Control risk assessment method for vehicle remote control driving, medium and product

Abstract

The invention provides a control risk assessment method for vehicle remote control driving, a medium and a product, which are applied to a remote driving platform, and the method comprises the following steps: obtaining a control quality assessment index of a vehicle, the control quality evaluation index is calculated at least based on communication process related data between the vehicle and the remote driving platform; and inputting the control quality evaluation index into a control risk evaluation model, and obtaining a control risk evaluation result for the vehicle output by the control risk evaluation model.

Description

Technical Field

[0001] This specification relates to the field of remote driving technology, and in particular to a control risk assessment method, medium, and product for remote control driving of a vehicle. Background Art

[0002] As a key branch of intelligent driving, remote driving technology leverages 5G communication technology to achieve driver-vehicle separation. This allows remote drivers to remotely control vehicles through a remote driving platform, demonstrating significant application value in scenarios such as smart transportation, industrial automation, and emergency rescue. In addition to the vehicle's inherent collision risk, remote drivers also face operational risks, distinct from these collision risks. The very act of remote driving itself impacts vehicle safety. Therefore, effective, accurate, and standardized assessments of operational risks associated with remote vehicle driving have become a pressing challenge for the industry.

[0003] In related technologies, the usual control risk assessment mechanism only considers traditional basic sensor data such as vehicle speed, rotation speed, throttle opening, brake opening, etc., which can only reflect limited remote driving conditions and remote driving behaviors, resulting in low accuracy in the assessment of control risks. Summary of the Invention

[0004] In view of this, this specification provides a control risk assessment method, medium and product for vehicle remote control driving to address the deficiencies in related technologies.

[0005] Specifically, this specification is implemented through the following technical solutions:

[0006] According to a first aspect of this specification, a method for assessing control risk of remote vehicle driving is provided, which is applied to a remote driving platform. The method includes:

[0007] Obtaining a handling quality evaluation index of the vehicle, the handling quality evaluation index being calculated based on at least data related to a communication process between the vehicle and the remote driving platform;

[0008] The control quality evaluation index is input into a control risk evaluation model, and a control risk evaluation result for the vehicle output by the control risk evaluation model is obtained.

[0009] According to a second aspect of this specification, a computer-readable storage medium is provided, on which a computer program is stored, and when the program is executed by a processor, the steps of the method described in the first aspect are implemented.

[0010] According to a third aspect of this specification, a computer program product includes a computer program / instruction, which implements the steps of the method described in the first aspect when executed by a processor.

[0011] The remote driving platform in this application overcomes the limitations of relying solely on basic sensor data by introducing control quality evaluation indicators calculated based on data related to the communication process between itself and the vehicle, and combining it with traditional vehicle sensor data to build a multi-dimensional evaluation system. It can more comprehensively and accurately capture the unique risk factors caused by communication process-related data such as communication delays, signal interference, and human-computer interaction anomalies during remote driving, thereby providing a standardized and normalized evaluation framework for remote driving safety control, and significantly improving the accuracy of the assessment of control risks for remote remote control driving of vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] To more clearly illustrate the technical solutions of this specification, the following briefly introduces the drawings required for use in the embodiments or prior art descriptions. Obviously, the drawings described below are merely some embodiments of this specification, and those skilled in the art can derive other drawings based on these drawings without inventive effort.

[0013] Figure 1 This is a schematic diagram of the architecture of a control risk assessment system for remote control driving of a vehicle, as shown in the embodiments disclosed in this specification;

[0014] Figure 2 This is a flowchart of a method for assessing control risk of remote control driving of a vehicle, as shown in the embodiments disclosed in this specification;

[0015] Figure 3 This is a schematic diagram of another method for assessing control risk of remote vehicle driving according to an embodiment disclosed in this specification;

[0016] Figure 4 is a schematic structural diagram of an electronic device shown in an embodiment of this specification;

[0017] Figure 5 This is a block diagram of a control risk assessment device for remote control driving of a vehicle shown in an embodiment of this specification. DETAILED DESCRIPTION

[0018] Exemplary embodiments are described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all possible embodiments consistent with this specification. Rather, they are merely examples of apparatuses and methods consistent with certain aspects of this specification.

[0019] The terms used in this specification are for the purpose of describing specific embodiments only and are not intended to limit this specification. As used in this specification and the appended claims, the singular forms "a," "an," "the," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0020] It should be understood that although the terms first, second, third, etc. may be used in this specification to describe various information, such information should not be limited to these terms. These terms are merely used to distinguish information of the same type from one another. For example, first information may also be referred to as second information, and similarly, second information may also be referred to as first information without departing from the scope of this specification. Depending on the context, the term "if" as used herein may be interpreted as "when," "when," or "in response to determining."

[0021] An embodiment of the method for assessing control risk of vehicle remote control driving according to the present specification is described in detail below with reference to the accompanying drawings.

[0022] Figure 1 This is a schematic diagram of the architecture of a control risk assessment system for remote control driving of a vehicle, as shown in the embodiment disclosed in this specification. Figure 1 As shown, the system may include a vehicle 11 and a remote driving platform 12 .

[0023] Vehicle 11, the remote control executor, can transmit data to the remote driving platform 12 using technologies such as 5G or vehicle-to-everything (V2X) wireless communication. It also receives information from the platform, including steering commands, throttle control signals, and risk assessment feedback parameters. Furthermore, vehicle 11 can be further adapted for various scenarios, including passenger vehicles, commercial vehicles, and special operation vehicles, and this specification does not limit this.

[0024] The remote driving platform 12 serves as the core interactive interface of the remote driver. It can be based on a cloud server or a local server of the simulated cockpit, as well as related software and hardware to form a complete platform. In addition to helping the remote driver remotely control the vehicle 11, it can also calculate and obtain the vehicle's control quality evaluation index based on the relevant data of the communication process between itself and the vehicle 11, and then determine the control risk assessment results of the above-mentioned control quality assessment indicators according to the control risk assessment model. At the same time, the remote driving platform also supports the control of the remote driving status of the corresponding vehicle according to the control risk assessment results, thereby forming a closed loop for handling control risks. Of course, the remote driving platform can present the above-mentioned core interactive interface through devices such as simulated cockpits, virtual reality glasses, mobile phones, tablets, etc., and this specification does not limit this.

[0025] Figure 2 This is a flowchart of a method for assessing control risk in remote vehicle driving, provided by an exemplary embodiment; the method is applied to a remote driving platform and may include the following steps:

[0026] Step S202: Acquire a control quality evaluation index of the vehicle, where the control quality evaluation index is calculated based on at least data related to the communication process between the vehicle and the remote driving platform.

[0027] In the case where the remote driver provides a remote driving platform to remotely drive the vehicle, the above-mentioned remote driving platform can calculate a control quality evaluation index based on at least relevant data of the communication process between the vehicle and the above-mentioned remote driving platform, so as to be used in determining the control risk assessment result in subsequent steps.

[0028] In this specification, the method for obtaining the above-mentioned control quality evaluation index may vary according to the specific content of the communication process-related data or other data.

[0029] In one embodiment, when the communication process-related data includes monitoring data reported by the vehicle, the calculation of the control quality evaluation index can be combined with the dynamic analysis of the reception status of the monitoring data. Specifically, a differentiated index extraction strategy is adopted for different types of monitoring data:

[0030] First, when the monitoring data includes data packets periodically reported by vehicle sensors such as steering angle sensors, wheel speed sensors, and inertial measurement units, the reception frequency and / or packet loss rate of the data packets can be calculated by comparing the continuity of the data packet sequence numbers and the time stamp intervals. The reception frequency can be calculated based on the ratio of the number of reported data packets received within a fixed time period to the duration of the fixed time period. The packet loss rate, used to quantify the loss of data integrity in the communication link, can be calculated by counting the ratio of lost data packets to the total number of data packets sent. These parameters, alone or in combination, can constitute a control quality assessment indicator reflecting the reliability of data packets during transmission.

[0031] Secondly, when the above-mentioned monitoring data includes a video stream transmitted back in real time by a vehicle environment perception camera, the picture clarity, picture refresh rate, picture delay time and picture reception completeness can be calculated by decoding timestamps and analyzing picture metadata. Among them, the above-mentioned picture clarity can be calculated by the ratio of the picture resolution to the preset bit rate, or the structural similarity index (SSIM) algorithm or edge sharpness analysis is used to perform image quality detection on the received video frames, and the brightness, contrast and structural features of the local area of ​​the image are calculated by comparing them with the original captured picture to quantify the degree of picture blur, noise interference or compression distortion. It can also be obtained based on the weighted calculation result of the above multiple calculation results; the above-mentioned picture refresh rate can be obtained by counting the number of complete frames actually received per unit time based on the timestamp interval of consecutive frames in the video stream. The above-mentioned image delay time can be obtained by embedding the acquisition timestamp on the vehicle camera end, and then recording the decoding display timestamp after the remote driving platform receives the image. The difference between the two can be calculated to obtain the end-to-end transmission delay. Furthermore, the delay fluctuation range of multiple consecutive frames can be statistically analyzed through a sliding window algorithm to identify sudden delay peaks; the above-mentioned image reception completeness can be calculated based on the ratio of the number of images received per unit time to the expected number of frames received. For example, if 30 frames should be received in theory but only 25 frames are actually received, the above-mentioned image reception completeness is 83.3%. Of course, based on the characteristics of video encoding, the sequence integrity of key frames and predicted frames can be detected to determine whether the video stream is faulted due to packet loss or network jitter. For example, in the H.264 or H.265 encoding format, the above-mentioned image reception completeness can be calculated based on the ratio of consecutively lost predicted frames (P frames) or key frames (I frames) to normal frames.

[0032] Those skilled in the art will appreciate that the camera image of the second aspect may also be used as at least a part of the sensor data packet of the first aspect to calculate the control quality evaluation index.

[0033] In another embodiment, when the control quality evaluation index is also calculated based on the control data issued to the vehicle, the control quality evaluation index can be calculated according to the execution status of the vehicle with respect to the control data.

[0034] Specifically, the remote driving platform can obtain the first time duration consumed by the above-mentioned vehicle to parse the above-mentioned control data and the second time duration consumed to respond to the control instruction corresponding to the control data; and finally use the total time duration of the above-mentioned first time duration and the above-mentioned second time duration as the above-mentioned control quality evaluation indicator. Among them, the execution status of the above-mentioned control data can be specifically divided into a first time duration and a second time duration, among which the first time duration is the time interval for the remote driving platform to send data such as steering angle instructions, throttle opening instruction control, etc. to the vehicle and complete the instruction parsing (that is, the vehicle's on-board control unit decodes the data packet into an executable instruction), hereinafter also referred to as the parsing delay, which can be achieved by embedding the platform-side sending timestamp T1 in the control data packet, and the vehicle-side recording the local timestamp T2 after completing the instruction parsing, and finally calculating the parsing delay by the time difference T2-T1. The second duration is the time interval between the vehicle completing command parsing and the time when the wire-controlled actuators, such as the steering motor and brake hydraulic unit, actually reach the command target state. This is also referred to as the response delay. The vehicle can trigger execution status monitoring after command parsing is complete, record the timestamp T3, and use the actuator's built-in sensor to provide real-time feedback on the matching result between the actual state and the target state. When the error enters the preset error threshold range, the completion timestamp T4 is recorded. The difference, T4-T3, serves as the response delay. The total duration, as the sum of the first duration and the second duration, reflects the end-to-end timeliness from the issuance of the control command to the actual execution of the vehicle.

[0035] The above embodiment further refines the calculation method of the control quality evaluation index based on the execution status of the control data. By disassembling the delay structure of the vehicle-side command processing and execution links, an end-to-end response timeliness evaluation system is constructed, thereby avoiding the limitation of only focusing on the one-way network delay.

[0036] In another embodiment, when the handling quality evaluation index is also calculated based on environmental data reported by the vehicle, the environmental data can be input into an environmental analysis model, and the environmental index output by the environmental analysis model can be determined as the handling quality evaluation index. The environmental analysis model can be an artificial intelligence model built based on a machine learning algorithm; the environmental data can essentially include data collected by sensors such as on-board millimeter-wave radars, as well as camera images captured by the aforementioned cameras. While the two types of data contain the same content, their purposes and the processes involved in calculating the corresponding handling quality evaluation index are different.

[0037] Specifically, after inputting the aforementioned environmental data into the aforementioned environmental analysis model, the vehicle's surrounding visibility information, road information, weather information, pedestrian information, and / or obstacle information output by the environmental analysis model can be used as the aforementioned control quality assessment indicators. Visibility information can be calculated based on onboard cameras, lidar point cloud data, and ambient light sensors, using fog concentration detection algorithms such as image contrast analysis and dark channel priors, and point cloud density statistics to determine effective viewing distance and ambient light level. The aforementioned road information can further include information such as road curvature, slope angle, and road adhesion coefficient. This information can be obtained by integrating high-precision map data with real-time positioning information such as GPS or RTK through road curvature calculation, slope detection, and road surface type identification. The aforementioned weather information can further include rainfall level, crosswind intensity, and icing risk index. This information can be determined by integrating onboard meteorological sensors such as rain sensors and temperature sensors with cloud-based weather API data through precipitation intensity classification, wind speed vector analysis, and temperature-humidity correlation models. Pedestrian and obstacle information can further include dynamic motion vectors such as target density, speed, and direction, as well as static distance. The difference between the two is that pedestrians, as "Vulnerable Road Users" (VRUs), have a higher computational weight for the control quality assessment indicator. For the environmental analysis model, the control quality assessment indicators in this embodiment can be normalized to a risk contribution value ranging from 0 to 1, and then weighted summed using a scenario weight matrix to generate a comprehensive environmental risk score. For example, in urban road scenarios, pedestrian density accounts for 40% and visibility accounts for 30%. In highway scenarios, crosswind intensity accounts for 50% and obstacle distance accounts for 30%.

[0038] The above embodiment further refines the control quality assessment mechanism driven by environmental data, converts the multi-dimensional environmental information around the vehicle into quantifiable risk characteristics through the environmental analysis model, and realizes the dynamic impact assessment of environmental complexity on remote control safety.

[0039] Step S204 : inputting the control quality assessment index into a control risk assessment model, and obtaining a control risk assessment result for the vehicle output by the control risk assessment model.

[0040] After determining the handling quality assessment index, it can be input into a handling risk assessment model, which then outputs a handling risk assessment result for the vehicle. Depending on the handling quality assessment index, the risk assessment model can be a deterministic rule-based model based on expert experience or preset mathematical relationships, or an artificial intelligence model based on a machine learning algorithm.

[0041] In one embodiment, the aforementioned handling risk assessment model can be a static weighted model. This model manually defines risk event types, such as sudden acceleration and line-crossing frequency, and their associated risk weights. Discrete handling quality assessment indicators are then input into a weighted scoring function to calculate the handling risk value. While this model relies on domain expertise to solidify the scoring logic, it is suitable for scenarios lacking accumulated time-series data. However, it struggles to capture the continuous dynamic correlations within driving behavior.

[0042] In another embodiment, the handling risk assessment model can be a recurrent neural network (RNN) model, which can perform normalization processing on the handling quality assessment indicators to obtain corresponding feature vector sequences, and input the feature vector sequences into the recurrent neural network model to obtain a handling risk score and / or a handling risk result used to characterize whether the vehicle has handling risk. The so-called normalization processing includes but is not limited to 1. Data alignment of different dimensional parameters in the handling quality assessment indicators, that is, using Min-Max normalization or Z-score normalization algorithms to map all indicators to a unified numerical range, such as a floating-point value in the range of 0-1, thereby eliminating the interference of dimensional differences on model training; 2. Slicing the continuously collected normalized indicators based on a preset time window to generate a feature vector sequence containing time series correlation. In addition, linear interpolation or forward filling algorithms can be used to supplement local data missing in the above feature vectors due to communication interruption, or sliding window mean filtering can be used to suppress sudden changes in the handling quality assessment indicators caused by sensor noise or network jitter, thereby retaining their trend change characteristics. For the above-mentioned recurrent neural network model, it takes into account the factors of continuous changes in time, and uses a large amount of data to train the model to eliminate the shortcomings of the above-mentioned static weight model. By using the above-mentioned feature vector sequence as the input vector, it efficiently outputs the control risk score or control risk result at the current time point as the above-mentioned control risk assessment result, effectively balancing the negative impact of delays on the vehicle network in the vehicle remote control driving scenario.

[0043] It is understandable that the above-mentioned handling risk assessment model can collect driving feature vector sequences corresponding to low handling risks and high handling risks as a training set, and manually mark the risk scores as the model output handling risk values. For example, the handling risk value range is 0-100 points, with 100 points being the highest risk driving. The above-mentioned handling risk assessment model can be obtained by training with this training set. Finally, it is deployed on the remote driving platform. The remote driver only needs to input the above-mentioned feature vector sequence to obtain the current handling risk value. Of course, if the output layer of the handling risk assessment model uses, for example, the sigmoid function of a binary classifier to convert the abnormal score between 0 and 1 as a handling risk result, 1 can indicate that the vehicle has a handling risk, and 0 indicates that the vehicle is normal.

[0044] Before determining the above-mentioned control risk assessment result, this manual can also determine at least one countermeasure for the above-mentioned vehicle control hidden danger when the above-mentioned control quality assessment index indicates that the above-mentioned vehicle has vehicle control hidden danger, and issue a response instruction for each countermeasure to the vehicle, wherein the vehicle control hidden danger is different from the control risk of remote control driving of the vehicle. The former refers to the potential hidden danger brought about by any control quality assessment index or part of the control quality assessment indicators meeting the preset control hidden danger conditions, and whether the corresponding potential hidden danger occurs or not also needs to consider factors such as other control quality assessment indicators; the latter refers to the trustworthy result obtained by the control risk assessment model based on the analysis of each input control quality assessment indicator.

[0045] For example, if the communication process-related data includes two control quality assessment indicators: the frequency of data packet reception and the clarity of the camera image, assuming that the reception frequency is lower than the minimum reception frequency set for the control risk condition, such as 10 Hz for three consecutive sampling periods, but the image clarity is higher than the minimum clarity set for the control risk condition, such as SSIM ≥ 0.6, then the preset correspondence between the control risk condition and the corresponding control risk content can be triggered to determine that the vehicle has a "communication bandwidth bottleneck risk." At this time, based on the preset correspondence between the control risk content and the countermeasure, at least one countermeasure for the vehicle control risk can be determined, such as dynamically reducing the video encoding resolution from 1080P to 720P and adjusting the I-frame interval to increase the reception frequency to 15Hz; or prioritizing the transmission of control command data packets at the communication protocol layer; and simultaneously initiating pre-rendering of the 3D environment on the remote driving platform to generate virtual perspective compensation visual information based on historical positioning data. By issuing the corresponding countermeasures, the image SSIM value can be stabilized above 0.7, eliminating the risk of command transmission delay caused by bandwidth competition before triggering a high-risk alarm, thereby achieving a dynamic balance between communication resource optimization and driving safety.

[0046] In addition to the aforementioned handling quality assessment indicators, basic sensor data directly derived from the vehicle's sensors, such as vehicle speed, rotational speed, throttle position, and brake position, can also be obtained and fed into the handling risk assessment model along with the aforementioned handling quality assessment indicators. The handling risk assessment model then outputs additional handling risk assessment results for the vehicle. The input parameters of the handling risk assessment model have been further expanded. While the existing handling quality assessment indicators reflect remote control stability, the model also analyzes the characteristics of the basic sensor data. For example, if the basic sensor data indicates a brake position of 80% and the vehicle speed decrease slope is below a predetermined threshold, the handling risk assessment model will still output a result indicating handling risk, even if the handling quality assessment indicator calculated based on the monitoring data reception status is normal. This dual data verification mechanism for basic sensor data and handling quality assessment indicators expands the risk assessment dimension of the handling risk assessment model from single-channel reliability to multi-dimensional collaborative state analysis, significantly improving the accuracy of handling risk assessments in complex scenarios such as mechanical failures, execution deviations, and sudden environmental disturbances.

[0047] Figure 3 This is a schematic diagram of another method for assessing the control risk of remote control driving of a vehicle according to the embodiment disclosed in this specification. Figure 3 As shown, the method includes the following steps:

[0048] In step S302 , the vehicle collects multi-source data and sends it to the remote driving platform.

[0049] In one embodiment, the vehicle can transmit sensor data packets including steering angle and wheel speed, camera video streams, control command response status including parsing latency and execution latency, and environmental perception data including lidar point clouds and weather information to the remote driving platform 12 in real time via a 5G communication link. Assuming the vehicle is driving in rainy or foggy weather, with the camera transmitting 25 frames per second of 720P footage, the SSIM is 0.75, the sensor data packet reception frequency is 12Hz, the total control command duration is 180ms, and the environmental data feedback visibility is 50 meters.

[0050] In step S304, the remote driving platform calculates a control quality evaluation index.

[0051] In one embodiment, the remote driving platform 12 can calculate the control quality evaluation index according to the classification of the received data type: for example, for sensor data packets, the receiving frequency of 12Hz and the packet loss rate of 2% can be calculated; for camera images, the image delay of 200ms, the refresh rate of 20fps and the clarity SSIM=0.75 can be calculated; for control instructions, the delay of 80ms and the execution delay of 100ms can be calculated as a total duration of 180ms; for environmental data, it can be input into the environmental analysis model to obtain the model output results of medium visibility level, medium rainfall intensity and low pedestrian density.

[0052] In step S306 , the remote driving platform generates a feature vector sequence according to the control quality evaluation index.

[0053] In one embodiment, the remote driving platform can standardize control quality assessment indicators: for example, parameters such as SSIM = 0.75, receiving frequency 12Hz, total duration 180ms, and visibility level (assuming a mapping of 0.5) are normalized to values ​​in the range of 0-1 using the Min-Max algorithm. A sliding window, such as a 10-second window, is then used to generate a sequence of feature vectors containing temporal correlations. To address temporary loss of sensor data packets due to rain or fog, a forward fill algorithm is used to fill in the missing values. A sliding mean filter is used to suppress fluctuations in image delay caused by communication jitter, for example, by correcting fluctuations of 200ms ± 50ms to a smooth curve. Finally, a feature vector sequence containing current and historical 5-second data is generated, which serves as the input to the RNN model in the next step.

[0054] Step S308: RNN model dynamic risk assessment and result output.

[0055] In one embodiment, the remote driving platform can input the feature vector sequence into a pre-trained recurrent neural network (RNN) model. The model can extract time series features based on a historical driving behavior dataset and then output a control risk score for the current time point. For example, the model recognizes the combined effect of a 200ms screen delay and a total duration of 180ms, and combines the environmental risks of medium visibility and medium rainfall intensity to output a comprehensive risk score of 65 points. Assuming a full score of 100 points, the corresponding risk level is medium. At the same time, the control risk assessment model predicts that the probability of lane departure risk caused by accumulated delays within the next 3 seconds is 35% based on the time series features, and outputs the corresponding control risk assessment result to indicate the presence of a medium-level control risk, which can trigger a corresponding level of control risk warning.

[0056] In step S310 , the remote driving platform performs processing according to the control risk.

[0057] In one embodiment, the remote driving platform can initiate a hierarchical response strategy based on the medium-risk output of the RNN model. For example, the following strategies may be implemented: 1. Adjusting communication priority, increasing the control command transmission bandwidth to 70%, and dynamically reducing the camera resolution from 720P to 480P to increase the SSIM to 0.85 and reduce image latency to 150ms; 2. Sending a pre-deceleration command to the vehicle, limiting the speed to no more than 40km / h; 3. Initiating virtual environment pre-rendering, generating a 3D scene model based on historical lidar point clouds, and using VR glasses to assist the driver in predicting obstacles in rain and fog. After executing these steps, the remote driving platform can monitor the new control quality assessment indicators, forming a closed-loop control system. Assuming that the reception frequency is increased to 15Hz and the total duration is reduced to 140ms, the verification risk score output after re-input into the RNN model will drop to 45 points.

[0058] In step S312, the remote driving platform optimizes the RNN model.

[0059] In one embodiment, the remote driving platform can store the feature vector sequence, risk score, and treatment effect data from the entire risk assessment and treatment process in a training set and update the RNN model parameters through incremental learning. For example, if the model learns that the "resolution reduction + speed control" combination strategy is effective in resolving maneuvering risks in rainy and foggy environments, it can prioritize this strategy when outputting maneuvering risk assessment results in similar scenarios in the future, improving the real-time nature of remote driver risk assessment and treatment.

[0060] Figure 4 This is a schematic structural diagram of an electronic device in an exemplary embodiment. Figure 4 At the hardware level, the electronic device includes a processor 402, an internal bus 410, a network interface 404, a memory 406, and a non-volatile memory 408, and may also include other necessary hardware. The processor reads the corresponding computer program from the non-volatile memory into the internal memory and then runs it, forming a risk code detection device at the logical level. Of course, in addition to software implementation, this specification does not exclude other implementation methods, such as logic devices or a combination of software and hardware. In other words, the execution of the following processing flow is not limited to individual logic units and can also be hardware or logic devices.

[0061] Figure 5 The embodiment of this specification shows a block diagram of a control risk assessment device for remote control driving of a vehicle. Figure 5 As shown, the device is applied to a remote driving platform and may include:

[0062] An evaluation index acquisition unit 502 is configured to acquire a handling quality evaluation index of the vehicle, wherein the handling quality evaluation index is calculated based on at least data related to the communication process between the vehicle and the remote driving platform;

[0063] The evaluation result acquisition unit 504 is configured to input the handling quality evaluation index into a handling risk evaluation model, and acquire a handling risk evaluation result for the vehicle output by the handling risk evaluation model.

[0064] Optionally, the device further includes:

[0065] A first index calculation unit calculates the control quality evaluation index according to a reception status of the monitoring data when the communication process related data includes the monitoring data reported by the vehicle.

[0066] Optionally, the first indicator calculation unit is specifically configured to:

[0067] When the monitoring data includes sensor data packets, calculating the reception frequency and / or packet loss rate of the data packets as the control quality evaluation index;

[0068] In a case where the monitoring data includes the camera image of the vehicle, the image clarity, image refresh rate, image delay time and / or image reception completeness of the camera image are calculated as the control quality evaluation index.

[0069] Optionally, the device further includes:

[0070] a second index calculation unit, configured to calculate the handling quality evaluation index according to an execution state of the vehicle with respect to the control data, when the handling quality evaluation index is also calculated based on the control data issued to the vehicle;

[0071] The third indicator calculation unit is used to input the environmental data into an environmental analysis model when the handling quality evaluation indicator is also calculated based on the environmental data reported by the vehicle, and determine the environmental indicator output by the environmental analysis model as the handling quality evaluation indicator.

[0072] Optionally, the second index calculation unit is specifically used to: obtain a first time duration consumed by the vehicle for parsing the control data and a second time duration consumed for responding to a control instruction corresponding to the control data; and use the total time duration of the first time duration and the second time duration as the control quality evaluation index; and / or,

[0073] The third indicator calculation unit is specifically used to: determine the environmental indicator output by the environmental analysis model as the control quality evaluation indicator, including: inputting the environmental data into the environmental analysis model, and using the visibility information, road information, weather information, pedestrian information and / or obstacle information around the vehicle output by the environmental analysis model as the control quality evaluation indicator.

[0074] Optionally, the evaluation result obtaining unit 504 is specifically configured to:

[0075] performing a normalization process on the control quality evaluation index to obtain a corresponding feature vector sequence;

[0076] The feature vector sequence is input into a recurrent neural network model to obtain a handling risk score and / or a handling risk result for characterizing whether the vehicle has a handling risk.

[0077] Optionally, the device further includes:

[0078] a handling risk response unit, configured to determine at least one response measure for the vehicle handling risk when the handling quality evaluation index indicates that the vehicle has a vehicle handling risk, and to issue a response instruction for each response measure to the vehicle;

[0079] Optionally, the device further includes:

[0080] A comprehensive evaluation result acquisition unit 504 is configured to acquire basic sensor data of the vehicle, wherein the basic sensor data is directly obtained based on sensors of the vehicle;

[0081] The basic sensor data and the control quality assessment index are input into the control risk assessment model, and other control risk assessment results for the vehicle output by the control risk assessment model are obtained.

[0082] The implementation process of the functions and effects of each unit in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, and will not be repeated here.

[0083] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial description of the method embodiments. The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this specification. A person of ordinary skill in the art can understand and implement it without paying any creative work.

[0084] Based on the same concept as the above method, this specification also provides an electronic device, including: a processor; a memory for storing processor-executable instructions; wherein the processor implements the steps of the method described in any of the above embodiments by running the executable instructions.

[0085] Based on the same concept as the above method, this specification also provides a computer-readable storage medium on which computer instructions are stored. When the instructions are executed by a processor, the steps of the method described in any of the above embodiments are implemented.

[0086] Based on the same concept as the above method, this specification also provides a computer program product, including a computer program / instruction, which implements the steps of the method described in any of the above embodiments when executed by a processor.

[0087] Embodiments of the subject matter and functional operations described in this specification may be implemented in the following: digital electronic circuits, tangibly embodied computer software or firmware, computer hardware including the structures disclosed in this specification and their structural equivalents, or a combination of one or more of them. Embodiments of the subject matter described in this specification may be implemented as one or more computer programs, i.e., one or more modules of computer program instructions encoded on a tangible, non-transitory program carrier to be executed by a data processing device or to control the operation of the data processing device. Alternatively or additionally, the program instructions may be encoded on an artificially generated propagation signal, such as a machine-generated electrical, optical, or electromagnetic signal, which is generated to encode information and transmit it to a suitable receiver device for execution by the data processing device. The computer storage medium may be a machine-readable storage device, a machine-readable storage substrate, a random or serial access memory device, or a combination of one or more of them.

[0088] The processes and logic flows described in this specification can be performed by one or more programmable computers executing one or more computer programs to perform the corresponding functions by operating on input data and generating output. The processes and logic flows can also be performed by, and apparatus can be implemented as, special purpose logic circuitry, such as an FPGA (field programmable gate array) or an ASIC (application-specific integrated circuit).

[0089] Computers suitable for executing computer programs include, for example, general-purpose and / or special-purpose microprocessors, or any other type of central processing unit. Typically, the central processing unit will receive instructions and data from a read-only memory and / or random access memory. The basic components of a computer include a central processing unit for implementing or executing instructions and one or more memory devices for storing instructions and data. Typically, a computer will also include one or more mass storage devices for storing data, such as magnetic disks, magneto-optical disks, or optical disks, or the computer will be operably coupled to such mass storage devices to receive data from them or to transmit data to them, or both. However, a computer does not necessarily have such devices. In addition, a computer can be embedded in another device, such as a mobile phone, a personal digital assistant (PDA), a mobile audio or video player, a game console, a global positioning system (GPS) receiver, or a portable storage device such as a universal serial bus (USB) flash drive, to name a few.

[0090] Computer-readable media suitable for storing computer program instructions and data include all forms of non-volatile memory, media, and storage devices, including, for example, semiconductor memory devices (e.g., EPROM, EEPROM, and flash memory devices), magnetic disks (e.g., internal hard disks or removable disks), magneto-optical disks, and CD ROM and DVD-ROM disks. The processor and memory can be supplemented by, or incorporated in, special purpose logic circuitry.

[0091] Although this specification includes many specific implementation details, these should not be interpreted as limiting the scope of any invention or the scope of protection claimed, but are mainly used to describe the features of specific embodiments of specific inventions. Certain features described in multiple embodiments within this specification may also be implemented in combination in a single embodiment. On the other hand, the various features described in a single embodiment may also be implemented separately in multiple embodiments or in any suitable sub-combination. In addition, although features may work in certain combinations as described above and even initially claimed as such, one or more features from the claimed combination may be removed from the combination in some cases, and the claimed combination may point to a sub-combination or a variation of the sub-combination.

[0092] Similarly, although operations are depicted in a particular order in the accompanying drawings, this should not be understood as requiring that these operations be performed in the particular order shown or performed sequentially, or that all illustrated operations be performed to achieve the desired results. In some cases, multitasking and parallel processing may be advantageous. In addition, the separation of various system modules and components in the above-described embodiments should not be understood as requiring such separation in all embodiments, and it should be understood that the described program components and systems can generally be integrated together in a single software product, or packaged into multiple software products.

[0093] Thus, specific embodiments of the subject matter have been described. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the particular order shown or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing may be advantageous.

[0094] The above description is only a preferred embodiment of this specification and is not intended to limit this specification. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of this specification should be included in the scope of protection of this specification.

Claims

1. A method for assessing control risk of remote vehicle driving, characterized in that: Applied to a remote driving platform, the remote driving platform is used to remotely drive a vehicle, the method comprising: Obtaining a handling quality evaluation index of the vehicle, the handling quality evaluation index being calculated based on at least data related to a communication process between the vehicle and the remote driving platform; The control quality evaluation index is input into a control risk evaluation model, and a control risk evaluation result for the vehicle output by the control risk evaluation model is obtained.

2. The method according to claim 1, characterized in that The obtaining of the vehicle handling quality evaluation index includes: In a case where the communication process-related data includes monitoring data reported by the vehicle, the control quality evaluation index is calculated according to a reception status of the monitoring data.

3. The method according to claim 2, characterized in that The calculating the control quality evaluation index according to the reception status of the monitoring data includes: When the monitoring data includes sensor data packets, calculating the reception frequency and / or packet loss rate of the data packets as the control quality evaluation index; In a case where the monitoring data includes the camera image of the vehicle, the image clarity, image refresh rate, image delay time and / or image reception completeness of the camera image are calculated as the control quality evaluation index.

4. The method according to claim 2, characterized in that The obtaining of the vehicle handling quality evaluation index includes: When the control quality evaluation index is also calculated based on the control data issued to the vehicle, the control quality evaluation index is calculated according to the execution state of the vehicle with respect to the control data; In a case where the handling quality evaluation index is also calculated based on environmental data reported by the vehicle, the environmental data is input into an environmental analysis model, and the environmental index output by the environmental analysis model is determined as the handling quality evaluation index.

5. The method according to claim 4, characterized in that Calculating the control quality evaluation index based on the vehicle's execution status of the control data includes: obtaining a first time duration consumed by the vehicle for parsing the control data and a second time duration consumed for responding to a control instruction corresponding to the control data; using the total of the first time duration and the second time duration as the control quality evaluation index; and / or, The method of determining the environmental indicators output by the environmental analysis model as the control quality evaluation indicators includes: inputting the environmental data into the environmental analysis model, and using the visibility information, road information, weather information, pedestrian information and / or obstacle information around the vehicle output by the environmental analysis model as the control quality evaluation indicators.

6. The method according to claim 1, characterized in that Inputting the handling quality assessment index into a handling risk assessment model and obtaining a handling risk assessment result for the vehicle output by the handling risk assessment model includes: performing a normalization process on the control quality evaluation index to obtain a corresponding feature vector sequence; The feature vector sequence is input into a recurrent neural network model to obtain a handling risk score and / or a handling risk result for characterizing whether the vehicle has a handling risk.

7. The method according to claim 1, characterized in that The method further comprises: When the handling quality evaluation index indicates that the vehicle has a vehicle handling risk, at least one countermeasure for the vehicle handling risk is determined, and a response instruction for each countermeasure is issued to the vehicle.

8. The method according to claim 1, characterized in that The method further comprises: Acquiring basic sensor data of the vehicle, where the basic sensor data is directly obtained based on sensors of the vehicle; The basic sensor data and the control quality assessment index are input into the control risk assessment model, and other control risk assessment results for the vehicle output by the control risk assessment model are obtained.

9. A computer-readable storage medium, characterized in that Computer instructions are stored thereon, and when the instructions are executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

10. A computer program product, characterized in that The method comprises a computer program / instruction, which implements the steps of the method according to any one of claims 1 to 8 when the computer program / instruction is executed by a processor.

Images