman filtering.
[0068] The fused \(S_{fused}=\{\mu_{fused},\Sigma_{fused}\}\) not only gives the most likely state of the target, but also gives the degree of uncertainty of this state, and the degree of uncertainty is described by the covariance matrix \(\Sigma_{fused}\).
[0069] In one embodiment, the method further includes: The physical area of the current road segment, such as a 100m x 100m intersection, is divided into 0.5m x 0.5m grids, forming a 200x200 risk field grid map R. The initial risk value R(x,y) of each grid (x,y) is 0.
[0070] For each fused object O_i to be evaluated in the scene, calculate the risk potential P_i(x,y) it generates for any grid (x,y) in space.
[0071]
[0072] in, Kinetic energy factor: Where m_i is the estimated mass and v_i is the velocity after fusion. The greater the kinetic energy, the greater the potential danger. TTI_i(x,y) is the estimated arrival time factor: it calculates the estimated time for the object O_i to arrive at the grid (x,y) in its current state. The shorter the time, the higher the risk. It can be represented as exp(-k*TTI). Prob_i(x,y) represents the trajectory coverage probability factor.
[0073] First, using an intent prediction model, such as an LSTM-based trajectory predictor, the input consists of the target's trajectory over the past few seconds and high-precision map information. This predicts multiple possible trajectories of the object to be evaluated within the next T seconds, along with their probabilities {(traj_1,p_1),(traj_2,p_2),...}. Then, the cumulative probability of these trajectories sweeping across a grid (x,y) within T seconds is calculated. The more high-probability trajectories cover a grid, the higher its risk value. This probability is also affected by the covariance Σ_fused of the fused state; the greater the uncertainty, the wider the trajectory bundle divergence and the broader the impact.
[0074] Then, special risk is injected. Combining information such as buildings and green belts in the high-precision map with the real-time location of vehicles, the area of visual obstruction is calculated. A basic "unknown risk" value is assigned to the edge of the obstructed area to remind vehicles to be wary of "ghost peeking out". If the prediction model identifies a high-probability violation, such as an imminent red light violation, it significantly increases the weight coefficient of its KE_i, creating an extremely high risk potential.
[0075] The risk potentials P_i(x,y) generated by all objects O_i to be evaluated are overlaid on a grid map R to obtain the final global risk field: R(x,y)=ΣP_i(x,y)+R_blindspot(x,y). The final R is a two-dimensional or three-dimensional floating-point matrix whose values represent the comprehensive risk index of the physical location. Then, the risk field is broadcast.
[0076] To reduce the amount of broadcast data, the risk field matrix R is compressed efficiently. Image compression algorithms such as JPEG-LS and WebP can be used, or a custom quantization and run-length encoding scheme can be employed. The compressed risk field data is then encapsulated in a custom V2X message, with the message header containing metadata such as timestamp, geographical extent of the risk field, and raster resolution.
[0077] This embodiment constructs a risk field grid map, then calculates the risk potential of the target to be evaluated to generate a risk field and broadcasts it, enabling vehicles to coordinate with each other and not interfere with each other in multi-vehicle interaction scenarios, greatly improving driving safety.
[0078] To perform the corresponding steps in the above embodiments and various possible methods, an implementation of a vehicle risk field generation device is given below. Please refer to [link / reference needed]. Figure 4 , Figure 4 This is a functional block diagram of a vehicle risk field generation device provided in an embodiment of the present invention. It should be noted that the basic principle and technical effects of the vehicle risk field generation device provided in this embodiment are the same as those in the above embodiments. For the sake of brevity, any parts not mentioned in this embodiment can be referred to the corresponding content in the above embodiments. The vehicle risk field generation device 400 includes: The data acquisition module 401 is used to acquire the perception data of each sensing device for the sensing area, and construct a sensing coordinate system corresponding to the sensing area based on the perception data; The vector construction module 402 is used to determine the object to be evaluated in the sensing coordinate system and construct the observation vector of the object to be evaluated based on the observation data of each sensing device on the object to be evaluated. The state calculation module 403 is used to obtain the dynamic confidence weight of each observation data, and fuse the dynamic confidence weight and the observation vector to obtain the final state of the object to be evaluated in the perception coordinate system. The risk field calculation module 404 is used to calculate multiple risk factors of the object to be evaluated in the perception coordinate system based on the final state, superimpose the multiple risk factors to obtain the risk field corresponding to the object to be evaluated, and broadcast it.
[0079] Optionally, the above modules can be stored in the form of software or firmware. Figure 1 The memory shown is either stored in or embedded in the operating system (OS) of the vehicle risk field generation device, and can be generated by... Figure 1 The processor executes the commands. Meanwhile, the data and program code required to execute these modules can be stored in memory.
[0080] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus embodiments described above are merely illustrative; for example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.
[0081] In addition, the functional modules in the various embodiments of the present invention can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.
[0082] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0083] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A vehicle risk field generation method characterized by, The method comprises the following steps: acquiring perception data of each perception device for a perception area, and constructing a perception coordinate system corresponding to the perception area according to the perception data; determining an object to be evaluated in the perception coordinate system, and constructing an observation vector of the object to be evaluated according to observation data of each perception device on the object to be evaluated; acquiring a dynamic confidence weight of each observation data, and fusing the dynamic confidence weight and the observation vector to obtain a final state of the object to be evaluated in the perception coordinate system; calculating a plurality of risk factors of the object to be evaluated in the perception coordinate system according to the final state, superimposing a plurality of the risk factors to obtain a risk field corresponding to the object to be evaluated, and broadcasting the risk field.
2. The vehicle risk field generation method according to claim 1, characterized by, The perception device comprises a road-end perception device and a vehicle, and the step of acquiring perception data of each perception device for a perception area comprises: acquiring first perception data collected by the road-end perception device, second perception data collected by the vehicle, and historical perception data; acquiring a first collection time of the first perception data, a second collection time of the second perception data, and a third collection time of the historical perception data; performing space-time alignment processing on the first perception data, the second perception data, and the historical perception data according to the first collection time, the second collection time, and the third collection time to obtain the perception data of each perception device for the perception area.
3. The vehicle risk field generation method according to claim 1, characterized by, The step of constructing a perception coordinate system corresponding to the perception area according to the perception data comprises: performing spatial alignment and splicing on the perception data collected by each perception device to obtain panoramic data of the perception area; determining coordinate system parameters according to the panoramic data, and constructing the perception coordinate system according to the coordinate system parameters.
4. The vehicle risk field generation method of claim 1, wherein, The step of constructing an observation vector of the object to be evaluated according to observation data of each perception device on the object to be evaluated comprises: acquiring perception data corresponding to the object to be evaluated from the perception data of each perception device as the observation data; calculating a vector corresponding to each observation data in the perception coordinate system, and fitting the vector corresponding to each observation data to obtain the observation vector.
5. The vehicle risk field generation method of claim 1, wherein, The step of acquiring a dynamic confidence weight of each observation data comprises: acquiring influence factors of the dynamic confidence weight, and assigning a corresponding dynamic weight to each influence factor; calculating the dynamic weight according to a property weight corresponding to a data property of each observation data; the property weight comprises a source type weight, a data quality weight, and a historical performance weight; calculating the dynamic confidence weight according to the influence factors and the dynamic weight.
6. The vehicle risk field generation method of claim 5, wherein, The step of calculating the dynamic weight according to a property weight corresponding to a data property of each observation data comprises: constructing an adjacency matrix according to each two observation data, and calculating Mahalanobis distances between each two observation data; determining a calculation result of the adjacency matrix according to the Mahalanobis distances. According to all the observation data corresponding to the to-be-evaluated object, a sum of calculation results of an adjacency matrix corresponding to the observation data is calculated; According to the sum of the calculation results, a cross-validation consistency weight of the observation data is determined, and the dynamic weight is calculated according to the cross-validation consistency weight, the source type weight, the data quality weight and the historical performance weight.
7. The vehicle risk field generation method of claim 1, wherein, The step of fusing the dynamic credibility weight and the observation vector to obtain the final state of the to-be-evaluated object in the perception coordinate system comprises: According to the dynamic credibility weight and the observation vector, an observation state corresponding to each observation vector is determined; A Gaussian distribution corresponding to the observation state is obtained, and a state mean value and a covariance corresponding to the observation state are calculated according to the Gaussian distribution; The final state is calculated according to the state mean value and the covariance through Kalman filtering.
8. A vehicle risk field generating device characterized by comprising: Comprise: A data acquisition module is configured to acquire perception data of each perception device for a perception region, and construct a perception coordinate system corresponding to the perception region according to the perception data; A vector construction module is configured to determine a to-be-evaluated object in the perception coordinate system, and construct an observation vector of the to-be-evaluated object according to observation data of the to-be-evaluated object observed by each perception device; A state calculation module is configured to acquire a dynamic credibility weight of each observation data, and fuse the dynamic credibility weight and the observation vector to obtain a final state of the to-be-evaluated object in the perception coordinate system; A risk field calculation module is configured to calculate a plurality of risk factors of the to-be-evaluated object in the perception coordinate system according to the final state, superimpose the plurality of risk factors to obtain a risk field corresponding to the to-be-evaluated object, and broadcast the risk field.
9. An electronic device, comprising: The processor and the memory, the memory stores machine executable instructions that can be executed by the processor, and the processor can execute the machine executable instructions to implement the vehicle risk field generation method in any one of claims 1-7.
10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the vehicle risk field generation method in any one of claims 1-7. The computer program is executed by the processor to implement the vehicle risk field generation method in any one of claims 1-7.