Context-aware parsing and cache updating method for heterogeneous vehicle-side devices

Abstract

This disclosure provides a context-aware parsing and cache update method for heterogeneous vehicle-side devices, including: acquiring vehicle-side cache data and dividing the vehicle-side cache data into scenarios to construct a security event chain; establishing a multi-dimensional value judgment mechanism for the contexts within the security event chain and generating multi-dimensional value results within the chain through the multi-dimensional value judgment mechanism; setting different partitions for the contexts within the same continuous security event chain based on the multi-dimensional value results within the chain to determine the cache occupancy boundary; based on the cache occupancy boundary, keeping the resident contexts continuously available in the cache and prohibiting unlimited addition of cache occupancy to the same security event chain to perform chained cache updates; when the continuous security event chain enters the end state, performing closed-loop verification on each partition to generate context-aware results and cache update results, avoiding long-term accumulation of security-related contexts and ensuring that auxiliary contexts can enter the cache in a timely manner.

Description

Technical Field

[0001] This disclosure relates to the field of artificial intelligence technology, and in particular to a context-aware parsing and cache update method for heterogeneous vehicle-side devices. Background Technology

[0002] In scenarios such as merging at the entrance of an urban elevated highway, sudden deceleration of the vehicle in front, vehicles in adjacent lanes crossing the line, and drivers simultaneously requesting a change of navigation route via voice, emergency braking, lane departure, target vehicle trajectory, navigation replanning, and cockpit prompts may all be marked as high time-sensitive contexts.

[0003] Currently, traditional technical solutions for the above scenarios increase the resident weight of safety scenario contexts, prioritizing their retention in the cache. However, they do not further distinguish the end boundaries between "single safety events" and "continuous safety event chains." If an emergency braking scenario is not yet released, and a subsequent lane departure or navigation replanning scenario is written to the cache, high-value cache blocks will continuously accumulate, easily leading to the cache layer being occupied by safety-related contexts for extended periods. This deficiency is difficult to resolve by increasing cache capacity because on-vehicle high-speed static storage resources are limited, and safety scenarios require deterministic responses, not temporary expansion or post-event compensation. Therefore, traditional technical solutions are prone to the technical problem that while safety contexts are retained, auxiliary contexts such as cockpit prompts, driver takeover alerts, and rendering feedback cannot be cached in a timely manner. Summary of the Invention

[0004] In view of this, in order to solve the problems brought about by the existing technology, this application provides a context-aware parsing and cache update method for heterogeneous vehicle-side devices.

[0005] Firstly, this disclosure provides a context-aware parsing and cache update method for heterogeneous vehicle-side devices, the method comprising: Acquire vehicle-side cached data and divide the vehicle-side cached data into scenarios to construct a security event chain; A multi-dimensional value judgment mechanism is established for the context within the security event chain, and multi-dimensional value results within the chain are generated through the multi-dimensional value judgment mechanism. Based on the multidimensional value results within the chain, different partitions are set for the context within the same chain of consecutive security events to determine the cache occupancy boundary. Based on the cache occupancy boundary, the resident context is kept continuously available in the cache, and unlimited addition of cache occupancy to the same security event chain is prohibited in order to perform chained cache updates; When the continuous security event chain enters the end state, a closed-loop review is performed on each partition to generate context-aware results and cache update results.

[0006] Furthermore, the step of acquiring vehicle-side cached data and segmenting the vehicle-side cached data into scenarios to construct a security event chain includes: The context in the vehicle-side cached data is time-aligned and cached to form an initial context entry set, and the security event scenario is identified through the initial context entry set. Extract the context corresponding to the security event scenario and determine whether the context corresponding to the security event scenario is merged into the same continuous security event chain. If so, merge the context corresponding to the security event scenario into the same continuous security event chain. Based on the result of grouping the contexts corresponding to the security event scenarios into the same continuous security event chain, a continuous security event candidate chain is generated, and the continuous security event candidate chain is filtered to construct a security event chain that meets the set requirements.

[0007] Furthermore, the step of establishing a multi-dimensional value judgment mechanism for the context within the security event chain, and generating multi-dimensional value results within the chain through the multi-dimensional value judgment mechanism, includes: The security event chain is expanded into multiple context entries according to the scenario classification type, and a maximum initial weight is set for a specified context to calculate the time sensitivity score within the chain based on the maximum initial weight. The system detects whether vehicle trajectory and navigation planning are involved in current control, and whether cockpit prompts, driver takeover prompts, and rendering feedback are displayed, in order to obtain detection results, and calculates the intra-chain control importance result based on the detection results.

[0008] Furthermore, the step of establishing a multi-dimensional value judgment mechanism for the context within the security event chain, and generating multi-dimensional value results within the chain through the multi-dimensional value judgment mechanism, further includes: The parsing time of the context entry in the cockpit chip and the driving domain controller is read, and the parsing time is converted with the cache access parameters to generate the in-chain calculation access result; The in-chain time sensitivity score, in-chain control importance result, and in-chain computation access result are fused to obtain a comprehensive value score. The context entries are then sorted according to the numerical order of the comprehensive value score to generate the in-chain multidimensional value result.

[0009] Furthermore, the step of setting different partitions for the context within the same chain of consecutive security events based on the multidimensional value results within the chain to determine cache occupancy boundaries includes: Obtain the unallocated capacity within the cache layer, and set the chain-level allocation capacity for the same continuous security event chain based on the unallocated capacity and the multidimensional value result within the chain; Within the chain-level allocation capacity, identify context entries of a specified scenario type and extract the comprehensive value score corresponding to the context entries to determine the permanent area based on the comprehensive value score; Within the chain-level allocation capacity, space is reserved for cockpit prompts, pilot takeover prompts, and rendering feedback to set up an auxiliary substitution area, and attenuation release determination is performed on the completed process to obtain the attenuation release area. The different partitions include the resident area, the auxiliary substitution area, and the decay release area, and the cache occupancy boundary is obtained by fixing the boundaries of the resident area, the auxiliary substitution area, and the decay release area.

[0010] Furthermore, based on the cache occupancy boundary, ensuring continuous availability of the resident context in the cache and prohibiting unlimited appending of cache occupancy to the same security event chain to perform chained cache updates includes: The updated context is matched with the chain of consecutive security events, and if the updated context belongs to the chain of consecutive security events, the updated context is replaced within the cache occupancy boundary. When the updated context does not belong to the continuous security event chain, a security event chain is established for the updated context, and the compressed security event chain is gradually removed.

[0011] Furthermore, when the continuous security event chain enters the end state, a closed-loop review is performed on each partition to generate context-aware results and cache update results, including: Based on the results of the chained cache update, the end state of each consecutive security event chain is reviewed, and the event chain closure score corresponding to each consecutive security event chain is determined after the review. When the event chain closure score reaches the preset closure judgment requirement, the continuous security event chain is switched from the high-speed permanent state to the value decay state, and the decay priority is determined when the continuous security event chain switches from the high-speed permanent state to the value decay state.

[0012] Furthermore, the step of performing a closed-loop review of each partition when the continuous security event chain enters the end state to generate context-aware results and cache update results also includes: Each partition is released uniformly according to the decay priority, and contexts that have entered the end state are processed first during the uniform release, while the cache release result is output. The cache space released is determined by the cache release result, and the context to be responded to is selected and added to the cache space to generate the context-aware result and cache update result.

[0013] Secondly, this disclosure provides a context-aware parsing and cache update apparatus for heterogeneous vehicle-side devices, used to implement the context-aware parsing and cache update method for heterogeneous vehicle-side devices as described in any one of the first aspects, the apparatus comprising: The event chain construction module is used to acquire vehicle-side cached data and divide the vehicle-side cached data into scenarios to construct a security event chain; The value result generation module is used to establish a multi-dimensional value judgment mechanism for the context within the security event chain, and to generate multi-dimensional value results within the chain through the multi-dimensional value judgment mechanism. The occupancy boundary determination module is used to set different partitions for the context within the same chain of continuous security events based on the multidimensional value results within the chain, so as to determine the cache occupancy boundary. The chained cache update module is used to ensure that the resident context remains continuously available in the cache based on the cache occupancy boundary, and to prohibit unlimited addition of cache occupancy to the same security event chain in order to perform chained cache updates. The perception update generation module is used to perform closed-loop verification on each partition when the continuous security event chain enters the end state, so as to generate context-aware results and cache update results.

[0014] Thirdly, this disclosure provides an electronic device including a memory and at least one processor, the memory storing a computer program, and the processor executing the computer program to implement the method of the first aspect described above.

[0015] Fourthly, this disclosure provides a computer storage medium storing a computer program that, when executed, implements the method described in the first aspect.

[0016] The beneficial effects of this disclosure are that, compared with the prior art, this disclosure has the following advantages: (1) This invention acquires vehicle-side cached data and divides the vehicle-side cached data into scenarios to construct a safety event chain. Then, a multi-dimensional value judgment mechanism is established for the context within the constructed safety event chain, and multi-dimensional value results within the chain are generated through the multi-dimensional value judgment mechanism. Subsequently, different partitions are set for the context within the same continuous safety event chain according to the generated multi-dimensional value results within the chain to determine the high-speed cache occupancy boundary. Instead of increasing the cache resident weight solely based on a single safety scenario such as emergency braking or lane departure, the continuous safety event chain is used as the cache scheduling object. Multi-dimensional value judgment based on time sensitivity, computational overhead, data importance, and access frequency fundamentally avoids the long-term accumulation of safety-related context.

[0017] (2) Based on the aforementioned cache occupancy boundaries, this invention ensures the continuous availability of resident contexts in the cache and prohibits unlimited addition of cache occupancy to the same security event chain, thereby performing chained cache updates. Finally, when a continuous security event chain enters its final state, a closed-loop review is performed on each partition to generate context-aware results and cache update results. Based on multi-dimensional value judgments of time sensitivity, computational overhead, data importance, and access frequency, the invention further identifies the start, continuation, mitigation, and end boundaries of security event chains, enabling contexts such as emergency braking, target vehicle trajectory, lane departure, navigation replanning, and cockpit prompts to maintain high-speed response during critical control phases and to release cache space promptly according to value decay results after the control loop ends. This ensures that auxiliary contexts such as driver takeover alerts, rendering feedback, and cockpit prompts can enter the cache in a timely manner. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0019] Figure 1 The diagram illustrates the flow chart of the context-aware parsing and cache update method for heterogeneous vehicle-side devices provided by the present invention.

[0020] Figure 2 A schematic diagram of the context-aware parsing and cache update device for heterogeneous vehicle-side devices provided by the present invention is shown.

[0021] The accompanying drawings have illustrated specific embodiments of this disclosure, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concepts of this disclosure to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0022] The present disclosure will be further described below with reference to the accompanying drawings. The following embodiments are only used to illustrate the technical solutions of the present disclosure more clearly, and should not be used to limit the scope of protection of the present disclosure.

[0023] The components of the embodiments of the invention described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0024] In the following, the terms “comprising,” “having,” and their cognates, which may be used in various embodiments of the invention, are intended only to indicate a particular feature, number, step, operation, element, component, or combination thereof, and should not be construed as excluding, firstly, the presence of one or more other features, numbers, steps, operations, elements, components, or combinations thereof, or adding the possibility of one or more features, numbers, steps, operations, elements, components, or combinations thereof.

[0025] Unless otherwise specified, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of the invention pertain. Terms (such as those defined in commonly used dictionaries) shall be interpreted as having the same meaning as in their contextual meaning in the relevant technical field and shall not be interpreted as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of the invention.

[0026] like Figure 1 As shown in one embodiment, a context-aware parsing and cache update method for heterogeneous vehicle-side devices includes the following steps: Step S110: Obtain vehicle-side cached data and divide the vehicle-side cached data into scenarios to construct a security event chain.

[0027] In some embodiments, the context-aware parsing and cache update method for heterogeneous vehicle-side devices provided by the present invention includes the following steps in step S110: Step S111: Time-align and cache number the context in the vehicle-side cached data to construct an initial context entry set, and identify security event scenarios through the initial context entry set.

[0028] Step S112: Extract the context corresponding to the security event scenario and determine whether the context corresponding to the security event scenario is merged into the same continuous security event chain. If so, merge the context corresponding to the security event scenario into the same continuous security event chain.

[0029] Step S113: Based on the result of grouping the contexts corresponding to security event scenarios into the same continuous security event chain, a continuous security event candidate chain is generated, and the continuous security event candidate chain is filtered to construct a security event chain that meets the set requirements.

[0030] In a specific embodiment, the context-aware parsing and cache update method for heterogeneous vehicle-side devices provided by the present invention includes steps 1 to 5: Step 1: Merge the event chains of the security-related contexts generated by heterogeneous vehicle-side devices.

[0031] The vehicle-side cache management module receives image context, point cloud context, and natural language semantic context generated by the cockpit chip, autonomous driving domain controller, and sensors, and labels them according to scenarios such as emergency braking, lane departure, target vehicle trajectory, navigation replanning, and cockpit prompts. Contexts generated within the same driving area, the same continuous window of time, the same vehicle control object, or the same driver interaction are not treated as independent high-time-sensitive cache blocks, but are merged into the same continuous safety event chain. Contexts that occur only once and do not trigger subsequent takeover alerts, navigation replanning, or rendering feedback are marked as single safety events, ultimately forming a safety event chain labeling result, including the following sub-steps:

[0032] Sub-step 1.1: Collect and unify the basic entries of security-related contexts.

[0033] Specifically, the vehicle-side cache management module first binds unified timestamps and cache block numbers to the contexts output by various heterogeneous vehicle-side devices, enabling each basic context entry to be traced back to its specific device origin. For example, image context comes from the camera and cockpit rendering chain, point cloud context comes from LiDAR or other perception sensors, and natural language semantic context comes from driver voice interaction. For each basic context entry, it is not directly marked as a high-time-sensitive cache block, but rather its generation time, driving area number, and vehicle control object number are retained as the basic input for subsequent event chain merging.

[0034] Sub-step 1.2: Mark the security scenario strength of the context base entries.

[0035] Specifically, the vehicle-side cache management module determines which category a context entry belongs to—emergency braking, lane departure, target vehicle trajectory, navigation replanning, or cockpit prompts—based on the context entry's scenario origin and functional relationships. When the same context entry is associated with multiple scenarios, manual priority override is not performed. Instead, when multiple safety scenarios are associated with the same context, the contribution of highly associated scenarios is enhanced, highlighting the contexts truly involved in the safety event, while ordinary cockpit displays or ordinary voice interactions are not mistakenly marked as core entries in a continuous safety event chain.

[0036] Sub-step 1.3: Calculate the chain association strength between adjacent contexts.

[0037] Specifically, for contexts whose safety scenario strength values ​​meet the initial screening criteria, it is further determined whether they should be grouped into the same continuous safety event chain. If two contexts occur in the same elevated merging area, the same lane departure process, the same vehicle decelerating suddenly ahead, or the same driver voice navigation change process, the chain association strength between the two is calculated, expressed as: ; In the formula, Indicates the first The context and the first The strength of the chain association between contexts is dimensionless. This indicates the time difference between the two generation times, in seconds. This represents the continuous time window parameter, ranging from 1.5 to 3.0 seconds. This indicates the distance between the centers of their corresponding driving areas, in meters; This represents the parameters of a continuous spatial window, ranging from 20 to 35 meters. This represents the relative speed difference between the two target vehicles, expressed in meters per second. This represents the continuous window parameter for velocity, ranging from 4 to 7 meters per second. This indicates a consistent flag for the vehicle control object, and can be either 0 or 1. This indicates a consistent driving area, and can be either 0 or 1. This indicates a continuous marker for driver interaction, and can be either 0 or 1. Indicates a scene conflict flag, which can be 0 or 1; , , The parameters representing consistency enhancement are all set between 0.20 and 0.35. The parameter represents the conflict suppression parameter, ranging from 0.30 to 0.50. This formula indicates that the closer the time, the closer the space, and the smaller the difference in target speed, the more likely they are to belong to the same continuous safety event chain; if the controlled object, driving area, or driver interaction process is consistent, the merging tendency is enhanced; if obvious scene conflicts occur, false merging is suppressed.

[0038] Sub-step 1.4: Perform chain merging of continuous security events according to the chain association strength.

[0039] Specifically, the vehicle-side cache management module uses the chain association strength results as a basis to group context base entries that meet the merging conditions into the same continuous security event chain. Merging is not based solely on the security level of a single cache block, but rather considers both the continuity of scenarios within the chain and changes in cache usage. For existing candidate event chains, contexts that appear consecutively and have strong correlations should be confirmed as event chains. However, when the cache block usage within a chain is too high, unlimited expansion needs to be suppressed to provide a pre-constraint for subsequent cache usage boundaries.

[0040] Sub-step 1.5: Distinguish between a continuous chain of security events and a single security event and output the labeling results.

[0041] Specifically, the vehicle-side cache management module makes the final determination of the candidate results for continuous safety event chains. If a candidate event chain contains two or more consecutive contexts and triggers at least one of subsequent takeover alerts, navigation replanning, or rendering feedback, it is marked as a continuous safety event chain. If only one emergency braking, lane departure, or single cockpit alert occurs and does not trigger subsequent takeover alerts, navigation replanning, or rendering feedback, it is marked as a single safety event. For continuous safety event chains, the module further distinguishes between core contexts such as emergency braking, lane departure, and target vehicle trajectory, as well as auxiliary contexts such as cockpit alerts, driver takeover alerts, navigation replanning, and rendering feedback. The final output safety event chain marking result will serve as the direct input for generating the multi-dimensional value result within the chain in step 2, so that subsequent cache value calculations are no longer oriented towards isolated cache blocks, but rather towards different context entries within the continuous safety event chain.

[0042] Step S120: Establish a multi-dimensional value judgment mechanism for the context within the security event chain, and generate multi-dimensional value results within the chain through the multi-dimensional value judgment mechanism.

[0043] In some embodiments, the context-aware parsing and cache update method for heterogeneous vehicle-side devices provided by the present invention includes the following steps in step S120: Step S121: Expand the security event chain into multiple context entries according to the scenario classification type, and set the highest initial weight for the specified context to calculate the time sensitivity score within the chain based on the highest initial weight.

[0044] Step S122: Detect whether the vehicle trajectory and navigation planning are involved in the current control and whether cockpit prompts, driver takeover prompts and rendering feedback are presented, in order to obtain the detection results, and calculate the intra-chain control importance result based on the detection results.

[0045] In some embodiments, the context-aware parsing and cache update method for heterogeneous vehicle-side devices provided by the present invention further includes the following steps in step S120: Step S123: Read the parsing time of the context entry in the cockpit chip and the driving domain controller, and convert the parsing time with the cache access parameters to generate the in-chain calculation access result.

[0046] Step S124: The on-chain time sensitivity score, on-chain control importance result, and on-chain computation access result are integrated to obtain a comprehensive value score. The context entries are then sorted according to the numerical order of the comprehensive value score to generate an on-chain multidimensional value result.

[0047] In a specific embodiment, the context-aware parsing and cache update method for heterogeneous vehicle-side devices provided by the present invention, step 2, generates multi-dimensional value results within the chain based on the safety event chain marking results. The cache value calculation module reads the safety event chain marking results formed in step 1 and establishes multi-dimensional value judgments on time sensitivity, computational overhead, data importance, and access frequency for each context within the chain. The contexts corresponding to emergency braking and lane departure are set to the highest time sensitivity level; the target vehicle trajectory and navigation replanning are determined to have data importance based on whether they still participate in the current control loop; the cockpit prompt interface, driver takeover reminder, and rendering feedback are determined to have auxiliary response value based on whether they need to be presented in real time; the rendering and natural language semantic parsing contexts are determined to have computational overhead based on their inference time on the cockpit chip and autonomous driving domain control. Thus, multi-dimensional value results within the chain are formed, including the following sub-steps: Sub-step 2.1: Expand the context entries within the security event chain.

[0048] Specifically, the cache value calculation module reads the continuous safety event chain markers, single safety event markers, the core context list within the chain, and the auxiliary context list within the chain. It then expands emergency braking, lane departure, target vehicle trajectory, navigation replanning, driver takeover alerts, cockpit prompts, and rendering feedback within the same safety event chain into independent context entries. For a single safety event, only its single context number and scene marker are retained; for a continuous safety event chain, it records whether the context is at the beginning, middle, or end of the chain, providing a basis for subsequent determination of whether it still participates in the current control loop.

[0049] Sub-step 2.2: Calculate the time sensitivity results of the in-chain context.

[0050] Specifically, the highest base weight is set for emergency braking and lane departure contexts, but adjustments are still made based on the interval between the generation time and the current time, whether the event chain has been mitigated, and whether the context is still at the end of the chain, to prevent old, highly sensitive contexts from lingering indefinitely.

[0051] In this embodiment, the expression for the time sensitivity score is: ; In the formula, Indicates the first Time sensitivity score of each chain context, dimensionless; Indicates emergency braking, indicated by a value of 0 or 1; This indicates a lane departure warning, and can be either 0 or 1. This represents the normalized value of the security scenario intensity, ranging from 0 to 1. This indicates that the chain tail is involved in the flag, and can be either 0 or 1. This indicates an event chain de-escalation flag, which can be either 0 or 1. This indicates the current time, in seconds. Indicates the context generation time, in seconds; This represents the time-sensitive decay parameter, ranging from 1.8 to 2.6 seconds, because the critical buffer responses to emergency braking and lane departure on urban roads are typically concentrated within a few seconds. This indicates the emergency braking enhancement parameter, ranging from 0.85 to 1.05; This represents the lane departure warning enhancement parameter, ranging from 0.75 to 0.95. This represents the scene intensity enhancement parameter, ranging from 0.25 to 0.40; This indicates that the chain tail participates in the enhancement parameter, which is set to 0.20-0.35; This represents the slow-release inhibition parameter, ranging from 0.45 to 0.65. This formula indicates that the higher the safety level and the closer to the current control endpoint, the higher the time sensitivity; however, the older the context or the slower the event chain, the lower the time sensitivity.

[0052] Sub-step 2.3 calculates the data importance and auxiliary response value of the in-chain context.

[0053] Specifically, the cache value calculation module determines whether the target vehicle trajectory and navigation replanning are still involved in the current control loop, and whether the cockpit prompt interface, driver takeover alert, and rendering feedback need to be presented in real time. For the target vehicle trajectory that is still involved in the current driving path determination, the data importance is increased; for navigation replanning that has completed path switching, the data importance is reduced; for takeover alerts and cockpit prompt interfaces that have not yet been presented, the auxiliary response value is increased.

[0054] In this embodiment, the expressions for data importance score and auxiliary response value score are as follows: ; ; In the formula, Indicates the first Data importance scores for each chain's context, dimensionless; This indicates that the target vehicle's trajectory is included in the marking, and can be either 0 or 1. Indicates the current control closed-loop participation flag, which can be 0 or 1; A marker indicating that navigation replanning still affects the current driving path, with a value of 0 or 1; A flag indicating that the context has been replaced by a subsequent context in the chain, which can be 0 or 1; A flag indicating that the context has finished rendering or has ended its control action, which takes the value 0 or 1; Indicates the first Auxiliary response scores for each chain context, dimensionless; This indicates a cockpit prompt interface marker, which can be either 0 or 1. The indicator is a driver takeover warning symbol, which can be either 0 or 1. Indicates the rendering feedback flag, which can be 0 or 1; This indicates the timing parameter for auxiliary presentation, ranging from 2.0 to 3.0 seconds. , , This represents the control importance enhancement parameter, ranging from 0.35 to 0.55; This represents the alternative suppression parameter, ranging from 0.45 to 0.70; This indicates the end of the suppression parameter, which is set to 0.60-0.85; , , This represents the auxiliary response enhancement parameter, ranging from 0.30 to 0.55; The suppression parameter is set to 0.50-0.75. The formula above indicates that contexts still involved in the vehicle control closed loop have high data importance; cockpit prompts, takeover alerts, and rendering feedback that have not yet been rendered in real-time have high auxiliary response value.

[0055] Sub-step 2.4: Determine the computational overhead and access frequency of the in-chain context.

[0056] Specifically, the cache value calculation module reads the parsing time of the context on the cockpit chip and the autonomous driving domain controller, and calculates it by combining the cache block size, access frequency, and recent access time. For rendering and natural language semantic parsing contexts, if their inference time on the cockpit chip or autonomous driving domain controller is high, the computational overhead score is increased; for contexts frequently accessed in a short period of time, the access popularity score is increased; for contexts with excessive size, occupancy suppression is implemented when calculating value to prevent large cache blocks from occupying the cache for a long time simply because of high access frequency.

[0057] Understandably, the longer the parsing and reconstruction take, the more worthwhile it is to cache; however, the larger the cache block, the higher the cost of occupying the cache; the more frequent the access and the closer it is to the current moment, the higher the access popularity.

[0058] In this embodiment, the expressions for calculating the cost score and the access popularity score are as follows: ; ; In the formula, Indicates the first The computational overhead score for each in-chain context is dimensionless. This indicates the time taken to resolve this context on the cockpit chip, in milliseconds; This indicates the time taken to resolve this context on the autonomous driving domain controller, in milliseconds; This indicates the time taken for the most recent reconstruction of this context, in milliseconds. This indicates the size of the context cache block, in megabytes. This represents the baseline for cockpit chip parsing time, taken as 18-25 milliseconds. This represents the baseline for autonomous driving domain control parsing time, taken as 12-18 milliseconds; This represents the baseline for context reconstruction time, taken as 20-30 milliseconds. This represents the baseline cache block size, ranging from 8 to 12 megabytes. This represents the visit popularity score, which is dimensionless. This indicates the number of visits within the recent statistics window, in units of times. This indicates a consecutive hit flag, which can be either 0 or 1. This indicates the time of the most recent access, in seconds. This represents the access popularity decay parameter, ranging from 2.5 to 4.0 seconds.

[0059] Sub-step 2.5 integrates and forms a multi-dimensional value result within the chain.

[0060] Specifically, the cache value calculation module comprehensively and weights time sensitivity, data importance, auxiliary response value, computational overhead, and access frequency to form a comprehensive value score for each context within the chain. During execution, the weights in the weighted fusion are fine-tuned according to the vehicle engineering calibration results, but time sensitivity and data importance should be higher than access frequency. This is because, for a continuous security event chain, ordinary high-frequency access contexts cannot overshadow the security closed-loop context. Furthermore, for a continuous security event chain, the comprehensive value score is not only used to determine whether to retain the context, but also to distinguish the core persistent area, the auxiliary waiting area, and the decay release area in subsequent step 3. Finally, the system sorts the contexts within the chain according to their comprehensive value scores from high to low, and outputs the multi-dimensional value results within the chain.

[0061] Step S130: Based on the multidimensional value results within the chain, set different partitions for the context within the same chain of continuous security events to determine the cache occupancy boundary.

[0062] In some embodiments, the context-aware parsing and cache update method for heterogeneous vehicle-side devices provided by the present invention includes the following steps in step S130: Step S131: Obtain the capacity to be allocated within the cache layer, and set the chain-level allocation capacity for the same continuous security event chain based on the capacity to be allocated and the multidimensional value results within the chain.

[0063] Step S132: Identify context entries of a specified scenario type within the chain-level allocation capacity, and extract the comprehensive value score corresponding to the context entries to determine the permanent area based on the comprehensive value score.

[0064] Step S133: Within the chain-level allocation capacity, reserve space for cockpit prompts, pilot takeover prompts, and rendering feedback to set up an auxiliary substitution area, and perform attenuation release determination on the completed process to obtain the attenuation release area.

[0065] The different partitions include the resident area, the auxiliary substitution area, and the decay release area. The cache occupancy boundary is obtained by fixing the boundaries of the resident area, the auxiliary substitution area, and the decay release area.

[0066] In a specific embodiment, the context-aware parsing and cache update method for heterogeneous vehicle-side devices provided by the present invention, step 3, generates a high-speed cache occupancy boundary based on the multi-dimensional value results within the chain. The cache scheduling module reads the multi-dimensional value results within the chain formed in step 2 and sets a core persistent area, an auxiliary waiting area, and a decay release area for contexts within the same continuous safety event chain. Contexts involving emergency braking, lane departure, and those still participating in target vehicle trajectory judgment enter the core persistent area; driver takeover alerts, cockpit prompt interfaces, and rendering feedback enter the auxiliary waiting area, with reserved space for caching; contexts that have completed navigation replanning, ended voice interaction, or are no longer participating in the current control loop enter the decay release area. This forms the high-speed cache occupancy boundary result, including the following sub-steps: Sub-step 3.1 generates the cache allocable capacity for a continuous chain of security events.

[0067] Specifically, the cache scheduling module reads the cache block size, comprehensive value score, time sensitivity score, and data importance score from the multi-dimensional value results within the chain, and combines this with the current allocable capacity of the cache layer to provide a chain-level allocable capacity for each consecutive security event chain. This capacity is not the total remaining space of the cache, but rather the upper limit allowed for that event chain to occupy within the current scheduling cycle, used to prevent consecutive security event chains from infinitely appending to the cache.

[0068] Sub-step 3.2: Determine the candidate boundaries of the core resident area based on the chain-level capacity initialization results.

[0069] Specifically, within the chain-level allocable capacity, the cache scheduling module prioritizes identifying contexts related to emergency braking, lane departure, and those still involved in determining the target vehicle's trajectory. For each in-chain context, it not only examines its overall value score but also combines time sensitivity scores, data importance scores, cache block size, and whether its control function has ended to determine its core resident priority.

[0070] Understandably, contexts related to emergency braking, lane departure, and target vehicle trajectory that are still within the control loop enter the core persistent area; contexts that only have cockpit prompts and rendering feedback attributes do not directly occupy the core persistent capacity.

[0071] Sub-step 3.3: Determine the boundary of the auxiliary waiting area to be filled based on the candidate results of the core permanent resident area.

[0072] Specifically, after defining the core persistent candidates, the cache scheduling module does not release all remaining space to ordinary cache blocks. Instead, it reserves space within the chain-level allocable capacity for driver takeover alerts, cockpit prompts, and rendering feedback. For auxiliary contexts, the module focuses on determining whether they have not yet been rendered in real time, whether they are related to the current driver interaction process, whether they have been accessed continuously in a short period of time, and whether their cache block size is suitable for caching. Among these, cockpit prompts, takeover alerts, and rendering feedback that have not yet been rendered, require immediate driver awareness, and have high reconstruction costs are given priority for allocation; however, auxiliary contexts that have already been rendered must not continue to occupy allocation space.

[0073] Sub-step 3.4: Determine the boundary of the attenuation release zone based on the candidate results of the auxiliary waiting zone.

[0074] Specifically, the cache scheduling module performs attenuation and release checks on contexts that have completed navigation replanning, ended voice interaction, been replaced by subsequent in-chain contexts, or are no longer involved in the current control loop. Attenuation and release do not involve immediate deletion; instead, a release priority is assigned. Contexts with high release priority and not belonging to the core persistent area or auxiliary waiting area are released into the attenuation and release area. Contexts that are more clearly ended, older, or more likely to be replaced should be released more frequently; however, contexts with high overall value, time sensitivity, and data importance should not be released prematurely.

[0075] Sub-step 3.5: Solidify the cache occupancy boundary results.

[0076] Specifically, the cache scheduling module solidifies the boundaries of candidate results for the core permanent area, candidate results for the auxiliary waiting area, and candidate results for the decay release area. The execution process is as follows: First, the core permanent area is loaded according to the priority of core permanent area from high to low, and the cumulative capacity shall not exceed the upper limit of core permanent area capacity; then, the auxiliary waiting area is formed according to the priority of auxiliary replacement from high to low, and the reserved capacity shall not exceed the replacement capacity of the auxiliary waiting area; finally, the decay release area is formed according to the release priority from high to low, and its release capacity is fed back to the chained cache update in step 4. For the case where the same context meets multiple area conditions at the same time, the unique ownership determination is made according to the order of core permanent area, auxiliary waiting area, and decay release area; for contexts that have been presented but still have driver takeover association, a delayed release flag is set to retain the replacement qualification for a short scheduling cycle.

[0077] During execution, it is necessary to determine whether the core resident area and auxiliary waiting area exceed the chain-level allocable capacity, and whether the review decay release area can reclaim the total occupancy within the chain. If the review value rises abnormally, it indicates that there is still a risk of cache crowding out in this event chain, and low-value cache blocks within the chain need to be replaced first when performing chained cache updates in step 4.

[0078] Step S140: Based on the cache occupancy boundary, the resident area context is kept continuously available in the cache, and unlimited appending of cache occupancy to the same security event chain is prohibited, so as to perform chained cache update.

[0079] In some embodiments, the context-aware parsing and cache update method for heterogeneous vehicle-side devices provided by the present invention includes the following steps in step S140: Step S141: Match the updated context with the chain of consecutive security events, and replace the updated context within the cache occupancy boundary if the updated context belongs to the chain of consecutive security events.

[0080] Step S142: When the updated context does not belong to the continuous security event chain, establish a security event chain for the updated context and loosely remove the compressed security event chain.

[0081] In a specific embodiment, the context-aware parsing and cache update method for heterogeneous vehicle-side devices provided by the present invention, step 4, performs chained cache updates based on the cache occupancy boundary results. The cache update module reads the cache occupancy boundary results formed in step 3, prioritizes maintaining the continuous availability of the core resident area context in the cache, and prohibits unlimited addition of cache occupancy to the same security event chain. When a new lane departure, target vehicle trajectory change, or navigation replanning context enters, it first determines whether it belongs to an existing continuous security event chain; if it belongs to the same event chain, the cache block in the low-value chain is replaced within the original cache occupancy boundary; if it belongs to a new security event chain, the core resident area is reallocated according to time sensitivity and data importance, and the old chain context that has been buffered is removed from the cache, finally forming a chained cache update result, including the following sub-steps: Sub-step 4.1: Determine the security event chain to which the new context belongs.

[0082] Specifically, the cache update module reads the core resident context list, auxiliary pending context list, decay release context list, and chain-level cache occupancy boundary formed in step 3, and matches the newly entering context with the existing continuous security event chain. The matching does not rely solely on identical scenario names; instead, it considers the temporal continuity, spatial continuity, control object consistency, time sensitivity score formed in step 2, data importance score, and cache block size to determine ownership.

[0083] Understandably, the closer a new context is to the end of an existing event chain and the more it participates in the current control loop, the more likely it is to be grouped into the same event chain; however, the higher the boundary occupancy of the existing event chain and the larger the new context cache block, the more necessary it is to suppress further additions.

[0084] Sub-step 4.2: Perform low-value cache block replacement within the same event chain boundary.

[0085] Specifically, when the new context chain ownership determination result shows that it belongs to an existing continuous safety event chain, the cache update module does not increase the chain-level allocable capacity of that event chain, but replaces it within the high-speed cache occupancy boundary fixed in step 3. If the new context is lane departure or still involved in target vehicle trajectory determination, it will prioritize entering the core persistent area; if the new context is navigation replanning or cockpit prompt linkage context, it will prioritize entering the auxiliary waiting area. The replacement object is selected from the decay release area, the tail of the auxiliary waiting area, and the in-chain cache block that has ended its control function within the same event chain. The new context is only allowed to be replaced within the original boundary if its value, time sensitivity, and data importance are indeed higher than the low-value cache block in the chain, and it will not significantly increase the capacity burden.

[0086] Sub-step 4.3: Perform core resident area reallocation on the new security event chain.

[0087] Specifically, when a new context does not belong to an existing continuous security event chain, and its time sensitivity and data importance meet the triggering conditions for a new chain, the cache update module establishes a new security event chain for it. At this time, the system does not directly add capacity to the cache layer, but reallocates the core resident area according to the time sensitivity and data importance of the new chain, and reclaims capacity from the old chain that is already in a buffered state.

[0088] Sub-step 4.4 removes the deferred old chain context from the cache.

[0089] Specifically, the cache update module performs a removal judgment on the compressed old chain based on the new chain core reallocation result. During the execution, contexts that have ended, become obsolete, or been replaced in the old chain are removed first; contexts that still serve as safety prompts or takeover alerts are released with a delay. That is, cache blocks in the old chain that have completed navigation replanning, ended voice interaction, are no longer involved in the current control loop, or have been replaced by subsequent contexts are removed first; cache blocks that are still related to driver takeover alerts or safety prompt presentation are set to release with a delay and are not removed immediately.

[0090] Sub-step 4.5: Consolidate the chained cache update results.

[0091] Specifically, the cache update module uniformly solidifies the replacement results within the same chain boundary, the core reallocation results of the new chain, and the delayed removal results of the old chain. During execution, it first confirms that the contexts in the core persistent area, such as emergency braking, lane departure, and those still involved in the target vehicle trajectory judgment, are continuously available; then it confirms that the driver takeover reminder, cockpit prompt interface, and rendering feedback in the auxiliary waiting area have room for replacement; finally, it confirms that the old chain contexts in the decay release area have been removed or have a delayed release flag set. If the actual cache occupancy exceeds the chain-level cache occupancy boundary formed in step 3, the low-value replacement within the same chain is re-executed instead of directly adding capacity. The final chain-like cache update result will serve as the direct input to the context-aware parsing and cache update result output in step 5, ensuring that the same security event chain remains updated within the boundary when consecutive security events overlap, rather than occupying the cache without limit.

[0092] Step S150: When the continuous security event chain enters the end state, a closed-loop review is performed on each partition to generate context-aware results and cache update results.

[0093] In some embodiments, the context-aware parsing and cache update method for heterogeneous vehicle-side devices provided by the present invention includes the following steps in step S150: Step S151: Based on the result of the chained cache update, review the end state of each consecutive security event chain, and determine the event chain closure score corresponding to each consecutive security event chain after review.

[0094] Step S152: When the event chain closure score reaches the preset closure judgment requirement, the continuous security event chain is switched from the high-speed resident state to the value decay state, and the decay priority is determined when the continuous security event chain switches from the high-speed resident state to the value decay state.

[0095] In some embodiments, the context-aware parsing and cache update method for heterogeneous vehicle-side devices provided by the present invention further includes the following steps in step S150: Step S153: Release each partition uniformly according to the decay priority, and prioritize the processing of contexts that have entered the end state during the uniform release, while outputting the cache release result.

[0096] Step S154: Determine the released cache space based on the cache release result, and select the context to be responded to to fill the cache space, so as to generate context-aware results and cache update results.

[0097] In a specific embodiment, the context-aware parsing and cache update method for heterogeneous vehicle devices provided by the present invention, step 5, outputs the context-aware parsing and cache update results based on the chained cache update results. The cache output module reads the chained cache update results formed in step 4, and performs closed-loop verification of the core resident area, auxiliary waiting area, and decay release area when the continuous safety event chain enters the end state. When the emergency braking has been released, the lane departure risk has subsided, the target vehicle trajectory no longer affects the current driving path, the navigation replanning has been completed, and the cockpit prompt has been presented, the corresponding safety event chain is switched from the high-speed resident state to the value decay state, and the high-speed cache space it occupies is released; at the same time, the contexts that still require immediate response in the driver takeover reminder, rendering feedback, and cockpit prompt interface are added to the high-speed cache. Finally, the context-aware parsing and cache update results for heterogeneous vehicle devices are output, so that high time-sensitive contexts no longer form a resident cache crowding chain when continuous safety events are superimposed, including the following sub-steps:

[0098] Sub-step 5.1: Verify the end status of the continuous security event chain.

[0099] Specifically, the cache output module reads the updated core resident area context list, auxiliary waiting area context list, decay release area context list, new chain number, old chain release record, and actual cache usage generated in step 4. During execution, for each consecutive safety event chain, it confirms whether emergency braking has been released, lane departure risk has subsided, target vehicle trajectory no longer affects the current driving path, navigation replanning has been completed, and cockpit prompts have been displayed. Therefore, only when safety actions, trajectory impact, navigation replanning, and cockpit prompts are all in a closed state, and there is no context still participating in the current control loop, is the event chain allowed to transition from the high-speed resident state to the value decay state.

[0100] Sub-step 5.2: Switch the closed event chain to a value decay state.

[0101] Specifically, when the event chain closure score meets the closure judgment requirements, the cache output module switches the event chain from a high-speed resident state to a value decay state. Core contexts such as emergency braking, lane departure, and target vehicle trajectory no longer maintain the highest resident weight, but are instead ranked according to whether their control function has ended, whether they have been replaced by subsequent contexts, and whether they still serve as takeover alerts. Therefore, contexts that have ended, been replaced, or completed their interactions are given priority in decay; contexts that still have high overall value, high time sensitivity, high data importance, or takeover alert functions are decayed later.

[0102] Sub-step 5.3: Release the cache space in the value decay state.

[0103] Specifically, the cache output module releases the core persistent area, auxiliary waiting area, and decay release area uniformly according to decay priority. During release, contexts that have completed navigation replanning, ended voice interaction, and are no longer involved in the current control loop are processed first; for contexts that still need to retain short-term takeover prompts, only a delayed release flag is set.

[0104] Understandably, the more attenuable contexts there are and the higher the attenuation priority, the larger the release capacity; however, contexts that still serve as auxiliary responses and takeover alerts need to be appropriately preserved to avoid the release action affecting the driver's perception.

[0105] Sub-step 5.4 adds auxiliary contexts that still require immediate response to the cache.

[0106] Specifically, after freeing up cache space, the cache output module selects driver takeover alerts, cockpit prompts, and rendering feedback contexts that still require immediate response from the auxiliary waiting area formed in step 4 and adds them to the cache. When filling in the gaps, priority is given to auxiliary contexts that have not yet been rendered, have high access frequency, high computational overhead, and moderate cache block size. That is, auxiliary contexts that still require immediate driver awareness and have high reconstruction costs are added first, while avoiding auxiliary contexts from crowding out core safety contexts.

[0107] Sub-step 5.5 outputs the context-aware parsing and cache update results.

[0108] Specifically, the cache output module performs a final check on the released capacity, the fill capacity, and the current actual cache occupancy. If a closed event chain has switched from a high-speed resident state to a value decay state, the released capacity can cover the auxiliary fill capacity, and the core resident area only retains the safety contexts still participating in the current control loop, then a continuous safety event chain crowding-out elimination flag is output. The final results show that highly time-sensitive contexts such as emergency braking, lane departure, and target vehicle trajectory remain highly available during critical phases; after their control loop ends, they no longer continue to resident and crowd out the cache; driver takeover alerts, rendering feedback, and cockpit prompts can be promptly added to the cache after space is released, thus achieving context-aware parsing and cache updates for heterogeneous vehicle-side devices.

[0109] The context-aware parsing and cache update apparatus for heterogeneous vehicle devices provided by the present invention is described below. The context-aware parsing and cache update apparatus for heterogeneous vehicle devices described below can be referred to in correspondence with the context-aware parsing and cache update method for heterogeneous vehicle devices described above.

[0110] like Figure 2 As shown in one embodiment, a context-aware parsing and cache update device for heterogeneous vehicle-side devices includes an event chain construction module, a value result generation module, an occupancy boundary determination module, a chained cache update module, and a perception update generation module.

[0111] The event chain building module is used to obtain vehicle-side cached data and divide the vehicle-side cached data into scenarios to build a security event chain.

[0112] The value result generation module is used to establish a multi-dimensional value judgment mechanism for the context within the security event chain, and to generate multi-dimensional value results within the chain through the multi-dimensional value judgment mechanism.

[0113] The occupancy boundary determination module is used to set different partitions for the context within the same chain of consecutive security events based on the multidimensional value results within the chain, in order to determine the cache occupancy boundary.

[0114] The chained cache update module is used to maintain the continuous availability of resident context in the cache based on cache occupancy boundaries, and to prevent unlimited appending of cache occupancy to the same security event chain in order to perform chained cache updates.

[0115] The awareness update generation module is used to perform closed-loop verification of each partition when the continuous security event chain enters the end state, in order to generate context-aware results and cache update results.

[0116] According to embodiments of this disclosure, an electronic device is also provided, which may include a processor, a communications interface, a memory, and a communication bus, wherein the processor, the communications interface, and the memory communicate with each other via the communication bus. The processor can invoke logical instructions stored in the memory to execute the methods provided in the above embodiments.

[0117] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, 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 this disclosure. 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.

[0118] On the other hand, this disclosure also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the methods provided in the above embodiments.

[0119] 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.

[0120] 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.

[0121] It should be understood that the above embodiments are only used to illustrate the technical solutions of this disclosure, and not to limit them; although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or make equivalent substitutions for 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 this disclosure.

Claims

1. A context-aware resolution and cache update method for heterogeneous vehicle-end devices, characterized in that, The method includes: Acquire vehicle-side cached data and divide the vehicle-side cached data into scenarios to construct a security event chain; A multi-dimensional value judgment mechanism is established for the context within the security event chain, and multi-dimensional value results within the chain are generated through the multi-dimensional value judgment mechanism. Based on the multidimensional value results within the chain, different partitions are set for the context within the same chain of consecutive security events to determine the cache occupancy boundary. Based on the cache occupancy boundary, the resident context is kept continuously available in the cache, and unlimited addition of cache occupancy to the same security event chain is prohibited in order to perform chained cache updates; When the continuous security event chain enters the end state, a closed-loop review is performed on each partition to generate context-aware results and cache update results.

2. The context-aware resolution and cache update method for heterogeneous vehicle-end devices according to claim 1, wherein, The process of acquiring vehicle-side cached data and dividing the vehicle-side cached data into scenarios to construct a security event chain includes: The context in the vehicle-side cached data is time-aligned and cached to form an initial context entry set, and the security event scenario is identified through the initial context entry set. Extract the context corresponding to the security event scenario and determine whether the context corresponding to the security event scenario is merged into the same continuous security event chain. If so, merge the context corresponding to the security event scenario into the same continuous security event chain. Based on the result of grouping the contexts corresponding to the security event scenarios into the same continuous security event chain, a continuous security event candidate chain is generated, and the continuous security event candidate chain is filtered to construct a security event chain that meets the set requirements.

3. The method of claim 1, wherein the method further comprises: The establishment of a multi-dimensional value judgment mechanism for the context within the security event chain, and the generation of multi-dimensional value results within the chain through the multi-dimensional value judgment mechanism, includes: The security event chain is expanded into multiple context entries according to the scenario classification type, and a maximum initial weight is set for a specified context to calculate the time sensitivity score within the chain based on the maximum initial weight. The system detects whether vehicle trajectory and navigation planning are involved in current control, and whether cockpit prompts, driver takeover prompts, and rendering feedback are displayed, in order to obtain detection results, and calculates the intra-chain control importance result based on the detection results.

4. The context-aware resolution and cache update method for heterogeneous vehicle-side devices according to claim 3, wherein, The step of establishing a multi-dimensional value judgment mechanism for the context within the security event chain, and generating multi-dimensional value results within the chain through the multi-dimensional value judgment mechanism, further includes: The parsing time of the context entry in the cockpit chip and the driving domain controller is read, and the parsing time is converted with the cache access parameters to generate the in-chain calculation access result; The in-chain time sensitivity score, in-chain control importance result, and in-chain computation access result are fused to obtain a comprehensive value score. The context entries are then sorted according to the numerical order of the comprehensive value score to generate the in-chain multidimensional value result.

5. The context-aware resolution and cache update method for heterogeneous vehicle-side devices according to claim 4, wherein, The step of setting different partitions for the context within the same chain of consecutive security events based on the multidimensional value results within the chain to determine the cache occupancy boundary includes: Obtain the unallocated capacity within the cache layer, and set the chain-level allocation capacity for the same continuous security event chain based on the unallocated capacity and the multidimensional value result within the chain; Within the chain-level allocation capacity, identify context entries of a specified scenario type and extract the comprehensive value score corresponding to the context entries to determine the permanent area based on the comprehensive value score; Within the chain-level allocation capacity, space is reserved for cockpit prompts, pilot takeover prompts, and rendering feedback to set up an auxiliary substitution area, and attenuation release determination is performed on the completed process to obtain the attenuation release area. The different partitions include the resident area, the auxiliary substitution area, and the decay release area, and the cache occupancy boundary is obtained by fixing the boundaries of the resident area, the auxiliary substitution area, and the decay release area.

6. The context-aware resolution and cache update method for heterogeneous vehicle-end devices according to claim 1, wherein, The step of maintaining continuous availability of resident context in the cache based on the cache occupancy boundary and prohibiting unlimited appending of cache occupancy to the same security event chain to perform chained cache updates includes: The updated context is matched with the chain of consecutive security events, and if the updated context belongs to the chain of consecutive security events, the updated context is replaced within the cache occupancy boundary. When the updated context does not belong to the continuous security event chain, a security event chain is established for the updated context, and the compressed security event chain is gradually removed.

7. The context-aware resolution and cache update method for heterogeneous vehicle-end devices according to claim 1, wherein, When the continuous security event chain enters the end state, a closed-loop review is performed on each partition to generate context-aware results and cache update results, including: Based on the results of the chained cache update, the end state of each consecutive security event chain is reviewed, and the event chain closure score corresponding to each consecutive security event chain is determined after the review. When the event chain closure score reaches the preset closure judgment requirement, the continuous security event chain is switched from the high-speed permanent state to the value decay state, and the decay priority is determined when the continuous security event chain switches from the high-speed permanent state to the value decay state.

8. The context-aware resolution and cache update method for heterogeneous vehicle-side devices according to claim 7, wherein, The step of performing a closed-loop review of each partition when the continuous security event chain enters the end state to generate context-aware results and cache update results also includes: Each partition is released uniformly according to the decay priority, and contexts that have entered the end state are processed first during the uniform release, while the cache release result is output. The cache space released is determined by the cache release result, and the context to be responded to is selected and added to the cache space to generate the context-aware result and cache update result.

9. An electronic device, comprising: The electronic device includes a memory and at least one processor, the memory storing a computer program, and the processor executing the computer program to implement the method of any one of claims 1-8.

10. A computer storage medium, characterized in that, It stores a computer program, which, when executed, performs the method according to any one of claims 1-8.

Images