A driver route selection guidance method and system based on VMS induced information

Abstract

The application discloses a driver path selection guide method and system based on VMS induced information, and particularly relates to the technical field of path selection guide, which comprises the following steps: obtaining Beidou positioning, driving direction and speed of a vehicle, and matching pre-stored road topology to solve the current road section and the current lane of the vehicle and output the state of the vehicle; reading the induced code displayed by a target VMS, and decoding the induced code into a target split point, an action type and an action parameter according to a pre-stored code table to output the code state; and encoding the VMS induced information into decodable action parameters, combining the Beidou positioning and the road topology to construct an executable constraint domain, and performing screening and time sequence rearrangement on an action sequence in the constraint domain to form a path operation instruction corresponding to the current position of the vehicle.

Description

Technical Field

[0001] This invention relates to the field of route selection guidance technology, and more specifically, to a driver route selection guidance method and system based on VMS guidance information. Background Technology

[0002] In existing road traffic guidance technologies, in response to situations such as mainline congestion, accidents, or construction, the industry typically uses variable message signs (VMS) at key nodes to issue detour suggestions or road condition information in the form of text, arrows, or brief graphics. The main solution is to provide unified route guidance to a large number of drivers without relying on in-vehicle equipment. This approach relies on fixed-point VMS displays in engineering, allowing drivers to visually obtain information and make their own understanding and decisions. It has become the mainstream configuration method for highways and urban expressways. However, in practical applications, such as in the main lane divergence zone of highways with high vehicle speeds and dense lanes, VMS is limited by the number of display bits, refresh rate and the driver's available attention. Its information expression must be highly compressed. At the same time, a large number of vehicles do not have vehicle-road coordination or personalized push capabilities. Drivers can only judge with their naked eyes and try to map the abstract prompts into lane changing or off-ramp actions that they can perform in a very short time. Under this hard constraint, the phenomenon that can be consistently observed is that even if the VMS prompts are objectively consistent with the actual road conditions, a large number of vehicles still do not follow the guidance, or have missed the key operation window when they do follow it. The root cause is that the mainstream practice only provides "directional suggestions" but cannot directly convert the guidance information into path instructions that strictly correspond to the current vehicle position and the action that can be performed under public display conditions, which leads to the systematic weakening of the effectiveness of the information on the driver's side. The technical problem this application aims to solve is: how to enable drivers to quickly and accurately reconstruct the guidance information issued by VMS into their currently executable route operation instructions without relying on vehicle-to-infrastructure communication and limited by the low display resources of VMS. Summary of the Invention

[0003] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a driver route selection guidance method and system based on VMS guidance information. By encoding VMS guidance information into decodable action parameters and combining it with BeiDou positioning and road topology to construct an executable constraint domain, the action sequence is filtered and rearranged in time within the constraint domain to form a route operation instruction corresponding to the vehicle's current position.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a driver route selection guidance method based on VMS guidance information, comprising: S1. Obtain the vehicle's BeiDou positioning, driving direction and speed, match it with the pre-stored road topology, solve the vehicle's current road segment and current lane, and output the vehicle status; S2. Read the inducement code displayed by the target VMS, and decode the inducement code into the target diversion point, action type and action parameters according to the pre-stored code table, and output the code state; S3. Construct a diversion anchor point graph based on the pre-stored road topology. The diversion anchor point graph uses diversion point identifiers as nodes and lane reachability relationships as edges. Record the lower limit of lane change distance and the boundary of the lane change restriction interval on the edges. Then, map the vehicle state to the diversion anchor point graph and solve the reachable anchor point path set, and output the anchor point state. S4. Calculate the window cone based on vehicle state and anchor point state. The window cone consists of an executable time window, executable lane change count, and executable trigger mileage. Output the window state. S5. Based on the code state query, the inducement code action grid consists of a hold action node, a lane change action node and a diversion action node. Each action node is configured with a trigger position and a trigger threshold. Then, the inducement code action grid and the window state are subjected to intersection filtering and time sequence rearrangement to generate a set of action trigger positions and a set of trigger thresholds and form a path instruction. S6. During vehicle operation, periodically acquire BeiDou positioning and update vehicle status accordingly. Perform satisfaction judgment on the action trigger position and trigger threshold in the path instruction one by one, output the action execution status and form the execution result.

[0005] In a preferred embodiment, S1 includes: S1-1. Obtain the vehicle's current BeiDou positioning, driving direction, and speed. Based on the pre-stored road topology, retrieve the candidate road segment set and candidate lane set, and output the candidate set. The pre-stored road topology includes the merging point set, merging point identifiers, lane set, lane connection relationship, no-lane-change zone boundary, and lower limit of lane-changing distance. The pre-stored road topology refers to structured data formed by writing roads into a topology library according to a unified coordinate system. The topology library includes the merging point set and merging point identifiers, lane set and lane boundary lines and lane center lines, lane connection relationship, no-lane-change zone boundary, and lower limit of lane-changing distance. The merging point set is obtained by performing bifurcation point identification and ramp connection point identification on the road centerline network. Each diversion point is assigned a diversion point identifier according to its location sequence. The lane boundary line and lane center line are formed by connecting the lane marking point set obtained from the survey according to the sampling order after performing coordinate calibration. The lane connection relationship is formed by performing connectivity determination on the lane center line endpoints at the endpoints of adjacent road segments and writing the endpoint pairs that meet the connectivity requirement into the directed connection record. The boundary of the no-lane-change section is formed by extracting the start and end mileage of the marking segments marked as solid lines in the lane marking type data and summarizing them into the section boundary. The lower limit of the lane-changing distance is formed by calculating the lane center line spacing of each adjacent lane pair in the lane-changing section and aligning it with the boundary of the no-lane-change section, then taking the minimum lane-changing length in the section and writing it into the edge attribute. S1-2. Based on the candidate set, calculate the lateral offset from the positioning point to the center line of the corresponding candidate lane, the angle difference between the driving direction and the direction of the corresponding candidate lane for each candidate lane, and calculate the forward projection point corresponding to the preset time step based on the speed and driving direction. Then, for each candidate lane, determine whether the lateral offset meets the offset upper limit, whether the angle difference meets the angle difference upper limit, and whether the forward projection point falls within the boundary range of the corresponding candidate lane. Based on the determination results, a lane score set is formed, which is used to output a lane-level guidance prompt benchmark consistent with the current lane of the vehicle during human-computer interaction. The preset time step refers to the fixed time interval for converting the current speed and driving direction of the vehicle into forward projection displacement. It is generated by setting it in the configuration parameters according to the Beidou positioning update cycle and the minimum response time required for lane determination and reading and calling it at runtime. S1-3. Select the target lane from the lane score set that simultaneously meets the upper limit of lateral offset, the upper limit of included angle difference, and the forward projection point falls within the lane boundary range, and determine the road segment to which the target lane belongs as the target road segment, and generate the vehicle state, which includes the vehicle's current road segment and the vehicle's current lane.

[0006] In a preferred embodiment, S2 includes: S2-1. Obtain the current guidance code displayed by the target VMS and retrieve the code table record that matches the guidance code from the pre-stored code table; the pre-stored code table refers to the lookup table that establishes a one-to-one mapping between the VMS guidance code and the diversion point identifier, action type and action parameters and stores them. The generation method is to first assign a fixed-length diversion point identifier code to each diversion point and a fixed-length action type code to each type of action, and then assign a fixed bit length to the action sequence number, trigger mileage offset, trigger threshold rule identifier and constraint identifier and agree on the splicing order. The above fields are spliced ​​together according to the fixed length and splicing order to generate the guidance code and write it into the code table library; S2-2. Read the diversion point identifier field, action type field, and action parameter field from the code table record, and extract the action sequence number field, trigger mileage offset field, trigger threshold rule identifier field, and constraint identifier field from the action parameter field according to the preset field segments in the code table record to form an action parameter set, which is used to generate prompt fields for trigger position and trigger threshold for subsequent actions during human-computer interaction. S2-3. Generate the target diversion point based on the diversion point identifier field, generate the action type based on the action type field, and write the target diversion point, action type and action parameter set into the code state.

[0007] In a preferred embodiment, S3 includes: S3-1. Based on the set of branch points and branch point identifiers in the pre-stored road topology, associate each branch point with its branch point identifier to construct a node set and output the node set; S3-2. Based on the lane connection relationships in the pre-stored road topology, determine that the connection direction of each lane connection relationship is consistent with the driving direction in the vehicle state and that the connection interval does not fall within the boundary coverage of the lane change restriction interval. When the determination is satisfied, write the corresponding starting divergence point identifier and ending divergence point identifier into the edge set. The starting divergence point identifier and ending divergence point identifier are the numbers of the starting divergence point and the ending divergence point of the lane connection relationship in the unified coordinate system, respectively. The generation method is to extract the starting lane centerline endpoint and the ending lane centerline endpoint of each lane connection relationship when constructing the pre-stored road topology, match them to the nearest divergence point node, and write them into the corresponding divergence point identifier field. S3-3. Write the starting and ending divergence point identifiers of the edge set into the edge set according to the lower limit of the lane change distance and the boundary of the lane change prohibited section to form a divergence anchor point map; S3-4. Based on the current road segment and current lane in the vehicle state, determine the starting node associated with the current lane in the diversion anchor point graph, and use the starting node as the initial set. Perform layer-by-layer expansion on the unvisited nodes according to the connection direction of the edge set. During expansion, only add the endpoint node corresponding to the edge that satisfies the lower limit constraint of lane change distance and does not cross the boundary of the no-lane-change section to the initial set. After no new nodes are added, summarize the node sequence from the starting node to each endpoint node as the reachable anchor point path set, which is used to limit the reachable diversion range of the prompt during human-computer interaction and avoid prompting to cross the boundary of the no-lane-change section. S3-5. Write the branch anchor point map and the set of reachable anchor point paths into the data field of the anchor point state to form the anchor point state.

[0008] In a preferred embodiment, the process of forming the diversion anchor point map in S3 further includes: S3-31. Based on the pre-stored road topology edge set, for each edge, retrieve the corresponding lane pair identifier and connection mileage segment according to the starting point divergence point identifier and the ending point divergence point identifier, and read the lane centerline spacing sequence and solid line marking mileage segment sequence of the corresponding lane pair within the connection mileage segment, and output the edge attribute candidate set; where the lane pair identifier is used as the field of lane connection relationship; and the connection mileage segment is used as the mileage field of lane connection relationship. S3-32. Based on the edge attribute candidate set, generate candidate boundary of no lane change interval for each edge and candidate lower limit of lane change distance. The candidate boundary of no lane change interval is formed by merging and deduplicating the sequence of solid line marking mileage segments and satisfies the monotonically increasing constraint of interval boundary. The candidate lower limit of lane change distance is formed by taking the lower limit value after aligning the sequence of lane centerline spacing by mileage and satisfies the non-negative constraint. Output the candidate attribute set. S3-33. Perform consistency checks and conflict resolution on the candidate attribute set to form a constraint set. The constraint set includes the endpoint constraint of the connecting mileage segment where the boundary of the no-lane-change section must not cover the edge, the unique constraint of the no-lane-change section boundary corresponding to the same starting point divergence point identifier and the same ending divergence point identifier, and the unique constraint of the lower limit of the lane-changing distance corresponding to the same starting point divergence point identifier and the same ending divergence point identifier. Generate a conflict set for candidate attribute records that violate the constraints, and output the constraint set and the conflict set. Among them, the endpoint constraint of the connecting mileage segment where the boundary of the no-lane-change section must not cover the edge refers to the constraint that for each edge in the edge set, the endpoint constraint of the connecting mileage segment is determined by the connecting mileage segment of that edge. The starting and ending mileages of each segment are determined. For each lane-change restriction candidate corresponding to that segment, the starting and ending mileages are determined to be neither equal to the starting mileage nor equal to the ending mileage. At the same time, it is determined that the lane-change restriction does not contain the starting mileage and does not contain the ending mileage. If there exists any lane-change restriction that satisfies the condition that the starting mileage is less than or equal to the starting mileage and the ending mileage is greater than or equal to the starting mileage, or the starting mileage is less than or equal to the ending mileage and the ending mileage is greater than or equal to the ending mileage, then the candidate is determined to violate the endpoint non-coverage constraint and the candidate record is written into the conflict set. The unique constraint of the no-lane-change interval boundary corresponding to the same starting point divergence point identifier and ending point divergence point identifier means that for all edges in the edge set that have the same starting point divergence point identifier and ending point divergence point identifier, their no-lane-change interval boundary candidates are summarized and boundary signatures are formed according to the number of intervals and the start and end mileage of each no-lane-change interval. It is determined that only one boundary signature is allowed as the write value under the starting point divergence point identifier and ending point divergence point identifier. If two or more different boundary signatures satisfy the endpoint non-overlap constraint at the same time, it is determined that the unique constraint is violated and the candidate records corresponding to different boundary signatures are written into the conflict set. Subsequently, through constraint solution, only the no-lane-change interval boundary corresponding to one boundary signature is retained as the write value. The unique constraint on the lower limit of the lane change distance corresponding to the same starting point branch point identifier and ending point branch point identifier means that for all edges in the edge set that have the same starting point branch point identifier and ending point branch point identifier, their candidate lower limit lane change distances are summarized and a set of values ​​is formed for the candidate values. It is determined that only one lower limit lane change distance value is allowed to be written as the value under the given starting point branch point identifier and ending point branch point identifier. If there are two or more different values ​​in the value set, it is determined that the unique constraint is violated and the corresponding candidate record is written to the conflict set. Subsequently, through constraint solving, one value is selected based on the sum of cost terms and determined as the lower limit lane change distance corresponding to the given starting point branch point identifier and ending point branch point identifier, and written to the edge set. S3-34. Based on the constraint set and conflict set, perform constraint solving to determine the written value. The constraint solving takes the sum of cost terms as the objective function, and the cost terms are composed of the adjustment amount of the restricted lane change interval boundary and the adjustment amount of the lower limit of the lane change distance. The stopping condition is that the sum of cost terms no longer decreases. Solve the restricted lane change interval boundary and the lower limit of the lane change distance corresponding to each edge. Write the solution results into the edge set according to the starting point diversion point identifier and the ending point diversion point identifier to form a diversion anchor point map and output it.

[0009] In a preferred embodiment, S4 includes: S4-1. For each path in the reachable anchor point path set, accumulate the edge mileage along the branch anchor point graph and form the path mileage from the vehicle's current position to the end of each path, and output the path mileage set; where the edge mileage is used as an edge attribute field or calculated from the connection mileage segment; S4-2. Calculate the arrival time for each item in the path mileage set based on speed, and align the arrival time with the preset time step to form an executable time window, and output the time window parameters. S4-3. For each path in the reachable anchor point path set, count the number of lane-changing edges contained in the path, and for each lane-changing edge, read its lower limit of lane-changing distance and the boundary of the lane-changing prohibited interval. Delete the path that crosses the boundary of the lane-changing prohibited interval, and summarize the number of lane-changing edges of the remaining paths as the number of lane-changing operations that can be performed. Output the lane-changing parameters. S4-4. Based on the time window parameters and lane change parameters, extract the trigger mileage interval that satisfies the executable lane change number constraint from the path mileage set to form the executable trigger mileage. Write the executable time window, executable lane change number, and executable trigger mileage into the window state to generate a window cone. The lane change number constraint refers to the upper limit of the number of lane changes that a vehicle is allowed to complete from the current lane to the target diversion point within the executable time window. The generation method is to calculate the available driving mileage based on the vehicle speed and the preset time step, and then accumulate the lower limit of the lane change distance of each lane change edge in the diversion anchor point map side by side to obtain the number of lane change edges that can be covered within the available driving mileage, and determine this number as the upper limit.

[0010] In a preferred embodiment, S5 includes: S5-1. Obtain the action type and action parameter set in the code state, and retrieve the action template corresponding to the action type based on the pre-stored code table; S5-2. Based on the action parameter set, the action number field is deduplicated and sorted to form an action number sequence. For each action number in the action number sequence, the corresponding node type is retrieved in the action template. When the node type is hold, a hold action node is generated; when the node type is lane change, a lane change action node is generated; when the node type is diversion, a diversion action node is generated. The action parameters in the action parameter set whose action number field is equal to the action number are written into the generated action node to form an action node sequence. The action node sequence is then identified as the inducement code action grid. S5-3. Based on the current road segment and current lane in the vehicle state, determine the reference mileage within the executable trigger mileage in the window state, and superimpose the trigger mileage offset field in the action parameter set with the reference mileage execution mileage to generate the trigger position of each action node, and output the trigger position set. S5-4. Take values ​​for each trigger threshold rule identifier field in the action parameter set, and retrieve the threshold rule record with the same value from the pre-stored threshold rule table to read the threshold field. Write the threshold field into the action node with the same action sequence number field to form the corresponding trigger threshold, and summarize the trigger thresholds of each action node to form a trigger threshold set. The pre-stored threshold rule table is a lookup table that establishes a one-to-one mapping between trigger threshold rule identifiers and trigger threshold values. The trigger threshold value consists of a threshold type field, a threshold value field, and a threshold unit field. The generation method is to pre-configure the corresponding threshold type for different action types, write the threshold value and threshold unit according to the road level and vehicle speed range to form a rule record and put it into the database. S5-5. Perform intersection filtering on the action node sequence in the induced code action grid. The intersection filtering uses the executable time window, executable lane change number and executable trigger mileage in the window state as constraints and deletes action nodes that do not meet the constraints. Perform time sequence rearrangement on the remaining action nodes according to the trigger position to form path instructions.

[0011] In a preferred embodiment, the process of determining the induced code action grid in S5 further includes: S5-21. Perform deduplication and sorting on the action number field in the action parameter set to form an action number sequence. Then, for each action number sequence, retrieve the node type field in the action template to form a node type sequence and output the candidate sequence. S5-22. Construct a constraint set based on the candidate sequence and perform constraint solving to reconstruct the action number sequence. The constraint set includes a one-to-one correspondence constraint between the action number field and the node type field, a consistency constraint between the node type sequence and the action template, a complete constraint on the action parameter field, and a conflict constraint on the constraint identifier field. The constraint solving takes the sum of cost terms as the objective and the cost terms no longer decreasing as the stopping condition, and outputs the reconstructed sequence. The one-to-one correspondence constraint between the action number field and the node type field means that for each action number in the action number sequence, the node type field is retrieved in the action template based on the action number, and the retrieval result is required to be a unique record. At the same time, the node type field corresponding to the same action number in the candidate sequence is counted and the count is required to be one. If the action template is empty, multiple records are retrieved, or the same action number in the candidate sequence corresponds to multiple node type fields, it is determined that the action number violates the one-to-one correspondence constraint, which is used to exclude the case where the mapping between the number and the node type cannot be determined. The consistency constraint between the node type sequence and the action template refers to generating a node type sequence in the order of the action number sequence, and reading the template node type sequence in the order of the same action number, and performing an equality judgment item by item; if the candidate node type field corresponding to any action number is not equal to the template node type field, it is determined that the consistency constraint is violated. This is used to ensure that the arrangement of node types in the action grid is consistent with the action structure defined in the action template. The action parameter field integrity constraint refers to determining that for records in the action parameter set where the action sequence number field is equal to the current action sequence number, the trigger mileage offset field, trigger threshold rule identifier field, and constraint identifier field all have values, and that the values ​​of each field satisfy the corresponding field segment length constraint in the pre-stored code table; if any field is missing or any field exceeds the segment length, it is determined that the integrity constraint is violated, which is used to ensure that the action parameters required for the subsequent generation of trigger positions and trigger thresholds have complete fields that can be decoded and executed; Conflict constraints in constraint identifier fields refer to parsing the constraint identifier field corresponding to the same action sequence number to obtain a set of constraint identifiers, and performing conflict determination with the node type field corresponding to the action sequence number. It requires that the action node does not contain a lane-changing constraint identifier, the lane-changing action node does not contain a hold constraint identifier, and the diversion action node does not contain a diversion prohibition constraint identifier. At the same time, it performs mutual exclusion determination within the constraint identifier set and requires that no mutually exclusive identifier pairs appear simultaneously. If any of the above conflicts or mutual exclusions occur, it is determined that the conflict constraint is violated, which is used to eliminate self-contradictory configurations between the action node type and the constraint identifier. S5-23. Perform consistency verification and error correction on the action parameter set based on the reconstruction sequence. The consistency verification includes performing field bit field verification on the action parameter field corresponding to each action number and performing back-off replacement on the action number that does not meet the field bit field verification. The back-off replacement uses the decrease of the sum of cost terms as the update criterion and the absence of back-off replacement as the stopping condition, and outputs the correction sequence. S5-24. Generate hold action nodes, lane change action nodes and diversion action nodes that are consistent with the node type field according to the correction sequence, and write the action parameters whose action sequence number field is equal to the corresponding action sequence number into the generated action nodes to form an action node sequence. Determine the action node sequence as the inducement code action grid and output it.

[0012] In a preferred embodiment, S6 includes: S6-1. Within a preset time step, aggregate multiple BeiDou positioning and calculate the position residual sequence. Solve the position confidence based on the position residual sequence and write the position confidence into the vehicle state to update the vehicle state. S6-2. Based on the pre-stored road topology, determine whether the BeiDou positioning falls within the lane boundary range of the current lane and whether the angle difference between the driving direction and the current lane direction is less than or equal to the angle difference threshold as a consistency verification condition. If the verification fails, search the candidate lane set in the BeiDou positioning neighborhood and calculate the lateral offset and angle difference for each candidate lane. Add the vehicle state corresponding to the candidate lane with the lateral offset less than or equal to the offset threshold and the angle difference less than or equal to the angle difference threshold to the candidate vehicle state set. Select the vehicle state corresponding to the first position of the position confidence ranking in the candidate vehicle state set as the target vehicle state to update the vehicle state and output the target vehicle state. S6-3. Calculate the remaining mileage from the current position to the trigger position for each action node according to the action node sequence of the path instruction. Use the remaining mileage being less than or equal to the trigger mileage threshold and the speed in the vehicle state meeting the speed threshold as the joint satisfaction judgment condition. When the condition is met, mark the action node as triggered and record the Beidou positioning and action sequence number fields at the trigger time. Summarize the marking results of each action node to form an action execution state set and output it. This is used to trigger the prompt output of the corresponding action node during human-computer interaction and record the prompt trigger basis. S6-4. Perform conflict resolution on the action execution state set, including sorting multiple action nodes marked as triggered within the same preset time step according to the remaining mileage and retaining only the action node at the top of the sort, marking the remaining action nodes as not triggered and recording the conflict reasons, and outputting the resolved action execution state set. S6-5. Determine the completion status of the centralized and diverted action nodes after the resolution of the action execution status, or determine if the executable time window has reached its upper limit. If so, generate and output the execution result at one time. The execution result includes a completion flag and a non-completion reason flag.

[0013] A driver route selection guidance system based on VMS guidance information includes a vehicle status positioning module, a guidance code decoding module, an anchor point map routing module, a window cone generation module, a route instruction generation module, and an execution verification output module. The vehicle status positioning module is used to obtain the vehicle's BeiDou positioning, driving direction and speed, and match it with the pre-stored road topology to solve the vehicle's current road segment and current lane, and output the vehicle status. The inducement code decoding module is used to read the inducement code displayed by the target VMS, and decode the inducement code into the target diversion point, action type and action parameters according to the pre-stored code table, and output the code state; The anchor point graph pathfinding module constructs a diversion anchor point graph based on the pre-stored road topology. The diversion anchor point graph uses diversion point identifiers as nodes and lane reachability relationships as edges. It records the lower limit of lane change distance and the boundary of lane change prohibited interval on the edges, and then maps the vehicle state to the diversion anchor point graph and solves the reachable anchor point path set, outputting the anchor point state. The window cone generation module calculates the window cone based on the vehicle state and the anchor point state. The window cone consists of an executable time window, an executable lane change count, and an executable trigger mileage, and outputs the window state. The path instruction generation module is based on the code state query of the inducement code action grid. The inducement code action grid consists of hold action nodes, lane change action nodes and diversion action nodes. Each action node is configured with a trigger position and a trigger threshold. Then, the inducement code action grid and window state are subjected to intersection filtering and time sequence rearrangement to generate a set of action trigger positions and a set of trigger thresholds and form a path instruction. The execution verification output module is used to periodically acquire BeiDou positioning during vehicle operation and update vehicle status accordingly. It performs satisfaction judgment on the action triggering position and triggering threshold in the path instruction one by one, outputs the action execution status and forms the execution result.

[0014] The technical effects and advantages of this invention are as follows: 1. This solution decodes the VMS guidance code into a set of action parameters and performs intersection filtering and temporal rearrangement on the guidance code action grid within the diversion anchor point map and window cone constraint domain, so that the common guidance information can converge into a triggerable path instruction corresponding to the current position of the vehicle, thereby relatively mitigating the failure of adoption caused by correct prompts but unexecutable ones; 2. Using the diversion point identifier as a node, write the boundary of the lane-no-lane-change section and the lower limit of the lane-change distance on the lane reachable edge. Combine this with the reachable traversal to output the reachable anchor point path set, so that the subsequent instruction generation has lane-level prohibition and restriction constraints, thereby relatively reducing the probability of generating an unexecutable lane-change sequence. 3. Calculate the executable time window based on the reachable anchor point path set and vehicle speed, and determine the number of executable lane changes and the executable trigger mileage by accumulating the lower limit of lane change distance side by side, forming a window cone constraint domain, so that the action trigger position and trigger threshold can be configured within the executable domain, thereby relatively improving the situation of missing the critical operation window; 4. The guidance code is split into action sequence number, trigger mileage offset, trigger threshold rule identifier and constraint identifier by adopting the field segment rules of the pre-stored code table, and the guidance code action grid is constructed accordingly, so that the VMS low display resources can carry the calculable action structure, thereby relatively improving the driver's certainty of restoring guidance information. 5. During the driving process, the vehicle's BeiDou positioning is periodically acquired and the vehicle status is updated. The path instructions are executed node by node, trigger judgment is performed, and conflict resolution is carried out to form a closed loop of action execution status and execution result. This keeps the instruction triggering aligned with the actual driving progress of the vehicle, thereby relatively suppressing the execution deviation caused by positioning drift or multiple action competition. Attached Figure Description

[0015] Figure 1 This is a flowchart outlining the method steps of the present invention; Figure 2 This is a schematic diagram of the system module structure of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Refer to the instruction manual appendix Figure 1-2 The present invention provides a driver route selection guidance method based on VMS guidance information, comprising: S1. Obtain the vehicle's BeiDou positioning, driving direction and speed, match it with the pre-stored road topology, solve the vehicle's current road segment and current lane, and output the vehicle status; This embodiment illustrates how, before a vehicle enters the variable message sign guidance coverage area, lane-level positioning and vehicle state generation are completed based on BeiDou positioning and pre-stored road topology, thereby providing directly readable current road segments and lanes for subsequent diversion anchor point mapping and window cone calculation. The principle is to project BeiDou positioning onto the road segment and lane space of the road topology to form a candidate set, and then perform geometric consistency and motion consistency calculations on the candidate set to obtain a uniquely determinable vehicle state. This implementation process includes the following steps: By acquiring the vehicle's current BeiDou positioning, driving direction, and speed, and retrieving pre-stored road topology, candidate road segment sets and candidate lane sets are formed, thus providing a calculable candidate space for subsequent consistency determination. Inputs include BeiDou positioning, driving direction, and speed. BeiDou positioning is represented using latitude and longitude or projected coordinates in a unified coordinate system, driving direction is represented using azimuth, and speed is represented by distance traveled per unit time. Processing involves performing a neighborhood search in the topology library centered on the BeiDou positioning to obtain a candidate road segment set, and then searching within the lane set covered by the candidate road segment set to obtain a candidate lane set, forming a candidate set. The neighborhood radius is given by a preset configuration and is consistent with the horizontal accuracy of the BeiDou positioning. Pre-stored road topology is written into the topology library and used as structured data storage. The topology library includes a set of branching points and branching point identifiers, lane sets and lane boundary lines and lane center lines, lane connection relationships, no-lane-change zone boundaries, and lower limits for lane-changing distances. The branching point set is determined by performing bifurcation point identification on the road centerline network. Ramp connection points are identified and each branch point is assigned a branch point identifier according to its position sequence. Lane boundary lines and lane center lines are formed by connecting lane marking point sets obtained from surveying according to the sampling order after coordinate calibration. Lane connection relationships are formed by performing connectivity determination on the lane center line endpoints at the endpoints of adjacent road segments and writing the endpoint pairs that meet the connectivity requirements into directed connection records. The boundary of the no-lane-change section is formed by extracting the start and end mileages of the marking segments marked as solid lines in the lane marking type data and summarizing them into the section boundary. The lower limit of lane-changing distance is formed by calculating the lane center line spacing of each adjacent lane pair within the lane-changing section, aligning it with the boundary of the no-lane-change section, and then taking the minimum lane-changing length within the section and writing it into the edge attribute. The output is a candidate set, which is written into the subsequent lane consistency calculation process for reading. Anomaly handling includes increasing the neighborhood radius by a preset expansion step size to the upper limit of the radius and then re-searching when the neighborhood is still empty, outputting a missing identifier and backtracking the output of road segment-level candidate results to ensure that subsequent steps can continue to be executed. The system determines the target lane and road segment by calculating the lateral offset, angle difference, and forward projection point on the candidate set and performing threshold judgment, thereby generating the vehicle status and writing it into the subsequent process. Inputs include the candidate lane set, BeiDou positioning, driving direction and speed, as well as a preset time step, offset upper limit, and angle difference upper limit. The preset time step is given by configuration parameters, and its value is determined based on the BeiDou positioning update cycle and the lower limit of the response time required for lane determination, and is read and called at runtime. The offset upper limit and angle difference upper limit are given by a preset rule table and retrieved according to road level or road segment type. Processing actions include calculating the lateral offset from the BeiDou positioning point to the candidate lane centerline for each candidate lane. The lateral offset is determined by searching for the first value of the distance from the point to the line segment on the centerline. The system also calculates the angle difference between the driving direction and the candidate lane direction. The candidate lane direction is determined by taking the tangential direction of the centerline at the projection point. Finally, the system calculates the forward projection point corresponding to the preset time step based on the speed and driving direction. The forward projection point is obtained by multiplying the speed by the preset time step. The forward displacement is superimposed on the BeiDou positioning according to the driving direction. Then, for each candidate lane, it is determined that the lateral offset is less than or equal to the offset upper limit, the angle difference is less than or equal to the angle difference upper limit, and the forward projection point falls within the candidate lane boundary range. The falling within the boundary is determined by executing the point within the polygon formed by the forward projection point and the lane boundary line. Based on this, a lane score set is formed, which includes at least candidate lane identifiers and passing identifiers. When there is only one candidate lane with the passing identifier, it is determined as the target lane. When there are more than one candidate lane with the passing identifier, it is sorted in ascending order by lateral offset and ascending order by angle difference, and the first in the sort order is selected as the target lane. When there are no candidate lanes with the passing identifier, the candidate lane corresponding to the first in the lateral offset order is selected as the downgraded target lane and the downgrade reason is recorded. The output includes vehicle status. The vehicle status is formed by determining the road segment to which the target lane belongs as the target road segment and writing it into the vehicle's current road segment field and vehicle's current lane field. The vehicle status is written into the subsequent diversion anchor point mapping process for reading. Through the above implementation process, even with complex BeiDou positioning errors and lane morphology, verifiable lane-level positioning results can be formed using offset upper limits, angle difference upper limits, and boundary fall-in judgments. Neighborhood expansion and downgrade selection mechanisms ensure that vehicle status can still be output even when candidates are empty or multiple solutions exist. This improves the input stability of subsequent calculations based on the diversion anchor point map and window cone and reduces the risk of false triggering during the guidance execution phase. In practical applications: when a vehicle approaches a ramp divergence point on an urban expressway, the onboard terminal periodically acquires BeiDou positioning and speed, searches the topology library for neighboring lanes to form a candidate set, and then, under the constraint of the forward projection point corresponding to the preset time step, completes offset and angle difference judgments and determines the target lane and target road segment. The vehicle status is then provided to subsequent diversion anchor point map traversal and window cone calculation, ensuring that the trigger positions for lane changing and diversion actions generated by the VMS guidance code are consistent with the vehicle's actual lane. This avoids generating unexecutable lane changing prompts within road segments where lane changing is prohibited and ensures stable output of path instructions.

[0018] S2. Read the inducement code displayed by the target VMS, and decode the inducement code into the target diversion point, action type and action parameters according to the pre-stored code table, and output the code state; This embodiment illustrates how, after a vehicle receives guidance information from a variable message sign (VMS), the currently displayed content of the target VMS is converted into a computable code state. This provides a directly readable set of target diversion points, action types, and action parameters for subsequent guidance code action grid generation, trigger position calculation, and threshold retrieval. The principle involves normalizing the VMS display content into guidance codes and performing a one-to-one mapping retrieval in a pre-stored code table. Then, the action parameter field is truncated according to preset field segments to obtain the action parameter set. Finally, the diversion point identifier field and the action type field are determined as the target diversion point and action type, respectively, and written into the code state. This implementation process includes the following steps: By acquiring the guidance code currently displayed by the target VMS and retrieving a matching record from the pre-stored code table, the guidance information is determined to be recorded in the code table, thus providing a definite data source for subsequent bit segment truncation and code state writing. Inputs include the current display content of the target VMS and the display time identifier. The current display content is identified by the vehicle terminal through the vehicle camera or received by the vehicle-to-everything (V2X) interface and uniformly converted into a guidance code string as the search key. The display time identifier is generated by the vehicle terminal's current clock to prevent cross-frame misuse. Processing actions include performing integrity checks on the guidance code and retrieving a matching code table record from the pre-stored code table. Integrity checks include determining whether the guidance code length meets the preset code length constraint and whether the guidance code character set meets the preset character set constraint. The preset code length constraint and preset character set constraint are given by the code table configuration file and read during runtime. The pre-stored code table establishes a one-to-one relationship between the VMS guidance code, the diversion point identifier, the action type, and the action parameters. The mapped and stored lookup table, with code table records including at least the inducement code field, the diversion point identifier field, the action type field, and the action parameter field, is generated by first assigning a fixed-length diversion point identifier code to each diversion point and a fixed-length action type code to each type of action. The action sequence number, trigger mileage offset, trigger threshold rule identifier, and constraint identifier are then assigned fixed bit lengths and their concatenation order is agreed upon. The above fields are then concatenated according to the fixed length and concatenation order to generate the inducement code and write it into the code table library. The fixed length and concatenation order are preset by the code table configuration file and solidified during the code table generation stage. The output is the code table record, which is written into the subsequent bit segment truncation process for reading. Anomaly handling includes outputting an invalid code identifier and triggering a re-acquisition of the display content when the inducement code integrity check fails; and outputting a miss identifier and triggering a fallback to a degradation path that generates a code state based on the default template of the action type when no matching code table record is found, to ensure that subsequent processes can continue to execute. By extracting action parameter fields from code table records according to preset field segments and writing them into the code state, the determination of the target diversion point, action type, and action parameter set from the code table records is achieved, thus forming a code state that can be directly read by the subsequent action grid generation process. The input includes code table records and field segment rules, where the field segment rules are given by the code table configuration file and include at least the start position, length, and order information of the action parameter fields, and are retrieved at runtime by the code table version number. The processing actions include reading the diversion point identifier field, action type field, and action parameter field from the code table records, and performing segment extraction on the action parameter fields according to the field segment rules to obtain the action sequence number field, trigger mileage offset field, trigger threshold rule identifier field, and constraint identifier field. The segment extraction method is to perform substring extraction on the action parameter fields according to the preset start position and preset length, or to perform mask extraction on the bit position, and write each extraction result into the action parameter set according to the field name. Subsequently, the action parameter set is subjected to field integrity verification, which includes verification of the action sequence number field, trigger mileage offset field, trigger threshold rule identifier field, and other fields. Both the segment and constraint identifier fields have value determination criteria and criteria for each field value to satisfy its fixed length constraint, which is provided by the field segment rules. Then, a target tributary point is generated based on the tributary point identifier field. The generation method is to use the tributary point identifier field as the index value of the target tributary point and use it to match the corresponding tributary point in the pre-stored road topology. An action type is generated based on the action type field. The generation method is to use the action type field as the enumeration value of the action type and use it to retrieve the action template later. The output is a code state, which is formed by writing the target tributary point, action type, and action parameter set into the data field of the code state. The code state is then written into the subsequent guidance code action grid generation process for reading. Anomaly handling includes recording the corresponding action sequence number field as invalid and deleting it from the action parameter set when the field integrity check fails, while recording the reason for invalidity for subsequent constraint solving and reconstruction. When the tributary point identifier field cannot match a tributary point in the pre-stored road topology, a tributary point missing identifier is output and a fallback is triggered to generate a degraded path based on the nearest tributary point of the current road segment to ensure that subsequent action grids can still be generated. Through the above implementation process, the VMS display content can be converted into a structured code state through deterministic retrieval and bit segment extraction. This ensures that the target diversion point, action type, and action parameter set have clear sources, clear value calibers, and verifiable integrity constraints. This provides stable input for subsequent action grid construction, trigger position calculation, and threshold retrieval, and reduces the risk of link interruption caused by display recognition errors or code table misses. In practical applications: when a vehicle approaches a diversion ramp on the elevated mainline and the VMS displays diversion guidance information, the on-board terminal normalizes the recognized display content into a guidance code and retrieves the code table record from the pre-stored code table. Then, according to the field bit segment rules, it extracts fields such as the action sequence number and trigger mileage offset from the action parameter field to form an action parameter set. The diversion point identifier field is written into the code state as the target diversion point, and the action type field is written into the code state as the action type. This enables the generation of trigger positions and trigger thresholds for lane changing and diversion actions under window cone constraints, and the formation of executable path instructions.

[0019] S3. Construct a diversion anchor point graph based on the pre-stored road topology. The diversion anchor point graph uses diversion point identifiers as nodes and lane reachability relationships as edges. Record the lower limit of lane change distance and the boundary of the lane change restriction interval on the edges. Then, map the vehicle state to the diversion anchor point graph and solve the reachable anchor point path set, and output the anchor point state. This embodiment illustrates how, after obtaining the vehicle status, a diversion anchor point graph is constructed based on pre-stored road topology, and a set of reachable anchor point paths is solved, thereby forming an anchor point status that can be directly read by window cone calculation and guidance code action grid filtering. The principle is to abstract diversion point identifiers as nodes, lane reachability relationships as edges, and write the boundaries of no-lane-change sections and the lower limit of lane-change distances into the edge attributes to constrain reachability traversal, ensuring that the set of reachable anchor point paths only contains executable cross-lane routes. This implementation process includes the following steps: By constructing a node set from the shunting point set and shunting point identifiers, and filtering lane connection relationships into an edge set, the skeleton of the shunting anchor point graph is generated, thus providing indexable nodes and edges for subsequent writing of edge attributes and reachability traversal. Inputs include pre-stored road topology and vehicle state. The pre-stored road topology includes at least the shunting point set, shunting point identifiers, lane connection relationships, and no-lane-change zone boundaries. The vehicle state includes at least the vehicle's current road segment, current lane, and driving direction. Processing actions include reading the shunting point identifier of each shunting point in the shunting point set and writing it into a node record to form a node set. The node record includes at least the shunting point identifier and shunting point coordinates for subsequent nearest neighbor matching. Subsequently, the connection direction field and connection mileage segment field of each lane connection relationship are read one by one, and a direction consistency determination is performed between the connection direction field and the driving direction in the vehicle state. The direction consistency determination is obtained by performing an angle difference threshold determination on the azimuth difference between the connection direction field and the driving direction. The angle difference threshold is determined by pre-stored... The configuration is given and searched by road level; then, a coverage determination is performed on the connection mileage segment field and the boundary of the no-lane-change section. The coverage determination is obtained by determining whether the connection mileage segment has an interval intersection with any no-lane-change section boundary; when the direction consistency determination passes and the coverage determination fails, the starting lane centerline endpoint and the ending lane centerline endpoint of the lane connection relationship are extracted, and the nearest divergence point node is searched in the node set to obtain the starting divergence point identifier and the ending divergence point identifier. The nearest neighbor search is obtained by calculating the Euclidean distance between the endpoint coordinates and the divergence point coordinates and selecting the node with the highest distance order. The starting divergence point identifier and the ending divergence point identifier are written into the edge record to form an edge set; the output includes the node set and the edge set, and the node set and the edge set are written into the subsequent edge attribute writing process for reading; the exception handling includes marking the lane connection relationship as invalid and removing it from the edge set when the endpoint nearest neighbor search does not hit the divergence point node, and recording the reason for invalidity for subsequent topology library verification; By writing the lower limit of lane-change distance and the boundary of the no-lane-change interval into the edge set and performing constrained reachability traversal on the graph, the reachable anchor point path set is solved and the anchor point state is formed, thus providing an executable path set for subsequent window cone calculation. The input includes the node set, edge set, the lower limit of lane-change distance and the boundary of the no-lane-change interval in the pre-stored road topology, and the current road segment and current lane of the vehicle in the vehicle state. The processing actions include retrieving the corresponding lower limit of lane-change distance and the boundary of the no-lane-change interval according to the starting point divergence point identifier and the ending point divergence point identifier of the edge set, and writing them into the edge record to form the divergence anchor point graph. When writing, the lower limit of lane-change distance is written. The incoming edge attribute field is used to write the boundary of the no-lane-change section into the edge attribute field for subsequent determination. Then, the starting node associated with the vehicle's current lane is determined in the directional anchor point graph. The starting node is determined by reading the lane centerline endpoint of the vehicle's current lane from the pre-stored road topology and retrieving the nearest directional point identifier from the node set. An initial set is then formed using the starting node and layer-by-layer expansion is performed. Layer-by-layer expansion includes retrieving the outgoing edge sets of nodes in the initial set that have not yet been expanded, and performing reachability determination on each outgoing edge. Reachability determination includes lane-change distance lower limit constraint determination and no-lane-change section boundary crossing determination, where lane-change... The distance lower limit constraint is determined by performing a lower limit satisfaction check on the length of the connecting mileage segment of the edge and the lower limit of the lane change distance of the edge. The lane change restriction interval boundary crossing determination is determined by checking whether there is an interval intersection between the connecting mileage segment of the edge and the lane change restriction interval boundary of the edge. When both types of determinations pass, the endpoint branch point identifier of the outgoing edge is added to the initial set and the outgoing edge is written to the predecessor record for path backtracking. After completing one expansion of the current layer node, if the expansion does not generate a new endpoint branch point identifier, the expansion is terminated. After termination, backtracking is performed on the predecessor record to summarize the path from the starting node to each endpoint node. The node sequence is written into the reachable anchor path set in the order of the branch point identifiers. The output includes a branch anchor graph, a reachable anchor path set, and an anchor state. The anchor state is formed by writing the branch anchor graph and the reachable anchor path set into the data field of the anchor state, and the anchor state is written into the subsequent window cone calculation process for reading. Anomaly handling includes: when the starting node search fails, the upstream adjacent branch point identifier of the vehicle's current lane is used as the replacement starting node and the replacement reason is recorded; when the reachable anchor path set is empty, an empty path identifier is output and the subsequent window cone calculation is triggered to enter the degraded path that only retains the hold action node. By retrieving lane pair identifiers and connecting mileage segments from each edge of the edge set and generating a candidate set of edge attributes, the computable generation of candidate boundary values ​​for no-lane-change zones and candidate lower limits for lane-change distances is achieved, thus providing candidate inputs for subsequent consistency checks and constraint solutions. The inputs include pre-stored road topology, lane pair identifier fields, connecting mileage segment fields, lane centerline spacing sequences, and solid line marking mileage segment sequences from the edge set and topology library. The lane pair identifier, as a field representing lane connection relationships, is used to index adjacent lane pairs, and the connecting mileage segment, as a mileage field representing lane connection relationships, is used to limit the data range. The processing actions include processing each edge in the edge set... The system reads the starting and ending divergence point identifiers from the edge, and retrieves the corresponding lane connection relationship records from the pre-stored road topology to read the lane pair identifiers and connection mileage segments. Then, based on the lane pair identifiers, it reads the lane centerline spacing sequence of the lane pair from the topology library, and performs interval truncation on the lane centerline spacing sequence based on the connection mileage segments to obtain the spacing subsequences within the connection mileage segments. The interval truncation is obtained by filtering the spacing sequence based on the starting and ending mileage axes. Finally, based on the lane pair identifiers, it reads the solid line marking mileage segment sequence, and based on the connection mileage segments... The process involves: 1) Pruning the solid line marking mileage segment sequence to obtain subsequences of solid line marking mileage segments within connected mileage segments; 2) Writing the spacing subsequence and the solid line marking mileage segment subsequence into the edge attribute candidate set to form candidate records for each edge; 3) Merging and deduplication of the solid line marking mileage segment subsequence for each edge to generate candidate boundaries for lane-change prohibited sections. Merging and deduplication is achieved by performing interval union on adjacent or overlapping intervals according to their start and end mileages, and the generated intervals are sorted in ascending order by their starting mileage to satisfy the monotonically increasing constraint of the interval boundaries; 4) Aligning the spacing subsequence of each edge by mileage to generate candidate lower limits for lane-change distances. Mileage alignment is achieved through... An aligned sequence is obtained by resampling the sampling mileage of the spacing subsequence at equal intervals. The candidate lower limit of lane change distance is obtained by calculating the lower limit value of the spacing value of the aligned sequence and performing a non-negativity check. If the non-negativity check fails, the candidate is marked as invalid. The output includes a candidate set of edge attributes and a candidate attribute set. The candidate attribute set is written into the subsequent consistency check and constraint solution process for reading. Anomaly handling includes marking the edge as invalid and removing it from the candidate attribute set when the lane connection relationship retrieval is not found, and marking the candidate lower limit of lane change distance as missing and recording the reason for the missing for subsequent constraint solution to replace it. By performing consistency checks and conflict resolution on the candidate attribute set and determining the written values ​​through constraint solving, the unique writing of the no-lane-change interval boundary and the lower limit of the lane-change distance is achieved, thereby ensuring that the edge attributes of the diversion anchor point graph have verifiable consistency under the same pair of starting diversion point identifiers and ending diversion point identifiers. The input includes the candidate attribute set, edge set, and connection mileage segment field, as well as the preset cost term calculation rules and stopping conditions. The cost term calculation rules and stopping conditions are given by the preset configuration and are fixed with the topology library version. The processing actions include reading the connection mileage segment of each edge to determine the starting mileage and ending mileage, and performing endpoint non-coverage judgment on each interval of the no-lane-change interval boundary candidate of the edge. The point non-coverage determination includes situations where the starting mileage and ending mileage of the interval are both not equal to the starting mileage and neither are equal to the ending mileage, and the interval does not contain the starting mileage and does not contain the ending mileage. If a violation exists, the candidate record is written into the conflict set. Subsequently, for edges in the edge set with the same starting shunt point identifier and ending shunt point identifier, their no-lane-change interval boundary candidates are summarized and boundary signatures are generated. The boundary signature consists of the number of intervals and the starting and ending mileages of each interval. A unique constraint determination is then performed. The unique constraint determination is obtained by determining that only one boundary signature is retained as a writing candidate under the given starting shunt point identifier and ending shunt point identifier. If multiple boundary signatures exist, the corresponding candidate record is written into the conflict set. Then, for the same group of identical starting shunt points... The edges of the point-to-point and endpoint-to-point branching point identifiers are aggregated to determine the candidate values ​​for the lower limit of the lane-change distance, and a unique constraint determination is performed. If multiple different values ​​exist, the corresponding candidate record is written into the conflict set. After forming the constraint set and conflict set, constraint solving is performed to determine the written value. The constraint solving uses the sum of cost terms as the objective function and stops when the sum of cost terms no longer decreases. The adjustment amount of the no-lane-change interval boundary is composed of the minimum interval truncation amount where the candidate boundary and the endpoint do not cover the constraint. The adjustment amount of the lower limit of the lane-change distance is composed of the absolute value of the difference between the candidate value and the selected value under the non-negative constraint and the unique constraint. The solution process iteratively updates the candidates for each group of the same starting point branching point identifier and endpoint branching point identifier. Iterative updates include selecting a conflict record in the conflict set and adjusting the corresponding candidate according to the principle of minimum adjustment. If all constraints are satisfied after adjustment, it is written into the solution set; otherwise, the next conflict record is selected until the sum of cost terms no longer decreases. Outputs include the determined boundaries of the lane-change prohibited intervals and the lower limit of the lane-change distance. The solution results are written into the edge set according to the starting point and ending point identifiers to form a flow-change anchor graph, and the flow-change anchor graph is output. Anomaly handling includes marking the edges corresponding to the starting point and ending point identifiers as unusable and removing them from the flow-change anchor graph when the conflict set cannot be resolved to empty through iteration. The reason for unusability is recorded for offline verification and revision by the topology library. Through the above implementation process, a diversion anchor point graph can be constructed in the lane-level road topology using computable candidate generation, verifiable constraint sets, and stopable constraint solving. Under the dual constraints of the no-change zone boundary and the lower limit of the lane-changing distance, a set of reachable anchor point paths can be solved, giving the anchor point state a clear source and consistent boundary. This provides an executable path foundation for subsequent window cone calculation and guidance action selection. In practical applications: when a vehicle approaches the ramp diversion area on the main line of an expressway, the onboard terminal reads the diversion point set and lane connection relationship from the topology library based on the vehicle state to construct a node set and edge set. The no-change zone boundary and the lower limit of the lane-changing distance, obtained offline or online verification, are written into the edge attributes to form a diversion anchor point graph. Subsequently, constrained layer-by-layer expansion is performed using the starting node associated with the vehicle's current lane to output a set of reachable anchor point paths and form an anchor point state. This allows the subsequent window cone to prune non-changeable paths within the executable time window and provide a triggerable diversion path range for the guidance code action grid, thereby avoiding the output of non-executable lane-changing instructions on solid-line no-change road sections and improving the consistency between guidance instructions and road constraints.

[0020] S4. Calculate the window cone based on vehicle state and anchor point state. The window cone consists of an executable time window, executable lane change count, and executable trigger mileage. Output the window state. This embodiment illustrates how, after obtaining the anchor point state, a window cone is calculated based on the reachable anchor point path set and the vehicle state. This provides directly readable executable time windows, executable lane change counts, and executable trigger mileage for subsequent intersection filtering and temporal reordering of the guidance code action grid. The principle is as follows: first, the edge mileage of the reachable anchor point path set is accumulated on the split anchor point graph to form a path mileage set. Then, the path mileage is converted into arrival time by combining the speed in the vehicle state and aligned with a preset time step to form an executable time window. Next, the executable lane change count is determined under path constraints and lane change constraints, and the trigger mileage interval is extracted to generate the window cone. This implementation process includes the following steps: By accumulating edge mileage for each path in the reachable anchor path set and converting the path mileage into arrival time, a time window parameter is formed, thus providing a unified time caliber for subsequent lane change constraints and trigger mileage interception. Inputs include the diversion anchor graph in the anchor state, the reachable anchor path set, and the speed in the vehicle state, as well as a preset time step. The diversion anchor graph includes an edge set and edge attribute fields. The edge attribute fields include at least an edge mileage field or a connection mileage segment field for calculating the edge mileage. The preset time step is given by configuration parameters and determined based on the BeiDou positioning update cycle and the lower limit of the response time generated by the path command. Processing actions include sequentially locating the edges corresponding to adjacent node pairs for each path in the reachable anchor path set according to the node sequence, and reading the edge mileage field of that edge as the cumulative increment. When the edge mileage field is missing, the edge mileage is calculated based on the connection mileage segment field. The calculation method is to subtract the starting mileage from the ending mileage of the connection mileage segment to obtain the edge mileage. The edge mileage of each path is accumulated... The calculation results are written into the path mileage set, which includes at least the path identifier and path mileage. Then, based on the speed in the vehicle state, the arrival time is calculated for each item in the path mileage set. The arrival time is calculated by dividing the path mileage by the speed to obtain the time value and outputting the arrival time. When the speed is missing or equal to zero, the arrival time is marked as missing and the reason for the missing is recorded. The arrival time is then aligned with a preset time step to form an executable time window. The alignment method is to divide the arrival time by the preset time step and round up to obtain the number of time steps, then multiply the number of time steps by the preset time step to obtain the aligned time window length. The aligned time window length is written into the time window parameters. The output includes the path mileage set and the time window parameters, and the time window parameters are written into the subsequent lane-changing parameter calculation process for retrieval. Exception handling includes outputting an empty mileage identifier when the path mileage set is empty and triggering window cone generation to enter a degraded path that only outputs an empty executable time window to ensure that subsequent action grid filtering can still be completed. By statistically analyzing the number of lane-changing edges in the reachable anchor point path set and pruning paths under lane-change restriction constraints, and then deriving the upper limit of lane-changing times from the available driving mileage and extracting the trigger mileage interval, lane-changing parameters and executable trigger mileage are formed. This allows the executable time window, executable lane-changing times, and executable trigger mileage to be written into the window state to generate a window cone. Inputs include the reachable anchor point path set, the lower limit of lane-changing distance and the boundary of the lane-change restriction interval in the diversion anchor point graph, the speed in the vehicle state, time window parameters, and a preset time step. The time window parameters are obtained from preorder alignment and represent the length of the executable time window. Processing actions include reading the node sequence path by path in the reachable anchor point path set and locating each node on the path. For each path, the number of lane-changing edges is determined based on its attribute fields. Lane-changing edge determination is achieved by checking if the lane change flag corresponding to the edge has been changed or by checking if the lower limit of the lane-changing distance field of the edge has a value. The number of lane-changing edges is then counted for each path to form a lane-changing edge count sequence. Next, for each lane-changing edge of each path, the boundary of the no-lane-change interval is read, and it is determined whether the connecting mileage segment of the edge intersects with the boundary of the no-lane-change interval. If an intersection exists, the path is determined to cross the boundary of the no-lane-change interval and is removed from the reachable anchor point path set. The number of lane-changing edges for the remaining paths after deletion is summarized to form lane-changing parameters, which include at least the path flag and the number of lane-changing edges. Finally, based on the vehicle status... The available driving mileage is calculated using speed and time window parameters. The calculation method involves multiplying the speed by the time window parameter to obtain the available driving mileage. Then, in the diversion anchor point map, the lower limit of the lane-change distance for each lane-change edge on the path is read sequentially and accumulated to form a cumulative lane-change distance sequence. The maximum number of lane-change edges that can be covered within the cumulative lane-change distance sequence, not exceeding the available driving mileage, is determined as the upper limit of the lane-change count constraint, and this upper limit is written into the executable lane-change count. Subsequently, based on the time window parameter and the executable lane-change count, a trigger mileage interval is extracted from the path mileage set. The extraction method is to determine that for each path, its path mileage is less than or equal to the available driving mileage and the number of lane-change edges on the path is less than or equal to the executable lane-change count. The system calculates the path mileage that meets the criteria to form a candidate set of trigger mileages. The mileage range of this candidate set is then written into the executable trigger mileage to form a trigger mileage interval. Outputs include lane-changing parameters, the number of executable lane changes, and the executable trigger mileage. The executable time window, the number of executable lane changes, and the executable trigger mileage are written into the data field of the window state to generate a window cone. This window state is then written into the subsequent code action grid filtering process for reading. Exception handling includes outputting an empty path flag when the remaining path is empty, setting the number of executable lane changes to zero, and setting the executable trigger mileage to an empty interval, so that subsequent processes only retain hold action nodes and avoid generating non-executable lane-changing and diversion instructions. Through the above implementation process, the reachable anchor point path set can be converted into window cone parameters under the time alignment and lane change restriction constraints. This provides a unified constraint basis for subsequent action grid selection, including executable time windows, executable lane change counts, and executable trigger mileage. This reduces the risk of stability failure caused by induced actions falling into lane change restriction zones or exceeding available driving mileage, and improves the consistency between path instructions and road constraints. In practical applications: when a vehicle is traveling on the mainline and approaches the divergence area, the onboard terminal reads the reachable anchor point path set from the anchor point state and accumulates the edge mileage on the divergence anchor point map to form... The path mileage set is then converted to the arrival time based on the vehicle's current speed and aligned with a preset time step to obtain an executable time window. At the same time, under the boundary constraints of the lane-change-restricted section, non-executable paths are deleted, and the upper limit of the number of lane changes is determined based on the available driving mileage and the lower limit of the lane-change distance of each lane-change side. Finally, the trigger mileage interval that meets the lane-change number constraint is extracted and written into the window state, so that the trigger position generated by the subsequent guidance code action grid can fall into the executable trigger mileage and be completed within the executable time window. Thus, executable path instructions can still be output in scenarios with dense traffic or a high proportion of lane-change-restricted sections.

[0021] S5. Based on the code state query, the inducement code action grid consists of a hold action node, a lane change action node and a diversion action node. Each action node is configured with a trigger position and a trigger threshold. Then, the inducement code action grid and the window state are subjected to intersection filtering and time sequence rearrangement to generate a set of action trigger positions and a set of trigger thresholds and form a path instruction. This embodiment illustrates how, after obtaining the code state and window state, an induction code action grid is constructed and path instructions are generated, thereby transforming the target VMS induction information into a triggerable, verifiable, and reorderable sequence of action nodes. The principle is as follows: first, the action type drives the action template retrieval, and the action parameter set generates an action node sequence to form the induction code action grid. Then, under window state constraints, the trigger position and trigger threshold are calculated, and intersection filtering and temporal reordering are performed to obtain the path instructions. Furthermore, when action parameters are missing or conflicting, the action sequence is reconstructed through constraint solving and error correction processes to ensure the action grid is executable. This implementation process includes the following steps: By retrieving action templates based on action types and mapping action parameter sets to action node sequences, the initial construction of the induction code action grid is achieved, thus providing an indexable action node carrier for subsequent writing of trigger positions and trigger thresholds. Input includes the action type and action parameter set in the code state, as well as the action template index field in the pre-stored code table. The action type is an enumerated value, and the action parameter set includes at least an action sequence number field, a trigger mileage offset field, a trigger threshold rule identifier field, and a constraint identifier field. The processing involves retrieving action templates from the pre-stored code table or a template table in the same database as the pre-stored code table using the action type as the search key. Each action template includes at least corresponding records with an action sequence number field and a node type field. The retrieved action templates are written to a cache for subsequent sequential reading. Then, the action sequence number field is extracted from the action parameter set, and deduplication and sorting are performed. Deduplication is achieved by merging records with the same action sequence number value to obtain a unique sequence number set. Sorting is achieved by arranging the action sequence number in ascending order to obtain an action sequence number sequence. Finally, the current action sequence number is read item by item, and the corresponding node type is retrieved from the action template. The node type is... The system handles one of three types of actions: hold, lane change, and diversion. When the node type is hold, a hold action node is generated; when the node type is lane change, a lane change action node is generated; and when the node type is diversion, a diversion action node is generated. The action node is generated by creating a node record containing an action sequence number field and a node type field, and writing it to the node cache. Then, action parameter records whose action sequence number field equals the current action sequence number are selected from the action parameter set. The selected trigger mileage offset field, trigger threshold rule identifier field, and constraint identifier field are written into the action node to form the node parameter field. The action nodes are then summarized in the order of their action sequence numbers to obtain the action node sequence, and this sequence is used as the inducement code action grid. The output includes the inducement code action grid, which is written into the subsequent trigger position calculation process for later reading. Anomaly handling includes outputting a template missing flag and triggering a fallback to generate only hold action nodes when the action template retrieval fails. When the action sequence number retrieval fails or the node type is missing in the action template, the corresponding action sequence number is marked as invalid and removed from the action sequence number sequence, while the reason for invalidity is recorded for subsequent constraint solving. By determining the baseline mileage within the executable trigger mileage in window mode and writing the trigger position and trigger threshold, the trigger elements required for path instructions are generated, thus enabling action nodes to have calculable trigger positions and verifiable trigger thresholds. Inputs include the current road segment and current lane in vehicle mode, the executable trigger mileage and executable time window in window mode, the action node sequence in the guidance code action grid, and the trigger mileage offset field and trigger threshold rule identifier field in the action parameter set. The processing action includes determining the baseline mileage within the executable trigger mileage. The baseline mileage is determined by retrieving the corresponding current lane mileage reference point from the pre-stored road topology based on the current road segment and current lane in vehicle mode. The process involves several steps: first, determining the consistency between the reference point mileage and the starting mileage of the executable trigger mileage; if the determination passes, the starting mileage of the executable trigger mileage is set as the base mileage; if the determination fails, the current lane mileage reference point mileage is cropped into the executable trigger mileage range and set as the base mileage. Then, the action node sequence is processed node by node, reading its action sequence number field. Action parameter records with action sequence numbers equal to the specified action sequence number are selected from the action parameter set to read the trigger mileage offset field. The trigger mileage offset field is then superimposed with the base mileage execution mileage to obtain the trigger position of the action node, and this trigger position is written into the action node to form a trigger position set. Finally, the trigger threshold in the action parameter set is... The rule identifier field is valued item by item, and the value is used to retrieve the threshold rule record in the pre-stored threshold rule table to read the threshold field. The threshold field includes at least a threshold type field, a threshold value field, and a threshold unit field. The pre-stored threshold rule table is generated by pre-configuring threshold types for different action types and writing threshold values ​​and threshold units according to road level and vehicle speed range to form rule records and store them in the database. After a retrieval hits, the threshold field is written to the action node with the same action sequence number field to form the corresponding trigger threshold, and the trigger thresholds of all action nodes are summarized to form a trigger threshold set. Subsequently, the action node sequence in the guidance code action grid is subjected to intersection filtering. The intersection filtering is based on the executable time window and executable... The number of lane changes and the executable trigger mileage serve as constraints. The executable time window limits the upper limit of the trigger timing of action nodes, the number of executable lane changes limits the upper limit of the number of lane-changing action nodes, and the executable trigger mileage limits the trigger positions from falling into the range. During filtering, action nodes whose trigger positions do not fall into the executable trigger mileage are deleted, and the number of remaining lane-changing action nodes is counted. If the count exceeds the number of executable lane changes, the excess lane-changing action nodes are deleted from farthest to nearest trigger position. The remaining action nodes after filtering are rearranged in ascending order of trigger position to obtain the path instructions. The output includes the trigger position set, the trigger threshold set, and the path instructions, and the path instructions are written into the subsequent satisfaction determination process for reading.Anomaly handling includes setting the base mileage as the start mileage of the executable trigger mileage and recording the alternative reason when the base mileage cannot be found in the pre-stored road topology; and marking the corresponding action node as missing threshold and removing it from the path instruction when the threshold rule retrieval fails to find the target to avoid unverifiable triggering. By forming a candidate sequence from the action sequence number field and the action template, and reconstructing the action sequence number sequence through constraint solving, convergent reconstruction of the action lattice structure is achieved under conditions of missing, duplicate, or conflicting elements. This ensures that subsequent node generation and threshold writing have a consistent sequence number skeleton. The input includes the action parameter set, the action template, and field segment length constraints in the pre-stored code table. The action template at least includes the sequential definition of the template node type sequence, and the field segment length constraints are used to verify the decodeability of the action parameter field. The action processing includes extracting the action sequence number field from the action parameter set, performing deduplication and sorting to form the action sequence number sequence, and processing the action sequence number item by item in... The node type field is retrieved from the action template to form a node type sequence. The action number sequence and the node type sequence are combined to output a candidate sequence. Subsequently, a constraint set is constructed based on the candidate sequence, and constraint solving is performed to reconstruct the action number sequence. The constraint set includes four types of constraints: a one-to-one correspondence constraint between the action number field and the node type field, which requires that each action number matches only one record in the action template when retrieving the node type field, and that the node type count corresponding to the same action number in the candidate sequence is one; and a node type sequence consistency constraint, which requires that the node type sequence generated from the candidate sequence matches the template node type sequence defined in the action template. The column is assigned an equal value for each item. The action parameter field completeness constraint requires that the trigger mileage offset field, trigger threshold rule identifier field, and constraint identifier field all have values, and the value length meets the field segment length constraint. The constraint identifier field conflict constraint requires that action nodes do not contain lane change constraint identifiers, lane change action nodes do not contain hold constraint identifiers, and diversion action nodes do not contain prohibit diversion constraint identifiers, and that no mutually exclusive identifier pairs appear simultaneously within the constraint identifier set. The constraint solution uses the sum of cost terms as the objective and stops when the cost terms no longer decrease. The cost terms are counted by the sequence number of the violation of the one-to-one correspondence constraint and the node count of the violation of the consistency constraint. The solution consists of the count of fields violating integrity constraints and the count of flags violating conflicting constraints. The solution process performs iterative updates on the candidate sequence. The iterative updates include deleting action numbers that violate one-to-one correspondence constraints, replacing the positions of action numbers that cause consistency constraint failures, removing action parameter records that violate integrity constraints, and deleting constraint flags that violate conflicting constraints, until the sum of cost terms no longer decreases and the reconstructed sequence is output. The output includes the reconstructed sequence, which is written into the subsequent consistency verification and error correction process for reading. Anomaly handling includes outputting a reconstruction failure flag and triggering a fallback to a degradation path that only retains the action nodes when the constraint solution outputs an empty sequence. By performing field segment verification and rollback replacement based on the reconstructed sequence and generating a correction sequence accordingly, the system achieves decodeable error correction of action parameters and final generation of action nodes, thereby outputting an inducement code action grid that can be stably used through intersection filtering and temporal rearrangement. The inputs include the reconstructed sequence, the action parameter set, field segment rules in the pre-stored code table, and the action template. The field segment rules at least include the start position and length of each subfield of the action parameter field. The processing actions include reading the action sequence number one by one from the reconstructed sequence and filtering action parameter records in the action parameter set whose action sequence number field equals the action sequence number. Field segment verification is performed on the filtered records. This verification includes determining whether the trigger mileage offset field length, trigger threshold rule identifier field length, and constraint identifier field length meet the required segment length. If any verification fails, the action sequence number is marked as needing rollback. Rollback replacement is performed on the action sequence numbers that need rollback. The rollback replacement uses the action parameter records of adjacent action sequence numbers under the same action template as the replacement source, with the selection of the replacement source decreasing according to the sum of cost terms. As an update criterion, the cost term consists of the number of fields with failed bit segment verification and the number of conflict constraint failure flags. When the sum of the cost terms decreases after replacement, the replacement result is written and the action parameter record is updated. When the sum of the cost terms does not decrease after replacement, the original record is maintained and the reason for the rollback failure is recorded. The output of the correction sequence is based on the condition that no rollback replacement occurs. Then, according to the correction sequence, the node type field is retrieved one by one in the action template and the keep action node, lane change action node and diversion action node with the same node type field are generated. The action parameters whose action sequence number field in the action parameter set is equal to the corresponding action sequence number are written into the generated action node to form the action node sequence. The action node sequence is determined as the inducement code action grid and output. The output includes the correction sequence and the inducement code action grid. The inducement code action grid is written into the trigger position and trigger threshold generation process for reading. The exception handling includes marking the action node corresponding to the action sequence number as invalid and removing it from the action node sequence when the rollback replacement still cannot meet the bit segment verification, so as to ensure that the output inducement code action grid only contains decodeable and executable action nodes. Through the above implementation process, the action types and action parameter sets in the code state can be transformed into verifiable induced code action grids. Under window state constraints, trigger positions and trigger thresholds are generated, forming path instructions. Simultaneously, in cases of missing action parameters, conflicts, or bit segment anomalies, convergent correction sequences are obtained through constraint solving and backoff replacement, thereby reducing the risk of link interruption caused by non-executable action nodes and improving the consistency between path instructions and executable constraints. In practical applications: when a vehicle is traveling on the mainline and the window state provides executable trigger mileage and executable lane change counts, the on-board terminal reads the action type from the code state and retrieves the action template. Then, it generates hold, lane change, and diversion action nodes from the action parameter set. The system generates a trigger code action grid. After determining the base mileage within the executable trigger mileage, it superimposes the trigger mileage offset to generate the trigger position. Simultaneously, it retrieves the threshold field from the pre-stored threshold rule table according to the trigger threshold rule and writes it into the trigger threshold. Then, it filters the action node sequence based on the executable time window, the number of executable lane changes, and the executable trigger mileage, and rearranges it according to the trigger position to form the path instruction. When the action parameters have duplicate sequence numbers, missing fields, or conflicting constraint identifiers, it first performs constraint solving to reconstruct the sequence number and performs segment verification and backtracking replacement to obtain the correction sequence before outputting the trigger code action grid. This ensures that the final path instruction remains triggerable, verifiable, and stably executed under the limited mileage and lane change prohibition constraints before the ramp diversion.

[0022] S6. During vehicle operation, periodically acquire BeiDou positioning and update vehicle status accordingly. Perform satisfaction judgment on the action triggering position and triggering threshold in the path instruction one by one, output the action execution status and form the execution result. This embodiment illustrates how, after path instructions are generated, the vehicle status is continuously updated during vehicle movement, and the satisfaction of action triggering conditions is determined, thereby outputting the action execution status and result. The principle is as follows: first, residual modeling is performed on multiple BeiDou positioning operations within a preset time step to form positional confidence and complete lane consistency verification and vehicle status correction; then, the remaining mileage is calculated according to the action node sequence of the path instructions, and the action trigger is determined based on the joint condition of the remaining mileage and a speed threshold; subsequently, conflict resolution is performed on multiple triggering cases within the same time step, and a terminateable execution result is output. This implementation process includes the following steps: By performing residual calculations on BeiDou positioning within a preset time step and completing consistency verification and correction under road topology constraints, continuous and reliable updates of the vehicle state are achieved. This provides the current road segment and current lane with confidence support for subsequent action triggering decisions. The inputs include multiple BeiDou positioning data collected within the preset time step, the corresponding sampling time sequence, the current lane and driving direction in the vehicle state, and the lane boundary lines and lane center lines in the pre-stored road topology. The preset time step is given by configuration parameters and is consistent with or an integer multiple of the BeiDou positioning update cycle. The processing actions include sorting the multiple BeiDou positioning data within the preset time step according to the sampling time to form a positioning sequence, and calculating the position based on the positioning sequence. The residual sequence is calculated by using the initial location of the positioning sequence as a reference location and calculating the coordinate difference vector between each subsequent location and the reference location. The magnitude of the coordinate difference vector is then written into the residual value to form the residual sequence. The location confidence is then determined based on the location residual sequence. The location confidence is calculated by taking the median of the residual sequence and comparing it with a preset residual threshold. Residuals with a median less than or equal to the residual threshold are identified as high confidence, while those with a median greater than the residual threshold are identified as low confidence. The residual threshold is given by a preset configuration and varies with road level and satellite signal quality level. The location confidence is then written into the vehicle state to update the vehicle state. The updated vehicle state is used as the input for consistency verification. Then, based on the pre-stored road topology, the updated vehicle state is checked to determine whether the BeiDou positioning falls within the lane boundary of the current lane and whether the angle difference between the driving direction and the current lane direction is less than or equal to the angle difference threshold. The "falling within" condition is determined by checking within the polygon formed by the BeiDou positioning point and the current lane boundary line. The angle difference is calculated by checking the angle difference between the driving direction in the vehicle state and the tangential direction of the current lane centerline at the projection point. The angle difference threshold is given by a preset configuration and varies with the road level. When the consistency verification passes, the updated vehicle state is output as the target vehicle state and written into the subsequent action triggering process for reading. When consistency is achieved... If the validity check fails, a set of candidate lanes is retrieved within the BeiDou positioning neighborhood. The neighborhood radius is given by a preset configuration and is linked to the positioning confidence level. When the confidence level is low, the neighborhood radius is increased to the upper limit of the radius; when the confidence level is high, the default radius is used. For each candidate lane, the lateral offset and the angle difference are calculated. The lateral offset is obtained by calculating the minimum point-to-line distance from the BeiDou positioning point to the center line of the candidate lane, and the angle difference is obtained by calculating the angle difference between the driving direction and the tangential direction of the candidate lane center line. The vehicle states corresponding to the candidate lanes with lateral offset less than or equal to the offset threshold and angle difference less than or equal to the angle difference threshold are added to the candidate vehicle state set. The offset threshold is given by a preset configuration and is determined according to the road level and positioning confidence level.The candidate vehicle state set is sorted by location confidence and the vehicle state corresponding to the first position of the sort is selected as the target vehicle state to update the vehicle state. The target vehicle state is output and written into the subsequent action trigger judgment process for reading. Anomaly handling includes keeping the vehicle state unchanged at the previous time step and outputting a retention flag when the candidate vehicle state set is empty, while marking the location confidence as low confidence to trigger the subsequent re-search with expanded neighborhood radius. By calculating the remaining mileage according to the path instructions and jointly determining the action trigger based on the remaining mileage and speed threshold, and then performing multi-trigger conflict resolution within the same window to generate a terminateable execution result, the verifiable output of the path instructions during the driving process is realized. The input includes the target vehicle state, the sequence of action nodes in the path instructions, the trigger position and trigger threshold of each action node, the executable time window in the window state, and the current BeiDou positioning and current speed within a preset time step. The trigger mileage threshold and speed threshold are obtained by writing the trigger threshold into the action node, and their values ​​are derived from the threshold field of the pre-stored threshold rule table. The processing action includes reading the action nodes one by one according to the action node sequence of the path instructions and calculating the current position up to the trigger of the action. The remaining mileage is calculated by projecting the current BeiDou positioning onto the center line of the current lane of the target vehicle in the pre-stored road topology to obtain the current mileage, and then calculating the difference between the trigger location mileage of the action node and the current mileage to obtain the remaining mileage. For each action node, a joint satisfaction determination is performed. The joint satisfaction determination is based on the condition that the remaining mileage is less than or equal to the trigger mileage threshold and the speed in the target vehicle state meets the speed threshold. The speed meeting the speed threshold is obtained by performing an inequality determination on the speed and the speed threshold according to the threshold type field. When the joint satisfaction determination passes, the action node is marked as triggered, and the BeiDou positioning and action sequence number fields at the trigger time are recorded and written into the trigger record. When the joint satisfaction... If a decision fails, the action node is marked as not triggered, and the reason for not triggering is recorded in the trigger record. The trigger records of all action nodes are summarized to form an action execution status set, which is then output. Subsequently, conflict resolution is performed on the action execution status set. Conflict resolution involves filtering multiple action nodes marked as triggered within the same preset time step and sorting them in ascending order of remaining mileage. Only the action node at the top of the sorted list is retained as triggered, while the remaining action nodes are marked as not triggered, and the reason for the conflict is recorded. The resolved action execution status set is output and written into the subsequent result generation process for retrieval. Finally, the completion status of the action nodes in the resolved action execution status set is determined, or the executable time window is determined to have reached its upper limit. The completion status is determined by retrieval... The trigger mark of the diversion action node is obtained by triggering. The determination of the time window reaching the upper limit is obtained by comparing the accumulated execution time since the path instruction was generated with the execution upper limit of the executable time window. When either determination is met, the execution result is generated and output. The execution result includes a completion flag and a non-completion reason flag. The completion flag is assigned based on the completion status of the diversion action node, and the non-completion reason flag is assigned based on the reason that the time window has reached the upper limit and the diversion action node has not been completed. Anomaly handling includes marking the remaining mileage as missing and marking the action node as not triggered when the current lane centerline projection fails. At the same time, the rollback is triggered by the approximate calculation of the mileage converted from the Euclidean distance between the Beidou positioning and the trigger position and the rollback reason is recorded. Through the above implementation process, even with fluctuations in BeiDou positioning and frequent lane changes, a consistency verification and correction mechanism driven by positioning reliability can stably update the vehicle status. Action trigger determination is completed using a combination of remaining mileage and speed thresholds. Simultaneously, by resolving conflicts from multiple triggers within the same window and determining the upper limit of the executable time window, a terminateable execution result is output, thereby reducing false triggers and missed triggers and improving the verifiability of the path command execution chain. In practical applications: when a vehicle approaches a diversion area and the path command includes lane changing and diversion action nodes, the onboard terminal aggregates multiple BeiDou positioning data within each preset time step and calculates the position residual sequence to obtain... The system determines the location confidence level, then performs lane entry and angle difference determination under road topology constraints, and re-searches candidate lanes to correct vehicle status when necessary. Subsequently, it calculates the remaining mileage from the current position to the trigger position according to the action node sequence, and performs joint determination with the trigger mileage threshold and speed threshold to mark the trigger action. When multiple triggers occur within the same time step, only the action node with the smallest remaining mileage is retained as the trigger. Finally, when the diversion action node is completed or the executable time window reaches the upper limit, the execution result containing the completion flag and the reason for non-completion flag is output, so that the path selection guidance process can still be implemented and traced and verified under traffic flow disturbance and positioning jitter scenarios.

[0023] Working principle: The system transforms the prompts from variable message signs into a series of actions that can actually be executed in the current lane. First, it uses BeiDou positioning, direction, and speed to determine the vehicle's current road segment and lane within the road topology, forming a vehicle state. Then, it reads the guidance code from the variable message sign and decodes it according to the code table to obtain the desired diversion point and corresponding action parameters, forming a code state. Next, the system converts the road topology into a constrained diversion anchor point map, writing the no-lane-change zones and the distances required for lane changes into the map, calculating the diversion path reachable from the current lane. Combining this with the vehicle speed, it calculates the remaining execution time range, the maximum number of lane changes, and the allowed mileage range for triggering actions, forming a window state. Finally, the system generates action nodes such as hold, lane change, and diversion according to the action template, calculates the trigger position for each action, writes the trigger threshold, uses the window state to filter out non-executable actions, and sorts them by trigger position to form path instructions. During the journey, it continuously updates the BeiDou positioning, determining whether the trigger position has been reached and the threshold is met. If it is, it triggers and outputs the execution status until diversion is completed or the executable time range is exceeded, at which point the result is output. For example, when a vehicle approaches a ramp on the main line of an expressway, a variable message sign ahead indicates a right-side lane change, but the section near the ramp is a solid line prohibiting lane changes. The system first confirms the vehicle's lane, then decodes the corresponding lane change point and action parameters for the target ramp. In the lane change anchor point map, lane change edges that cross the solid line are directly excluded. At the same time, the system determines the maximum number of lane changes allowed within the remaining mileage based on the distance required for the lane change. Then, the system provides the mileage range within which the action can be triggered in a windowed manner. Thus, the system only retains hold, lane change, and lane swerving actions that can be completed within this range, and prompts the system in sequence according to the trigger location. As the vehicle moves forward, the system continuously verifies the location using BeiDou positioning. Once the vehicle reaches the trigger location and the conditions are met, the system prompts the vehicle to execute the action. If there is insufficient time or the conditions are not met, the system will stop issuing unexecutable lane change instructions and output the reason for the incomplete operation.

[0024] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A driver route selection guidance method based on VMS guidance information, characterized in that, include: S1. Obtain the vehicle's BeiDou positioning, driving direction and speed, match it with the pre-stored road topology, solve the vehicle's current road segment and current lane, and output the vehicle status; S2. Read the inducement code displayed by the target VMS, and decode the inducement code into the target diversion point, action type and action parameters according to the pre-stored code table, and output the code state; S3. Construct a diversion anchor point graph based on the pre-stored road topology. The diversion anchor point graph uses diversion point identifiers as nodes and lane reachability relationships as edges. Record the lower limit of lane change distance and the boundary of the lane change restriction interval on the edges. Then, map the vehicle state to the diversion anchor point graph and solve the reachable anchor point path set, and output the anchor point state. S4. Calculate the window cone based on vehicle state and anchor point state. The window cone consists of an executable time window, executable lane change count, and executable trigger mileage. Output the window state. S5. Based on the code state query, the inducement code action grid consists of a hold action node, a lane change action node and a diversion action node. Each action node is configured with a trigger position and a trigger threshold. Then, the inducement code action grid and the window state are subjected to intersection filtering and time sequence rearrangement to generate a set of action trigger positions and a set of trigger thresholds and form a path instruction. S6. During vehicle operation, periodically acquire BeiDou positioning and update vehicle status accordingly. Perform satisfaction judgment on the action trigger position and trigger threshold in the path instruction one by one, output the action execution status and form the execution result.

2. The driver route selection guidance method based on VMS guidance information according to claim 1, characterized in that: S1 includes: S1-1. Obtain the vehicle's current BeiDou positioning, driving direction and speed, and retrieve the candidate road segment set and candidate lane set based on the pre-stored road topology, and output the candidate set. The pre-stored road topology includes the diversion point set, diversion point identifier, lane set, lane connection relationship, no-lane-change section boundary and lane-change distance lower limit. S1-2. Based on the candidate set, calculate the lateral offset from the positioning point to the center line of the corresponding candidate lane, the angle difference between the driving direction and the direction of the corresponding candidate lane for each candidate lane, and calculate the forward projection point corresponding to the preset time step based on the speed and driving direction. Then, for each candidate lane, determine whether the lateral offset meets the offset upper limit, whether the angle difference meets the angle difference upper limit, and whether the forward projection point falls within the boundary range of the corresponding candidate lane. Based on the determination results, form a lane score set. S1-3. Select the target lane from the lane score set that simultaneously meets the upper limit of lateral offset, the upper limit of included angle difference, and the forward projection point falls within the lane boundary range, and determine the road segment to which the target lane belongs as the target road segment, and generate the vehicle state, which includes the vehicle's current road segment and the vehicle's current lane.

3. The driver route selection guidance method based on VMS guidance information according to claim 2, characterized in that: S2 includes: S2-1. Obtain the inducement code currently displayed by the target VMS, and retrieve the code table record that matches the inducement code from the pre-stored code table; S2-2. Read the diversion point identifier field, action type field and action parameter field from the code table record, and extract the action sequence number field, trigger mileage offset field, trigger threshold rule identifier field and constraint identifier field from the action parameter field according to the preset field segments in the code table record to form an action parameter set; S2-3. Generate the target diversion point based on the diversion point identifier field, generate the action type based on the action type field, and write the target diversion point, action type and action parameter set into the code state.

4. The driver route selection guidance method based on VMS guidance information according to claim 3, characterized in that: S3 includes: S3-1. Based on the set of branch points and branch point identifiers in the pre-stored road topology, associate each branch point with its branch point identifier to construct a node set and output the node set; S3-2. Based on the lane connection relationship in the pre-stored road topology, determine that the connection direction of each lane connection relationship is consistent with the driving direction in the vehicle state and that the connection interval does not fall within the boundary coverage of the lane change prohibited interval. When the determination is satisfied, write the corresponding starting point divergence point identifier and ending point divergence point identifier into the edge set. S3-3. Write the starting and ending divergence point identifiers of the edge set into the edge set according to the lower limit of the lane change distance and the boundary of the lane change prohibited section to form a divergence anchor point map; S3-4. Based on the current road segment and current lane in the vehicle state, determine the starting node associated with the current lane in the diversion anchor point graph, and use the starting node as the initial set. Perform layer-by-layer expansion on the unvisited nodes according to the connection direction of the edge set. During the expansion, only add the endpoint node corresponding to the edge that satisfies the lower limit constraint of lane change distance and does not cross the boundary of the lane change prohibited section to the initial set. After no new nodes are added, summarize the node sequence from the starting node to each endpoint node as the reachable anchor point path set. S3-5. Write the branch anchor point map and the set of reachable anchor point paths into the data field of the anchor point state to form the anchor point state.

5. A driver route selection guidance method based on VMS guidance information according to claim 4, characterized in that: The process of generating the split anchor point diagram in S3 also includes: S3-31. Based on the pre-stored road topology edge set, for each edge, retrieve the corresponding lane pair identifier and connection mileage segment according to the starting point divergence point identifier and the ending point divergence point identifier, and read the lane centerline spacing sequence and solid line marking mileage segment sequence of the corresponding lane pair within the connection mileage segment, and output the edge attribute candidate set. S3-32. Based on the edge attribute candidate set, generate candidate boundary of no lane change interval for each edge and candidate lower limit of lane change distance. The candidate boundary of no lane change interval is formed by merging and deduplicating the sequence of solid line marking mileage segments and satisfies the monotonically increasing constraint of interval boundary. The candidate lower limit of lane change distance is formed by taking the lower limit value after aligning the sequence of lane centerline spacing by mileage and satisfies the non-negative constraint. Output the candidate attribute set. S3-33. Perform consistency verification and conflict resolution on the candidate attribute set to form a constraint set. The constraint set includes the endpoint constraint of the connecting mileage segment where the boundary of the no-lane-change section must not cover the edge, the unique constraint of the no-lane-change section boundary corresponding to the same starting point divergence point identifier and the same ending divergence point identifier, and the unique constraint of the lower limit of the lane-changing distance corresponding to the same starting point divergence point identifier and the same ending divergence point identifier. Generate a conflict set for the candidate attribute records that violate the constraints, and output the constraint set and the conflict set. S3-34. Based on the constraint set and conflict set, perform constraint solving to determine the written value. The constraint solving takes the sum of cost terms as the objective function, and the cost terms are composed of the adjustment amount of the restricted lane change interval boundary and the adjustment amount of the lower limit of the lane change distance. The stopping condition is that the sum of cost terms no longer decreases. Solve the restricted lane change interval boundary and the lower limit of the lane change distance corresponding to each edge. Write the solution results into the edge set according to the starting point diversion point identifier and the ending point diversion point identifier to form a diversion anchor point map and output it.

6. A driver route selection guidance method based on VMS guidance information according to claim 5, characterized in that: S4 includes: S4-1. For each path in the reachable anchor point path set, accumulate the edge mileage along the branch anchor point graph and form the path mileage from the current position of the vehicle to the end point of each path, and output the path mileage set; S4-2. Calculate the arrival time for each item in the path mileage set based on speed, and align the arrival time with the preset time step to form an executable time window, and output the time window parameters. S4-3. For each path in the reachable anchor point path set, count the number of lane-changing edges contained in the path, and for each lane-changing edge, read its lower limit of lane-changing distance and the boundary of the lane-changing prohibited interval. Delete the path that crosses the boundary of the lane-changing prohibited interval, and summarize the number of lane-changing edges of the remaining paths as the number of lane-changing operations that can be performed. Output the lane-changing parameters. S4-4. Based on the time window parameters and lane change parameters, extract the trigger mileage interval that satisfies the executable lane change number constraint from the path mileage set to form the executable trigger mileage, and write the executable time window, executable lane change number and executable trigger mileage into the window state to generate a window cone.

7. A driver route selection guidance method based on VMS guidance information according to claim 6, characterized in that: S5 includes: S5-1. Obtain the action type and action parameter set in the code state, and retrieve the action template corresponding to the action type based on the pre-stored code table; S5-2. Based on the action parameter set, the action number field is deduplicated and sorted to form an action number sequence. For each action number in the action number sequence, the corresponding node type is retrieved in the action template. When the node type is hold, a hold action node is generated; when the node type is lane change, a lane change action node is generated; when the node type is diversion, a diversion action node is generated. The action parameters in the action parameter set whose action number field is equal to the action number are written into the generated action node to form an action node sequence. The action node sequence is then identified as the inducement code action grid. S5-3. Based on the current road segment and current lane in the vehicle state, determine the reference mileage within the executable trigger mileage in the window state, and superimpose the trigger mileage offset field in the action parameter set with the reference mileage execution mileage to generate the trigger position of each action node, and output the trigger position set. S5-4. Take the value of each trigger threshold rule identifier field in the action parameter set, and retrieve the threshold rule record with the same value in the pre-stored threshold rule table to read the threshold field. Write the threshold field into the action node with the same action sequence number field to form the corresponding trigger threshold, and summarize the trigger thresholds of each action node to form a trigger threshold set. S5-5. Perform intersection filtering on the action node sequence in the induced code action grid. The intersection filtering uses the executable time window, executable lane change number and executable trigger mileage in the window state as constraints and deletes action nodes that do not meet the constraints. Perform time sequence rearrangement on the remaining action nodes according to the trigger position to form path instructions.

8. A driver route selection guidance method based on VMS guidance information according to claim 7, characterized in that: The process of determining the priming code action lattice in S5 also includes: S5-21. Perform deduplication and sorting on the action number field in the action parameter set to form an action number sequence. Then, for each action number sequence, retrieve the node type field in the action template to form a node type sequence and output the candidate sequence. S5-22. Construct a constraint set based on the candidate sequence and perform constraint solving to reconstruct the action number sequence. The constraint set includes a one-to-one correspondence constraint between the action number field and the node type field, a consistency constraint between the node type sequence and the action template, a complete constraint of the action parameter field, and a conflict constraint between the constraint identifier field. The constraint solving takes the sum of cost terms as the objective and the fact that the cost terms no longer decrease as the stopping condition, and outputs the reconstructed sequence. S5-23. Perform consistency verification and error correction on the action parameter set based on the reconstruction sequence. The consistency verification includes performing field bit field verification on the action parameter field corresponding to each action number and performing back-off replacement on the action number that does not meet the field bit field verification. The back-off replacement uses the decrease of the sum of cost terms as the update criterion and the absence of back-off replacement as the stopping condition, and outputs the correction sequence. S5-24. Generate hold action nodes, lane change action nodes and diversion action nodes that are consistent with the node type field according to the correction sequence, and write the action parameters whose action sequence number field is equal to the corresponding action sequence number into the generated action nodes to form an action node sequence. Determine the action node sequence as the inducement code action grid and output it.

9. A driver route selection guidance method based on VMS guidance information according to claim 8, characterized in that: S6 includes: S6-1. Within a preset time step, aggregate multiple BeiDou positioning and calculate the position residual sequence. Solve the position confidence based on the position residual sequence and write the position confidence into the vehicle state to update the vehicle state. S6-2. Based on the pre-stored road topology, determine whether the BeiDou positioning falls within the lane boundary range of the current lane and whether the angle difference between the driving direction and the current lane direction is less than or equal to the angle difference threshold as a consistency verification condition. If the verification fails, search the candidate lane set in the BeiDou positioning neighborhood and calculate the lateral offset and angle difference for each candidate lane. Add the vehicle state corresponding to the candidate lane with the lateral offset less than or equal to the offset threshold and the angle difference less than or equal to the angle difference threshold to the candidate vehicle state set. Select the vehicle state corresponding to the first position of the position confidence ranking in the candidate vehicle state set as the target vehicle state to update the vehicle state and output the target vehicle state. S6-3. Calculate the remaining mileage from the current position to the trigger position for each action node according to the action node sequence of the path instruction. Use the remaining mileage being less than or equal to the trigger mileage threshold and the speed in the vehicle state meeting the speed threshold as the joint satisfaction judgment condition. When the condition is met, mark the action node as triggered and record the Beidou positioning and action sequence number fields at the trigger time. Summarize the marking results of each action node to form an action execution state set and output it. S6-4. Perform conflict resolution on the action execution state set, including sorting multiple action nodes marked as triggered within the same preset time step according to the remaining mileage and retaining only the action node at the top of the sort, marking the remaining action nodes as not triggered and recording the conflict reasons, and outputting the resolved action execution state set. S6-5. Determine the completion status of the centralized and diverted action nodes after the resolution of the action execution status, or determine if the executable time window has reached its upper limit. If so, generate and output the execution result at one time. The execution result includes a completion flag and a non-completion reason flag.

10. A driver route selection guidance system based on VMS guidance information, comprising a driver route selection guidance method based on VMS guidance information as described in claim 9, including a vehicle status positioning module, a guidance code decoding module, an anchor point map routing module, a window cone generation module, a route instruction generation module, and an execution verification output module, characterized in that: The vehicle status positioning module is used to obtain the vehicle's BeiDou positioning, driving direction and speed, and match it with the pre-stored road topology to solve the vehicle's current road segment and current lane, and output the vehicle status. The inducement code decoding module is used to read the inducement code displayed by the target VMS, and decode the inducement code into the target diversion point, action type and action parameters according to the pre-stored code table, and output the code state; The anchor point graph pathfinding module constructs a diversion anchor point graph based on the pre-stored road topology. The diversion anchor point graph uses diversion point identifiers as nodes and lane reachability relationships as edges. It records the lower limit of lane change distance and the boundary of lane change prohibited interval on the edges, and then maps the vehicle state to the diversion anchor point graph and solves the reachable anchor point path set, outputting the anchor point state. The window cone generation module calculates the window cone based on the vehicle state and the anchor point state. The window cone consists of an executable time window, an executable lane change count, and an executable trigger mileage, and outputs the window state. The path instruction generation module is based on the code state query of the inducement code action grid. The inducement code action grid consists of hold action nodes, lane change action nodes and diversion action nodes. Each action node is configured with a trigger position and a trigger threshold. Then, the inducement code action grid and window state are subjected to intersection filtering and time sequence rearrangement to generate a set of action trigger positions and a set of trigger thresholds and form a path instruction. The execution verification output module is used to periodically acquire BeiDou positioning during vehicle operation and update vehicle status accordingly. It performs satisfaction judgment on the action triggering position and triggering threshold in the path instruction one by one, outputs the action execution status and forms the execution result.

Images