Cloud supported vehicle active safety control method and system

Abstract

The application provides a cloud-supported vehicle active safety control method and system, the method comprising: receiving a cloud risk control strategy issued by the cloud, the cloud risk control strategy being obtained by the cloud based on risk analysis results of fusion perception data uploaded to a cloud control basic platform and combined with a vehicle driving automation level to generate a differentiated control strategy, the fusion perception data being obtained by the cloud control basic platform based on fusion of vehicle perception data and roadside device perception data uploaded to an edge cloud; generating a vehicle control instruction based on the cloud risk control strategy, or based on the cloud risk control strategy and a local risk control decision, according to the vehicle driving automation level; the local risk control decision being generated based on the vehicle perception data when the vehicle driving automation level is greater than or equal to a preset level. The application can improve the comprehensiveness and accuracy of environmental perception, avoid visual range obstruction, and realize fine and differentiated risk control.

Description

Technical Field

[0001] This invention relates to the field of vehicle control technology, and in particular to a cloud-supported active safety control method and system for vehicles. Background Technology

[0002] With the rapid development of environmental perception and sensing technologies, the level of vehicle intelligence has significantly improved. Real-time perception of the surrounding environment using various high-precision sensors has become the foundation for realizing advanced driver assistance functions. To ensure traffic safety, cloud-supported active safety control technologies for vehicles have emerged and are widely used in various driving scenarios. These technologies aim to effectively reduce the incidence of traffic accidents by monitoring road conditions in real time and proactively controlling and avoiding hazards in critical moments.

[0003] Currently, cloud-supported active safety control solutions based on in-vehicle sensor networks are widely adopted. These solutions rely on various environmental perception sensors on the vehicle itself, such as millimeter-wave radar, lidar, and in-vehicle cameras, to collect real-time data on the driving environment around the vehicle. This includes the vehicle's motion status, such as speed, acceleration, and steering angle, as well as information on the relative distance and speed of surrounding obstacles. The collected multi-source sensor data is then fused to construct a local environmental model of the vehicle's surroundings. Based on a built-in preset algorithm model, potential collision risks are assessed. Once a rear-end collision or collision risk is determined, an audible and visual alarm is immediately triggered, or the vehicle's braking system is directly intervened to execute active collision avoidance control.

[0004] However, the aforementioned control scheme is primarily limited to perception and control from a single vehicle's intelligent perspective. Due to the physical blind spots of onboard sensors and their susceptibility to adverse weather conditions and complex road obstructions, the perception range and accuracy are limited. Furthermore, this scheme lacks beyond-line-of-sight perception capabilities and a global collaborative decision-making mechanism, making it unable to anticipate hidden dangers beyond the line-of-sight range and hindering interaction and collaboration with other traffic participants. In complex traffic scenarios such as sudden highway accidents or intersection obstructions, relying solely on the vehicle's own perception often leads to untimely warnings and passive control strategies, thus limiting further improvements in the vehicle's active safety performance and failing to meet the safety requirements of future high-level intelligent driving. Summary of the Invention

[0005] This invention provides a cloud-supported vehicle active safety control method and system to address the shortcomings of existing technologies that rely solely on the vehicle's own perception, resulting in untimely and inaccurate warnings. It can improve the comprehensiveness and accuracy of environmental perception, avoid visual obstruction, achieve refined and differentiated risk control, and enhance the applicability of the strategy.

[0006] This invention provides a cloud-supported active safety control method for vehicles, comprising: receiving a cloud-based risk control strategy issued by the cloud, wherein the cloud-based risk control strategy is obtained by combining risk analysis results obtained from fused perception data uploaded by the cloud control platform with the vehicle's driving automation level to generate a differentiated control strategy; the fused perception data is obtained by fusing vehicle perception data uploaded by the cloud control platform from the edge cloud and roadside device perception data; generating vehicle control commands based on the vehicle's driving automation level and the cloud-based risk control strategy, or based on the cloud-based risk control strategy and local risk control decisions; the local risk control decisions are generated based on vehicle perception data collected by the vehicle itself when the vehicle's driving automation level is greater than or equal to a preset level.

[0007] According to the present invention, a cloud-supported active safety control method for vehicles includes a cloud-based risk control strategy comprising risk analysis results and a cloud-based control strategy. The risk analysis results are obtained by performing spatiotemporal trajectory extrapolation, collision risk assessment, and risk point prediction on the fused perception data uploaded by the cloud control platform based on vehicle identification. The cloud-based control strategy is obtained by making active safety decisions based on deterministic risks in the risk analysis results and making early warning decisions based on non-deterministic risks in the risk analysis results when the cloud determines that the vehicle identification corresponds to a driving automation level greater than or equal to L3 autonomous driving assistance level.

[0008] According to the present invention, a cloud-supported active safety control method for vehicles generates vehicle control commands based on the vehicle's driving automation level and a cloud-based risk control strategy, or based on a cloud-based risk control strategy and a local risk control decision. The method includes: performing a safety risk analysis based on perception data collected by the vehicle, given that the vehicle's driving automation level is greater than or equal to L3 (Level 3) of automated driving assistance, and generating a local risk control decision; based on the safety risk analysis results, when a safety risk is determined to exist, comparing the values ​​of corresponding preset control indicators in the local risk control decision and the cloud-based risk control strategy, selecting the value with the higher safety coefficient, and combining it with the local risk control decision to generate vehicle control commands.

[0009] According to the present invention, a cloud-supported active safety control method for vehicles generates vehicle control commands based on a cloud-based risk control strategy, or based on a cloud-based risk control strategy and local risk control decisions, according to the vehicle's driving automation level. The method further includes: generating vehicle control commands based on a cloud-based risk control strategy when it is determined that no safety risk exists based on the results of a safety risk analysis.

[0010] According to the present invention, a cloud-supported active safety control method for vehicles performs spatiotemporal trajectory extrapolation, collision risk assessment, and risk point prediction in the cloud. The method includes: parsing fused perception data uploaded by the cloud control platform based on previously received vehicle identifiers; extracting motion state vectors of the main vehicle and surrounding targets corresponding to the vehicle identifiers; predicting the trajectories of the main vehicle and motor vehicle targets; and performing multimodal trajectory prediction for pedestrians and non-motorized vehicles to obtain corresponding trajectory prediction results; performing a collision risk assessment based on the trajectory prediction results of the main vehicle and the surrounding targets to obtain a collision risk assessment result; and determining trajectory segments where the collision risk exceeds a preset risk threshold and extracting the center coordinate sequence based on the collision risk assessment result. The center coordinate sequence is then mapped to a high-precision map road network topology to determine the specific lane marker location and corresponding road curvature features where the risk occurs, thus obtaining a risk point prediction result.

[0011] According to the present invention, a cloud-supported active safety control method for vehicles makes active safety decisions based on deterministic risks in risk analysis results. The method includes: obtaining the collision probability, collision time, collision location, and collision severity corresponding to the deterministic risks based on the risk analysis results; determining the minimum safe deceleration and maximum lateral offset required to eliminate the collision risk based on the current driving state of the main vehicle; constructing a cloud-based virtual control constraint set based on the minimum safe deceleration and maximum lateral offset, combined with the dynamic constraints of the main vehicle and the road adhesion coefficient; and performing multi-objective trajectory planning within the cloud-based virtual control constraint set, with the optimization objectives of minimizing driving trajectory conflicts and minimizing passenger impact, to obtain a globally desired trajectory containing a target speed sequence and a target path curvature sequence, and generating a corresponding cloud-based control strategy.

[0012] According to the present invention, a cloud-supported active safety control method for vehicles makes early warning decisions based on uncertain risks identified in risk analysis results. The method includes: analyzing the risk source characteristics corresponding to the uncertain risks based on the risk analysis results, extracting the risk source type, relative distance, and speed change trends, and identifying blind spots or traffic conflict areas where the risk sources are located by combining high-precision map road network topology data; determining the early warning trigger time and early warning coverage area for the uncertain risks based on the risk source type and the location information of the blind spots or traffic conflict areas; and matching the corresponding early warning level and early warning prompting method based on the early warning trigger time, early warning coverage area, and the current driving status of the vehicle, thereby generating a corresponding cloud control strategy.

[0013] According to a cloud-supported active safety control method for vehicles provided by the present invention, the cloud-based risk control strategy includes risk analysis results and cloud-based control strategies. The risk analysis results are obtained by the cloud based on vehicle identification and spatiotemporal trajectory extrapolation, collision risk assessment, and risk point prediction of fused perception data uploaded by the cloud control platform. The cloud-based control strategy is obtained by making active safety decisions based on deterministic risks in the risk analysis results when the cloud determines that the vehicle identification corresponds to a vehicle driving automation level of L2 assisted driving level; or, the cloud-based control strategy is obtained by making early warning decisions based on uncertain risks in the risk analysis results when the cloud determines that the vehicle identification corresponds to a vehicle driving automation level of less than L2 automated assisted driving level.

[0014] According to the present invention, a cloud-supported active safety control method for vehicles generates vehicle control commands based on the vehicle's driving automation level and a cloud-based risk control strategy, or based on a cloud-based risk control strategy and a local risk control decision. The method further includes: based on the vehicle's driving automation level being L2 assisted driving level, performing safety risk analysis based on perception data collected by the vehicle, and generating a local risk control decision; determining, based on the local risk control decision and the cloud-based risk control strategy, that the cloud-based collision risk is greater than the vehicle-side collision risk, and that the difference between the cloud-based collision risk and the vehicle-side collision risk exceeds a preset threshold, determining that a perception blind spot exists, and generating vehicle control commands based on the cloud-based risk control strategy.

[0015] According to the present invention, a cloud-supported active safety control method for vehicles generates vehicle control commands based on the vehicle's driving automation level and a cloud-based risk control strategy, or based on a cloud-based risk control strategy and local risk control decisions. The method further includes: if the vehicle's driving automation level is less than L2 autonomous driving assistance level, parsing the warning information in the cloud-based risk control strategy, and generating vehicle control commands based on the warning information to control the vehicle to alarm and prompt the driver to take over manually.

[0016] This invention also provides a cloud-supported active safety control system for vehicles, comprising: a strategy receiving module, which receives a cloud-based risk control strategy issued by the cloud, wherein the cloud-based risk control strategy is obtained by combining risk analysis results obtained from fused perception data uploaded by the cloud control platform with the vehicle's driving automation level to generate differentiated control strategies; the fused perception data is obtained by fusing vehicle perception data uploaded by the cloud control platform based on edge cloud and roadside device perception data; and a safety control module, which generates vehicle control commands based on the vehicle's driving automation level and the cloud-based risk control strategy, or based on the cloud-based risk control strategy and local risk control decisions; the local risk control decisions are generated based on vehicle perception data collected by the vehicle itself when the vehicle's driving automation level is greater than or equal to a preset level.

[0017] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the cloud-supported active safety control method for vehicles as described above.

[0018] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the cloud-supported active safety control method for vehicles as described above.

[0019] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the cloud-supported active safety control method for vehicles as described above.

[0020] The cloud-supported active safety control method and system for vehicles provided by this invention introduces perception data from roadside equipment to obtain perspectives that cannot be observed by the vehicle's own sensors, improving the comprehensiveness and accuracy of environmental perception and avoiding line-of-sight obstruction. This allows the cloud to perform in-depth analysis based on the fusion of vehicle perception data and roadside equipment perception data, not only identifying current risks but also predicting potential risks in long-distance or complex traffic flows, such as congestion or accidents several kilometers ahead. This enhances the ability to analyze macro-level traffic risks and potential hazards. Furthermore, based on the vehicle's level of driving automation, differentiated control strategies are generated, providing assistance to vehicles with weaker capabilities and granting higher-level control to vehicles with stronger capabilities. This ensures that the control strategy matches the actual performance of the vehicle, achieving refined and differentiated risk control, improving the applicability of the strategy, and facilitating dynamic adjustments by the vehicle based on the risk control strategy issued from the cloud and the vehicle's level of intelligence, ensuring system reliability under various levels of automation and network environments. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0022] Figure 1 This is one of the flowcharts of the cloud-supported active safety control method for vehicles provided by the present invention; Figure 2 This is the second flowchart of the cloud-supported active safety control method for vehicles provided by the present invention; Figure 3 This is a schematic diagram of the structure of the cloud-supported vehicle active safety control system provided by the present invention; Figure 4This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0024] Figure 1 This is a flowchart illustrating the cloud-supported active safety control method for vehicles provided by the present invention, as shown below. Figure 1 As shown, the method includes the following: S11 receives cloud-based risk control strategies from the cloud. The cloud-based risk control strategies are derived from risk analysis results obtained by combining the vehicle's driving automation level with the vehicle's risk analysis results obtained by the cloud-based control platform based on the fusion perception data uploaded by the cloud control platform. The fusion perception data is obtained by the cloud control platform based on the vehicle perception data and roadside equipment perception data uploaded by the edge cloud. S12, based on the vehicle's driving automation level and cloud-based risk control strategy, or based on cloud-based risk control strategy and local risk control decision, generate vehicle control commands; local risk control decision is generated based on vehicle perception data collected by the vehicle when the vehicle's driving automation level is greater than or equal to a preset level.

[0025] It should be noted that this method is executed on the vehicle side, as detailed below. Figures 2-3 The present invention describes a cloud-supported active safety control method for vehicles.

[0026] Step S11: Receive the cloud-based risk control strategy sent from the cloud. The cloud-based risk control strategy is obtained by combining the risk analysis results obtained from the fusion perception data uploaded by the cloud control platform with the vehicle driving automation level to generate differentiated control strategies. The fusion perception data is obtained by the cloud control platform by fusing vehicle perception data and roadside equipment perception data uploaded by the edge cloud.

[0027] It should be added that the vehicles and roadside equipment will send the collected perception data to the edge cloud. The vehicle perception data includes the vehicle's own data and the data of each target in the perception field of view around the vehicle. The data includes positioning information, speed information and acceleration information. The targets include motor vehicles, non-motor vehicles, pedestrians and / or other obstacles, etc., which are determined according to the actual road conditions involved. The roadside equipment perception data includes the vehicle identification, speed, acceleration, heading angle, vehicle type, lateral and longitudinal relative position of the target relative to the main vehicle, and lane position of each target in the perception field of view of the roadside equipment.

[0028] Accordingly, the edge cloud sends the received vehicle perception data and roadside device perception data to the cloud control platform for fusion processing. Specifically, this includes: determining the corresponding vehicle position and driving direction based on the vehicle perception data; matching roadside devices within the vehicle's field of vision, roadside devices outside the vehicle's field of vision and at a first preset distance from the vehicle along the driving direction, and roadside devices outside the vehicle's field of vision and at a second preset distance from the vehicle along the reverse driving direction, based on the vehicle position, and obtaining the corresponding roadside device perception data, where the first preset range is larger than the second preset range; and registering the vehicle perception data and the corresponding roadside device perception data to the global coordinate system for time synchronization and spatial alignment to obtain fused perception data.

[0029] Furthermore, the cloud-based control platform can also: traverse all targets in the fused perception data, determine the positional relationship and motion trend of each target relative to the main vehicle, divide the targets into related targets within the vehicle's field of vision and environmental targets outside the field of vision, and mark them with corresponding association identifiers; based on the association identifiers, the fused perception data is structured into a standardized data structure containing vehicle-related data and global environmental data. Vehicle-related data is used to represent trajectory information that directly affects the current vehicle, and global environmental data is used to represent the surrounding traffic situation, so that the cloud-based control application platform can directly perform trajectory extrapolation based on the standardized data structure. Through closed-loop control of hierarchical fusion and decision control, it can achieve efficient integration and collaborative utilization of multi-source data, as well as real-time interaction and dynamic adjustment of commands, effectively improving the accuracy and real-time performance of active safety control, thereby better addressing safety risks in complex traffic scenarios.

[0030] In addition, the differentiated control strategy includes: the cloud identifies the vehicle's risk perception and control capabilities based on the vehicle's level of driving automation; for vehicles below Level 2 assisted driving, such as traditional vehicles equipped only with connected devices, the cloud generates warning information for uncertain risks; for Level 2 assisted driving autonomous vehicles, the cloud generates proactive safety decision information for deterministic risks; for Level 3 assisted driving and above, the cloud generates comprehensive control information that includes both deterministic risk decisions and uncertain risk warnings. The cloud-based risk control strategy includes suggested speed, suggested deceleration, the distance between the risk vehicle identifier and the relative driver vehicle, and warning levels, for the vehicle to conduct safety arbitration.

[0031] Accordingly, the cloud-based risk control strategy includes risk analysis results and cloud-based control strategies. The risk analysis results are obtained by the cloud based on vehicle identification and spatiotemporal trajectory extrapolation, collision risk assessment, and risk point prediction of the fused perception data uploaded by the cloud control platform. The cloud-based control strategy is obtained by the cloud making proactive safety decisions based on deterministic risks in the risk analysis results and making early warning decisions based on uncertain risks in the risk analysis results when the cloud determines that the vehicle identification corresponds to a vehicle with a driving automation level greater than or equal to L3 autonomous driving assistance level.

[0032] Furthermore, deterministic risk is used to characterize risks for which the probability, timing, location, and severity of a collision can be determined, while nondeterministic risk is used to characterize risks that are difficult to predict accurately based on the random behavior of traffic participants.

[0033] It should be noted that the cloud-based system extrapolates the spatiotemporal trajectory of vehicles to identify potential collision risks and risk points in advance, improving the accuracy and foresight of risk prediction. For vehicles with a driving automation level of L3 or higher, it performs proactive safety and early warning decisions. This proactive safety decision-making overcomes the shortcomings of traditional early warning systems that rely on driver reaction speed. By quantifying specific collision parameters such as time, location, and severity, it precisely triggers control measures to minimize accident losses. Furthermore, early warning decisions prevent misjudgments or accidents caused by random behaviors that are difficult to quantify, such as pedestrians suddenly crossing the road, vehicles illegally changing lanes, or sudden deceleration. This addresses the control challenges caused by unmodelable random behaviors, expanding the system's safety coverage. Through the fusion and collaborative control of multi-source data from the vehicle, road, and cloud, the system fully leverages the empowering potential of the integrated vehicle-road-cloud system for proactive safety control. By utilizing the complementary advantages of cloud-based beyond-line-of-sight risk perception and vehicle-side close-range safety monitoring, it comprehensively enhances the effectiveness of proactive safety control for connected vehicles.

[0034] In addition, the cloud control strategy is derived by making proactive safety decisions based on the deterministic risks in the risk analysis results when the cloud determines that the vehicle identification corresponds to the L2 assisted driving level. This fully considers the longitudinal and lateral control capabilities already possessed by L2 level vehicles, enabling L2 level vehicles to automatically avoid risks at critical moments, rather than just reminding the driver. This allows the vehicle to cope with dangerous scenarios that exceed its own perception capabilities, achieving a leapfrog improvement in safety capabilities.

[0035] Furthermore, vehicles below Level 2 typically lack the ability to directly control braking or steering via external cloud commands. If an active safety decision is forcibly issued, the vehicle may be unable to execute it or malfunction due to system compatibility issues. Therefore, the cloud control strategy is derived from the cloud's determination that the vehicle's driving automation level is lower than Level 2 autonomous driving assistance level, based on the uncertain risks identified in the risk analysis results. This strategy aims to overcome the problem that low-end vehicles lack the necessary actuators or control interfaces, preventing cloud commands from being implemented and avoiding safety hazards caused by underlying control failures.

[0036] It should be noted that by providing targeted control and monitoring for different vehicle classes, the active safety performance of vehicles of different classes can be improved while meeting diverse safety control needs. Specifically, the cloud-based system performs spatiotemporal trajectory extrapolation, collision risk assessment, and risk point prediction. This includes: parsing the fusion perception data uploaded by the cloud control platform based on previously received vehicle identifiers, extracting the motion state vectors of the main vehicle and surrounding targets corresponding to the vehicle identifiers, and predicting the trajectories of the main vehicle and motor vehicle targets, as well as performing multimodal trajectory prediction for pedestrians and non-motorized vehicle targets, to obtain the corresponding trajectory prediction results; conducting a collision risk assessment based on the trajectory prediction results of the main vehicle and the surrounding targets, to obtain the collision risk assessment results; and identifying trajectory segments where the collision risk exceeds a preset risk threshold and extracting the center coordinate sequence based on the collision risk assessment results. The center coordinate sequence is then mapped to the high-precision map road network topology to determine the specific lane marker location and corresponding road curvature features where the risk occurs, thus obtaining the risk point prediction results.

[0037] It should be noted that motor vehicles generally follow traffic rules and their trajectories are relatively regular, so conventional prediction is sufficient. However, the behavior of pedestrians and non-motorized vehicles is highly random and multimodal. Therefore, the cloud performs trajectory prediction separately for motor vehicles, non-motorized vehicles, and pedestrians to overcome the limitation of not being able to adapt to both regular (motor vehicles) and random (pedestrian / non-motorized vehicle) movements simultaneously. This ensures coverage of more types of potential behaviors, avoids omissions or misjudgments of the behavioral evolution of vulnerable traffic participants, improves the accuracy of trajectory prediction in complex traffic scenarios, and further conducts collision risk assessment based on the corresponding trajectory prediction results to achieve dynamic collision detection from a global perspective, eliminate the hidden dangers of perception blind spots, and filter trajectory segments based on preset risk thresholds for risk point prediction. This overcomes the resource waste and false alarm interference caused by responding to all potential anomalies, achieves key focus on risk locations, and optimizes the utilization rate of cloud resources.

[0038] It should be added that trajectory prediction can be obtained according to the actual design method. For example, the trajectory prediction of the main vehicle and motor vehicle targets can be achieved by the constant speed and velocity model (CTRV), and the multimodal trajectory prediction of pedestrians and non-motorized vehicle targets can be achieved by the interactive multi-model algorithm (IMM). No further restrictions are made here, so as to obtain the predicted trajectory and confidence level of the corresponding main vehicle and target. The risk point prediction results include the location information and risk level of the risk point.

[0039] In addition, a collision risk assessment is performed based on the trajectory prediction results of the main vehicle and the trajectory prediction results of surrounding targets to obtain the collision risk assessment results, including: when overlapping trajectories are determined based on the trajectory prediction results of the main vehicle and the trajectory prediction results of surrounding targets, the collision time and collision location are determined based on the overlapping trajectories; the collision severity is assessed based on the relative speed between the main vehicle and the target corresponding to the overlapping trajectories; the collision risk is determined based on the confidence level of the corresponding trajectory prediction results and the collision time; and the collision risk assessment results are obtained based on the collision time, collision location, collision severity, and collision risk.

[0040] Furthermore, the collision risk can be obtained by weighted summation based on the ratio of 1 to the collision time, combined with the corresponding confidence level.

[0041] In addition, the cloud-based system makes proactive safety decisions based on the deterministic risks identified in the risk analysis results. This includes: obtaining the collision probability, collision time, collision location, and collision severity corresponding to the deterministic risks based on the risk analysis results, and determining the minimum safe deceleration and maximum lateral offset required to eliminate the collision risk based on the current driving status of the main vehicle; constructing a cloud-based virtual control constraint set based on the minimum safe deceleration and maximum lateral offset, combined with the dynamic constraints of the main vehicle and the road adhesion coefficient; and performing multi-objective trajectory planning within the cloud-based virtual control constraint set, with the optimization objectives of minimizing driving trajectory conflicts and minimizing passenger impact, to obtain a global expected trajectory containing the target speed sequence and the target path curvature sequence, and generating a corresponding cloud-based control strategy.

[0042] It should be noted that the cloud-based system quantifies specific collision parameters to precisely calculate the deceleration and offset required to avoid a collision. This means that the vehicle will neither collide with obstacles due to insufficient deceleration nor cause unnecessary energy waste or rear-end collision risks due to excessive braking. This overcomes the problem of traditional emergency braking strategies relying solely on fixed thresholds or simple distance judgments, which can lead to insufficient braking or excessive braking. It achieves precise calculation of the force of avoidance actions and, by introducing dynamic constraints and road adhesion coefficients, ensures that the strategies issued by the cloud are within the vehicle's capabilities and conform to the current physical characteristics of the road surface. This prevents the vehicle from losing control due to performing impossible tasks. The system then performs trajectory planning to ensure that the planned trajectory smoothly changes speed and direction while avoiding collisions, making the vehicle's movements more stable and human-like, and reducing the risk of secondary injuries caused by sudden braking and sharp turns.

[0043] It should be added that the collision probability can be determined based on the confidence level of the trajectory prediction results. Additionally, generating the corresponding cloud control strategy includes: generating corresponding cloud-based vertical and horizontal control commands based on the globally expected trajectory, serving as the corresponding cloud control strategy; or, performing a consistency check between the globally expected trajectory and the local planning intent uploaded from the edge cloud. If the check passes, then generating corresponding cloud-based vertical and horizontal control commands based on the globally expected trajectory, serving as the corresponding cloud control strategy.

[0044] Furthermore, the local planning intent is generated by the edge cloud identifying the micro-traffic flow characteristics around the vehicle based on vehicle perception data and roadside equipment perception data, predicting the short-term motion trend of surrounding targets, and calculating the longitudinal acceleration constraints and lateral displacement constraints required for the vehicle to avoid collision within a preset time window based on the short-term motion trend, thus generating a local planning intent containing local dynamic boundaries.

[0045] In addition, consistency verification is performed between the global expected trajectory and the local planning intent uploaded from the edge cloud. This includes checking the deviation parameters between the global expected trajectory and the local planning intent in the spatiotemporal dimensions, including velocity deviation, acceleration deviation, and path curvature deviation. If the deviation parameters exceed the corresponding preset safety threshold and the local planning intent contains an emergency avoidance indicator, the cloud determines that the global planning has failed, switches the control weight to the edge cloud, and generates a longitudinal emergency control command based on the local planning intent. If the deviation parameters do not exceed the corresponding preset safety threshold but exceed the efficiency threshold, the cloud uses the global expected trajectory as a benchmark, substitutes the local planning intent as a state constraint into the quadratic programming equation, performs local correction on the global expected trajectory, and generates a fusion control command. If the correction fails or the deviation parameters continue to diverge, the cloud generates a vehicle takeover command and sends it to the vehicle, and suspends sending cloud control strategies to the vehicle.

[0046] In addition, the cloud-based system makes early warning decisions based on uncertain risks identified in the risk analysis results. This includes: analyzing the risk source characteristics corresponding to the uncertain risks based on the risk analysis results, extracting the risk source type, relative distance, and speed change trends, and combining this with high-precision map road network topology data to identify blind spots or traffic conflict areas where the risk sources are located; determining the early warning trigger time and early warning coverage area based on the risk source type and the location information of the blind spots or traffic conflict areas; and matching the corresponding early warning level and early warning prompting method based on the early warning trigger time, early warning coverage area, and the current driving status of the main vehicle to generate corresponding cloud-based control strategies.

[0047] It should be noted that the cloud-based system associates risk source characteristics with the road network topology of high-precision maps, such as blind spots at intersections, building obstructions, and roadside vegetation obstructions, to intelligently identify whether the risk source is within the vehicle's blind spot. This allows the cloud to provide early warnings of blind spot risks before the driver or even the vehicle's sensors detect them, achieving deep identification of hidden risks, overcoming the hazards of perception blind spots, and dynamically calculating the alarm trigger time and the length of road segment requiring continuous monitoring based on whether the risk source is a pedestrian or a vehicle, and whether it is in an intersection blind spot or a roadside conflict zone. This ensures that the warning is neither too early, causing the driver to become complacent and ignoring it, nor too late, causing the driver to be unable to react in time, thus optimizing the spatiotemporal accuracy of the warning and reducing unnecessary disturbances.

[0048] In addition, the generated cloud control strategy includes warning information such as sound warning frequency, visual icon color, and vibration intensity. After generating the cloud control strategy, the warning information is mapped into the data format of the vehicle-road-cloud communication protocol to generate warning commands to be sent to the master vehicle.

[0049] Step S12: Based on the vehicle's driving automation level and cloud-based risk control strategy, or based on cloud-based risk control strategy and local risk control decision, generate vehicle control commands; local risk control decision is generated based on vehicle perception data collected by the vehicle when the vehicle's driving automation level is greater than or equal to a preset level.

[0050] In this embodiment, reference Figure 2 Based on the vehicle's driving automation level and cloud-based risk control strategies, or based on cloud-based risk control strategies and local risk control decisions, vehicle control commands are generated. This includes: based on the vehicle's driving automation level being greater than or equal to L3 autonomous driving assistance level, conducting safety risk analysis based on perception data collected by the vehicle and generating local risk control decisions; based on the safety risk analysis results, when a safety risk is determined to exist, comparing the values ​​of the corresponding preset control indicators in the local risk control decisions and cloud-based risk control strategies, selecting the value with the higher safety coefficient, and combining it with the local risk control decisions to generate vehicle control commands.

[0051] It should be noted that for L3 and above autonomous vehicles, the vehicle-side performs local safety risk analysis and generates corresponding risk control decisions to build a dual safety redundancy mechanism, which greatly improves system reliability. When a safety risk is identified at the vehicle-side, the values ​​of the corresponding preset control indicators in the local risk control decision and the cloud-based risk control strategy are compared. The preset control indicators are optimized through an optimization mechanism, and the vehicle is controlled in combination with the local risk control decision to avoid command conflicts. This fully utilizes the super computing power and global perspective of the cloud, while also leveraging the rapid response advantage of local decision-making. This ensures that the vehicle can respond to commands in a timely and reasonable manner when safety risks occur in different ranges, thus guaranteeing the safety of vehicle operation.

[0052] It should be added that the safety factor is used to characterize the vehicle's reserve capacity to ensure it escapes danger or the certainty of danger resolution for the corresponding preset control indicators. For example, if the preset control indicator is deceleration, the greater the deceleration, the higher the safety factor. Specifically, the corresponding safety factor can be configured based on the specific control indicators involved and prior experience; no further limitations are made here. In addition, after collecting perception data, the vehicle-side sensors transmit the corresponding vehicle perception data to the vehicle-side domain controller for local risk assessment and decision-making.

[0053] In addition, based on the vehicle's driving automation level, vehicle control commands are generated based on cloud-based risk control strategies, or based on cloud-based risk control strategies and local risk control decisions. This also includes generating vehicle control commands based on cloud-based risk control strategies when it is determined that there are no safety risks based on the results of safety risk analysis.

[0054] It is worth noting that when the vehicle's driving automation level is lower than the L2 assisted driving level, the driver can determine whether to follow the cloud-based or vehicle-based recommendations based on local risk control decisions and cloud-based risk control strategies, or control the vehicle based on the corresponding risks and the driver's experience, etc., without further limitations.

[0055] In addition, based on the vehicle's driving automation level, and based on cloud-based risk control strategies, or based on cloud-based risk control strategies and local risk control decisions, vehicle control commands are generated. This also includes: based on the vehicle's driving automation level being L2 assisted driving level, performing safety risk analysis based on perception data collected by the vehicle and generating local risk control decisions; based on local risk control decisions and cloud-based risk control strategies, determining that the cloud-based collision risk is greater than the vehicle-side collision risk, and that the difference between the cloud-based collision risk and the vehicle-side collision risk exceeds a preset threshold, determining that a perception blind spot exists, and generating vehicle control commands based on the cloud-based risk control strategy.

[0056] It should be added that, based on the vehicle's driving automation level, and based on cloud-based risk control strategies, or based on cloud-based risk control strategies and local risk control decisions, vehicle control commands are generated. This also includes: determining that the cloud-based collision risk is less than or equal to the vehicle-side collision risk, and generating vehicle control commands based on local risk control decisions.

[0057] In addition, after the vehicle-side determines the cloud-based risk control strategy and / or local risk control decision and generates vehicle control commands, it includes: constructing a vehicle dynamics constraint model based on the current vehicle load and road adhesion coefficient, and inputting the final control strategy determined by the cloud-based risk control strategy and / or local risk control decision into the vehicle dynamics constraint model; if the control parameters in the selected suggested command, such as the target deceleration and target yaw rate, exceed the allowable range of the vehicle dynamics constraint model, the control parameters are trimmed to the boundary values ​​of the allowable range, and the final execution command is generated to control the vehicle to perform deceleration or steering actions.

[0058] Furthermore, based on the vehicle's driving automation level and cloud-based risk control strategies, or a combination of cloud-based risk control strategies and local risk control decisions, vehicle control commands are generated. This also includes: for vehicles with a driving automation level lower than L2 (Level 2) autonomous driving assistance, parsing warning information from the cloud-based risk control strategy and generating vehicle control commands based on this information to trigger vehicle alarms and prompt the driver for manual intervention. It should be added that warnings can be displayed to the driver via a human-machine interface, suggesting vehicle speed and identifying potentially hazardous vehicles, thus achieving human-machine collaboration and allowing the driver to promptly understand the vehicle's safety status and take appropriate action.

[0059] In one optional embodiment, after generating vehicle control commands, the process includes: executing the control commands within a target period and sending vehicle status data back to the cloud to achieve underlying control execution.

[0060] In summary, this invention, by introducing roadside device perception data, facilitates the acquisition of perspectives that vehicle sensors cannot observe, improving the comprehensiveness and accuracy of environmental perception, avoiding line-of-sight occlusion, and enabling the cloud to perform in-depth analysis based on the fusion of vehicle perception data and roadside device perception data. This not only identifies current risks but also predicts potential risks in long-distance or complex traffic flows, such as congestion or accidents several kilometers ahead, enhancing the ability to analyze macro-level traffic risks and potential hazards. Furthermore, based on the vehicle's level of driving automation, differentiated control strategies are generated, providing assistance to vehicles with weaker capabilities and granting higher-level control to vehicles with stronger capabilities. This ensures that the control strategies match the actual performance of the vehicles, achieving refined and differentiated risk control, improving the applicability of the strategies, and facilitating dynamic adjustments by the vehicle based on the risk control strategies issued from the cloud and the vehicle's level of intelligence, ensuring system reliability under various levels of automation and network environments.

[0061] The cloud-supported active safety control system for vehicles provided by the present invention will be described below. The cloud-supported active safety control system for vehicles described below can be referred to in correspondence with the cloud-supported active safety control method for vehicles described above.

[0062] Figure 3 A schematic diagram of a cloud-supported active safety control system for vehicles is shown. The system includes: The strategy receiving module 31 receives the cloud risk control strategy sent from the cloud. The cloud risk control strategy is obtained by combining the risk analysis results obtained from the fusion perception data uploaded by the cloud control basic platform with the vehicle driving automation level to generate a differentiated control strategy. The fusion perception data is obtained by the cloud control basic platform by fusing the vehicle perception data and roadside equipment perception data uploaded by the edge cloud. The safety control module 32 generates vehicle control commands based on the vehicle's driving automation level and cloud-based risk control strategies, or based on cloud-based risk control strategies and local risk control decisions. The local risk control decisions are generated based on vehicle perception data collected by the vehicle itself when the vehicle's driving automation level is greater than or equal to a preset level.

[0063] It should be noted that the specific principles of the embodiments of the present invention are the same as those of the method embodiments described above. For details, please refer to the method embodiments above. More detailed explanations will not be repeated here.

[0064] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4 As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other through the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute a cloud-supported vehicle active safety control method. This method includes: receiving a cloud-based risk control strategy from the cloud, wherein the cloud-based risk control strategy is obtained by combining risk analysis results obtained from fused perception data uploaded by the cloud control platform with the vehicle's driving automation level to generate a differentiated control strategy; the fused perception data is obtained by fusing vehicle perception data uploaded by the cloud control platform from the edge cloud and roadside device perception data; generating vehicle control instructions based on the vehicle's driving automation level and the cloud-based risk control strategy, or based on the cloud-based risk control strategy and local risk control decisions; the local risk control decisions are generated based on vehicle perception data collected by the vehicle itself when the vehicle's driving automation level is greater than or equal to a preset level.

[0065] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part 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 the present 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.

[0066] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the cloud-supported vehicle active safety control method provided by the above methods. The method includes: receiving a cloud-based risk control strategy issued by the cloud, wherein the cloud-based risk control strategy is obtained by combining the risk analysis results obtained by the cloud based on the fusion perception data uploaded by the cloud control platform with the vehicle driving automation level to generate a differentiated control strategy; the fusion perception data is obtained by the cloud control platform based on the vehicle perception data uploaded by the edge cloud and the roadside equipment perception data; generating a vehicle control command based on the vehicle driving automation level and the cloud-based risk control strategy, or based on the cloud-based risk control strategy and the local risk control decision; the local risk control decision is generated based on the vehicle perception data collected by the vehicle when the vehicle driving automation level is greater than or equal to a preset level.

[0067] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a cloud-supported active vehicle safety control method provided by the above methods. This method includes: receiving a cloud-based risk control strategy from the cloud, wherein the cloud-based risk control strategy is obtained by combining risk analysis results obtained from fused perception data uploaded by the cloud control platform with the vehicle's driving automation level to generate a differentiated control strategy; the fused perception data is obtained by fusing vehicle perception data uploaded by the cloud control platform from the edge cloud and roadside device perception data; generating a vehicle control command based on the vehicle's driving automation level and the cloud-based risk control strategy, or based on the cloud-based risk control strategy and a local risk control decision; the local risk control decision is generated based on vehicle perception data collected by the vehicle itself when the vehicle's driving automation level is greater than or equal to a preset level.

[0068] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0069] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0070] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A cloud-supported active safety control method for vehicles, characterized in that, include: The cloud-based risk control strategy is received from the cloud. The cloud-based risk control strategy is obtained by combining the risk analysis results obtained by the cloud-based cloud control platform based on the fusion perception data uploaded by the cloud control platform with the vehicle driving automation level to generate a differentiated control strategy. The fusion perception data is obtained by the cloud control platform based on the vehicle perception data and roadside equipment perception data uploaded by the edge cloud. Based on the vehicle's driving automation level, and based on the cloud-based risk control strategy, or based on the cloud-based risk control strategy and local risk control decisions, vehicle control commands are generated. The local risk control decision is generated based on vehicle perception data collected by the vehicle itself when the vehicle's driving automation level is greater than or equal to a preset level.

2. The cloud-supported active safety control method for vehicles according to claim 1, characterized in that, The cloud-based risk control strategy includes risk analysis results and cloud-based control strategies. The risk analysis results are obtained by the cloud-based system performing spatiotemporal trajectory extrapolation, collision risk assessment, and risk point prediction on the fused perception data uploaded by the cloud control platform based on vehicle identification. The cloud-based control strategy is derived by the cloud determining that the vehicle's driving automation level is greater than or equal to L3 autonomous driving assistance level, making proactive safety decisions based on deterministic risks in the risk analysis results, and making early warning decisions based on non-deterministic risks in the risk analysis results.

3. The cloud-supported active safety control method for vehicles according to claim 2, characterized in that, Based on the vehicle's driving automation level and the cloud-based risk control strategy, or based on the cloud-based risk control strategy and local risk control decisions, vehicle control commands are generated, including: Based on the vehicle's driving automation level being greater than or equal to L3 autonomous driving assistance level, safety risk analysis is performed using perception data collected by the vehicle itself, and local risk control decisions are generated. Based on the security risk analysis results, when a security risk is determined to exist, the values ​​of the corresponding preset control indicators in the local risk control decision and the cloud-based risk control strategy are compared. The value with the higher security coefficient is selected, and combined with the local risk control decision, a vehicle control command is generated.

4. The cloud-supported active safety control method for vehicles according to claim 3, characterized in that, Based on the vehicle's driving automation level, and using the cloud-based risk control strategy, or based on the cloud-based risk control strategy and local risk control decisions, vehicle control commands are generated, further including: If, based on the security risk analysis results, it is determined that there is no security risk, vehicle control commands are generated based on the cloud-based risk control strategy.

5. The cloud-supported active safety control method for vehicles according to claim 2, characterized in that, The cloud-based system performs spatiotemporal trajectory simulation, collision risk assessment, and risk point prediction, including: Based on the previously received vehicle identification, the fused perception data uploaded by the cloud control platform is parsed, the motion state vectors of the main vehicle and surrounding targets corresponding to the vehicle identification are extracted, and trajectory prediction is performed on the main vehicle and motor vehicle targets, as well as multimodal trajectory prediction is performed on pedestrian and non-motor vehicle targets to obtain the corresponding trajectory prediction results. Based on the trajectory prediction results of the main vehicle and the trajectory prediction results of the surrounding targets, a collision risk assessment is performed to obtain a collision risk assessment result. Based on the collision risk assessment results, the trajectory segments with collision risks exceeding the preset risk threshold are identified and their center coordinate sequences are extracted. The center coordinate sequences are then mapped onto the high-precision map road network topology to determine the specific lane station location and corresponding road curvature features where the risk occurs, thus obtaining the risk point prediction results.

6. The cloud-supported active safety control method for vehicles according to claim 2, characterized in that, The cloud platform makes proactive security decisions based on the deterministic risks identified in the risk analysis results, including: Based on the risk analysis results, the collision probability, collision time, collision location and collision severity corresponding to the deterministic risk are obtained, and combined with the current driving state of the main vehicle, the minimum safe deceleration and maximum lateral offset required to eliminate the collision risk are determined. Based on the minimum safe deceleration and the maximum lateral offset, combined with the dynamic constraints of the main vehicle and the road adhesion coefficient, a cloud-based virtual control constraint set is constructed. Within the cloud-based virtual control constraint set, multi-objective trajectory planning is performed with the optimization objectives of minimizing driving trajectory conflicts and minimizing passenger impact. This yields a global desired trajectory containing the target speed sequence and the target path curvature sequence, and generates a corresponding cloud-based control strategy.

7. The cloud-supported active safety control method for vehicles according to claim 2, characterized in that, The cloud platform makes early warning decisions based on the uncertain risks identified in the risk analysis results, including: Based on the risk analysis results, the risk source characteristics corresponding to the uncertain risks are analyzed, the risk source type, relative distance and speed change trend are extracted, and combined with high-precision map road network topology data, the blind spots or traffic conflict areas where the risk sources are located are identified. Based on the type of risk source and the location information of the blind spot or the traffic conflict area, determine the warning trigger time and warning coverage area of ​​the uncertain risk; Based on the warning trigger time, the warning coverage area, and the current driving status of the main vehicle, the corresponding warning level and warning prompt method are matched to generate a corresponding cloud control strategy.

8. The cloud-supported active safety control method for vehicles according to claim 1, characterized in that, The cloud-based risk control strategy includes risk analysis results and cloud-based control strategies. The risk analysis results are obtained by the cloud-based system performing spatiotemporal trajectory extrapolation, collision risk assessment, and risk point prediction on the fused perception data uploaded by the cloud control platform based on vehicle identification. The cloud-based control strategy is obtained by making proactive safety decisions based on the deterministic risks in the risk analysis results when the cloud determines that the vehicle identification corresponds to the L2 assisted driving level of the vehicle. or, The cloud-based control strategy is derived by making a warning decision based on the uncertain risks in the risk analysis results when the cloud determines that the driving automation level of the vehicle corresponding to the vehicle identifier is less than the L2 autonomous driving assistance level.

9. The cloud-supported active safety control method for vehicles according to claim 8, characterized in that, Based on the vehicle's driving automation level, and using the cloud-based risk control strategy, or based on the cloud-based risk control strategy and local risk control decisions, vehicle control commands are generated, further including: Based on the vehicle's driving automation level of L2 assisted driving, safety risk analysis is performed using perception data collected by the vehicle, and local risk control decisions are generated. Based on the local risk control decision and the cloud-based risk control strategy, it is determined that the cloud-based collision risk is greater than the vehicle-side collision risk, and the difference between the cloud-based collision risk and the vehicle-side collision risk exceeds a preset threshold. It is then determined that there is a perception blind spot, and a vehicle control command is generated based on the cloud-based risk control strategy. Based on the vehicle's driving automation level, and using the cloud-based risk control strategy, or based on the cloud-based risk control strategy and local risk control decisions, vehicle control commands are generated, further including: Based on the fact that the vehicle's driving automation level is less than L2 autonomous driving assistance level, the warning information in the cloud-based risk control strategy is analyzed, and vehicle control commands are generated according to the warning information to control the vehicle to alarm and prompt the driver to take over manually.

10. A cloud-supported active safety control system for vehicles, characterized in that, include: The strategy receiving module receives cloud-based risk control strategies sent from the cloud. The cloud-based risk control strategies are obtained by combining the risk analysis results obtained from the fusion perception data uploaded by the cloud control platform with the vehicle driving automation level to generate differentiated control strategies. The fusion perception data is obtained by the cloud control platform by fusing vehicle perception data and roadside device perception data uploaded by the edge cloud. The safety control module generates vehicle control commands based on the vehicle's driving automation level and the cloud-based risk control strategy, or based on the cloud-based risk control strategy and local risk control decisions. The local risk control decision is generated based on vehicle perception data collected by the vehicle itself when the vehicle's driving automation level is greater than or equal to a preset level.

Images