Intelligent connected vehicle intelligent auxiliary driving method and system for complex road scene

Abstract

The application discloses an intelligent network connected automobile intelligent auxiliary driving method and system for complex road condition scenes and belongs to the technical field of intelligent network connected automobiles. Through acquisition of vehicle-mounted sensing information, roadside cooperative information and vehicle operation states, boundary segments, slippery signs and unblocking timing are extracted to form a dynamic passable area. On this basis, a continuously approaching target is identified and a conflict transmission chain is constructed. In combination with the slippery signs, the passable space is segmented and compressed to obtain a transferable passable window. Further in combination with the current speed of the vehicle, a first stable speed transfer point and a second regression takeover point are determined, and according to this, a limited-torque slow braking, lateral constraint and target yielding control instruction is output, and a minimum risk parking instruction is output when the passable condition is insufficient. Through continuous reconstruction of the passable space and cooperative processing of the multi-target constraint relationship, the safety and stability of the auxiliary driving under complex road conditions are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent connected vehicle technology, and more specifically to intelligent connected vehicle intelligent assisted driving methods and systems for complex road conditions. Background Technology

[0002] In short-distance merging scenarios where a tunnel exit meets a temporary construction area on a rainy night, glare from accumulated water weakens the identification of incomplete road markings, cone shifting causes the temporary boundary to drift continuously, and two-wheeled vehicles may suddenly cut across the construction opening after the obstruction of the vehicle in front is removed. Existing driver assistance systems typically make decisions based on fixed lanes and single-target recognition, making it difficult to simultaneously handle the coupling relationship between boundary drift, changes in the timing of obstruction, and braking margin in wet conditions. This can easily lead to problems such as accidentally entering the guide zone, yielding too late, or braking imbalance within a small area. Summary of the Invention

[0003] The purpose of this invention is to provide a smart assisted driving method and system for intelligent connected vehicles in complex road conditions, so as to overcome the shortcomings of the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a smart assisted driving method for intelligent connected vehicles in complex road conditions, comprising: Acquire vehicle-mounted perception information, roadside cooperative information and vehicle operating status, extract boundary segments, slippery signs and occlusion removal time sequence, and piece them together to form a dynamic passable area; Based on the dynamically accessible area, targets that are continuously approaching the dynamically accessible area are identified, and a conflict propagation chain is generated according to the probability of entry and the timing of occlusion removal. Based on the conflict transmission chain and slippery signs, the available passage range is compressed in segments along the dynamically passable area to obtain a passable window. Based on the vehicle's current speed and the available passage window, determine the initial stable speed handover point and the subsequent return to takeover point; Based on the first stable speed handover point, the second return to takeover point and the conflict transmission chain, at least one of the following assisted driving control commands is output: torque limiting braking, lateral constraint and target yielding. When the conflict propagation chain continues to lengthen and the available passage window is smaller than a preset value, a minimum risk parking instruction is output.

[0005] Preferably, the steps of acquiring vehicle-mounted perception information, roadside cooperative information, and vehicle operating status, and extracting boundary segments include: using the timestamps of each frame of the vehicle-mounted perception information as a reference, mapping the data with the closest time in the roadside cooperative information to the reference time; when there is a time interval between two adjacent roadside cooperative information, compensating the positions corresponding to the two roadside cooperative information according to the vehicle speed and heading changes within the time interval, so that they fall into the same ground reference position.

[0006] Preferably, the step of extracting boundary segments further includes: after completing segment-by-segment compensation, retaining candidate edges in continuous frames whose center position deviation after vehicle displacement compensation is no greater than 0.15 meters and whose extension direction change is no greater than 10 degrees; when the endpoint spacing of adjacent candidate edges is less than 1 / 4 of the vehicle width and the difference in extension direction is no greater than 10 degrees, splicing adjacent candidate edges into the same continuous boundary; for parts that cannot be directly spliced ​​but are consistent with the edge position indicated by the roadside coordination information, filling in the missing segments according to the edge direction given by the roadside coordination information to obtain boundary segments.

[0007] Preferably, the step of extracting slippery signs includes: dividing the road surface within the boundary segment along the vehicle's direction of travel into continuous strips; statistically analyzing the average brightness, brightness distribution fluctuation, number of identifiable texture edges, and continuous length of texture edges for each strip in three consecutive frames; when the same strip shows a continuous increase in average brightness, a continuous decrease in the number of texture edges, and a continuous shortening of continuous texture edge length in three consecutive frames, and the outer edge of the bright area in adjacent strips gradually expands along the vehicle's direction of travel, the area corresponding to that strip is identified as a candidate reflective area; only candidate reflective areas located between boundary segments or adjacent to the inner side of the boundary segments are retained, and slippery signs are determined based on their distribution range and expansion trend over time.

[0008] Preferably, the steps of generating the occlusion removal sequence and assembling it to form a dynamically passable area include: In the area where the slippery signs are located and the adjacent areas, the road surface with occlusion is selected as the occluded area. After the occluded area in the consecutive frames is compensated to the same ground reference position, the process of the background from missing to complete is compared. The residual marking edge, road joint edge, guardrail bottom edge or construction cone bottom edge with a contour position deviation of no more than 0.1 meters in 5 consecutive frames are used as stable structures. The moment of their first appearance is traced back to form the occlusion removal sequence. Each boundary segment is projected onto the same ground reference position and connected to form a continuous boundary zone. The area enclosed by the continuous boundary zone is used as the initial passable range. The boundary is contracted according to the proportion of wet and slippery signs covered by each segment. Based on the time sequence of the removal of obstruction, only the segments that have formed a continuous exposure sequence are retained and combined to form a dynamic passable area.

[0009] Preferably, the steps for generating a conflict propagation chain according to the entry probability and the timing of occlusion removal include: A unified position correction is performed on each target in consecutive frames, and the shortest distance from each target to the boundary of the dynamically passable area is calculated. When the shortest distance decreases sequentially in three consecutive frames, or the overall decrease exceeds 0.2 meters, the target is identified as a continuously approaching target. Combining the change in the angle between the current movement direction of each target and the boundary of the dynamically passable area, as well as the exposure order of the corresponding spatial position in the occlusion release sequence, the continuously approaching targets are prioritized for entry. Then, it is determined whether the movement path of the next target passes through the area currently occupied by the previous target or the area that it may occupy in the next two time intervals. If so, a sequential dependency relationship is established between the two, and dependencies that exceed the boundary of the dynamically passable area or are inconsistent with the occlusion release sequence are deleted to obtain the conflict propagation chain.

[0010] Preferably, based on the conflict transmission chain and slippery signs, the available passage range is segmented and compressed along the dynamically passable area to obtain a transferable passage window, including: The dynamically passable area is divided into continuous segments along the direction of vehicle travel, and each segment is marked as an affected segment or an unaffected segment according to the spatial location of each target in the conflict transmission chain. For the affected sections, the lateral occupancy range of each target in the corresponding section is determined in the order of the targets in the conflict transmission chain. An additional 0.2 meters is added to each side of the target’s maximum lateral width as a safety margin, and then deducted from the original passage width one by one. Calculate the distribution ratio of slippery signs for each section, and continue to shrink the passage boundary inward on the side where the slippery signs are located according to the distribution ratio; Finally, the remaining passage areas in each section are checked for connectivity and continuously spliced ​​together to form a passable passage window.

[0011] Preferably, by combining the vehicle's current speed with the available passage window, the initial stable speed handover point and the subsequent return-to-takeover point are determined, including: The passable passage window is divided into continuous segments along the vehicle's direction of travel. The distance traveled per unit time is calculated based on the vehicle's current speed. Multiple candidate positions are marked sequentially starting from the initial segment of the passable passage window. When the remaining passage width of the segment where the candidate position is located and the two subsequent consecutive segments are not less than the vehicle width and the width difference between adjacent segments does not exceed 0.2 meters, the candidate position is determined as the position that meets the speed stabilization condition. The position closest to the initial segment is selected first as the speed stabilization handover point. Starting from the section where the first stable speed handover point is located, the remaining passage width of subsequent consecutive sections is extracted in sequence. When there are three consecutive sections whose width increases segment by segment and the increase in each segment is not less than 0.15 meters, and the three consecutive sections are not within the influence range of any target in the conflict transmission chain in the next two time intervals, the first position that meets the conditions is determined as the re-takeover point.

[0012] Preferably, the steps for outputting a minimum-risk parking instruction when the conditions are met include: When the travel time from the vehicle's current position to the first stable speed yielding point is less than the preset time, torque limiting and slow braking are initiated. Between the first stable speed yielding point and the return to the takeover point, the target that needs to be passed first is determined according to the order of the targets in the conflict transmission chain, and the lateral position of the vehicle is constrained according to the remaining passage width of each section within the yieldable passage window, so as to output at least one of the following assisted driving control commands: torque limiting and slow braking, lateral constraint, and target yielding. When the number of targets in the collision propagation chain increases continuously over multiple consecutive frames, and the minimum remaining passage width of the passable window is less than the sum of the vehicle width and the preset safety margin, or the length of the continuous passable section is less than the preset travel distance calculated based on the current speed, a minimum risk stop command is output.

[0013] This invention also provides an intelligent assisted driving system for intelligent connected vehicles in complex road conditions, comprising: The perception fusion module acquires vehicle-mounted perception information, roadside cooperative information, and vehicle operating status, extracts boundary segments, slippery signs, and occlusion removal timing, and combines them to form a dynamic passable area. The conflict prediction module identifies targets that are continuously approaching the dynamically passable area based on the dynamic passable area, and generates a conflict propagation chain according to the probability of entry and the timing of the removal of obstruction. The passage window calculation module, based on the conflict transmission chain and slippery signs, segments and compresses the available passage range along the dynamically passable area to obtain a passable window; The yield decision module, in conjunction with the vehicle's current speed and the available yield window, determines the first steady-speed yield point and the subsequent return-to-take-over point. The driver assistance control module, based on the first speed stabilization and yield point, the second return to take over point and the conflict transmission chain, outputs at least one of the following driver assistance control commands: torque limiting and braking, lateral constraint, and target yield. The risk monitoring module outputs a minimum risk parking instruction when the conflict transmission chain continues to extend and the passable window is smaller than a preset value.

[0014] The technical effects and advantages provided by the present invention in the above technical solution are as follows: 1. This invention integrates boundary segments, slippery signs, and occlusion removal timing under a unified ground reference, transforming the previously fragmented perception of road boundaries, changes in road surface adhesion, and the process of occlusion disappearance into a dynamic passable area with continuous spatial constraints. Unlike existing technologies that rely solely on lane lines or single-frame target recognition for passage decisions, this solution utilizes boundary stability constraints to define the passable space boundary, then uses slippery signs to directionally shrink local adhesion capabilities, and simultaneously uses the occlusion removal timing to restore the true spatial opening sequence of occluded areas. Physically equivalent to a segment-by-segment reconstruction of the road's "passable space," this eliminates misjudgments caused by reflective interference, boundary drift, and abrupt occlusion changes in complex scenarios, ensuring vehicles always make decisions based on the true passable space and directly improving the continuity and reliability of passage boundary recognition.

[0015] 2. This invention transforms multi-target dynamic interference into a spatial occupancy sequence with sequential constraints by constructing a coupling compression mechanism of conflict transmission chain and yieldable passage window, and compresses it segment by segment along the dynamically passable area. Unlike existing technologies that rely solely on single-target avoidance or simple distance threshold control, this solution utilizes the target's future occupancy range to form continuous transmission constraints, which is physically equivalent to "segment-by-segment yielding" of the passage space, and simultaneously corrects lateral stability based on slippery conditions. This process directly causes the passable space to dynamically shrink and recover according to the target relationship and road surface conditions, enabling vehicles to determine the steady-speed yielding point and the return-to-take-over point in advance, achieving coordinated adjustment of speed and lateral position; triggering minimum-risk stopping when passage space is insufficient, thereby effectively avoiding sudden lateral conflicts and braking instability, and significantly improving traffic safety and control smoothness in complex road conditions. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 This is a flowchart of the intelligent assisted driving method for intelligent connected vehicles in complex road conditions according to the present invention.

[0018] Figure 2 This is a flowchart of the intelligent assisted driving system module for intelligent connected vehicles in complex road conditions according to the present invention.

[0019] Figure 3 This is a flowchart of the method for generating a transferable access window according to the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0021] Example 1, please refer to Figure 1 As shown in this embodiment, the intelligent assisted driving method for intelligent connected vehicles in complex road conditions includes: The system acquires vehicle-mounted sensing information, roadside cooperative information, and vehicle operating status, extracts boundary segments, slippery signs, and the timing of occlusion removal, and combines them to form a dynamic passable area.

[0022] The vehicle-mounted perception information is preferably continuous frame images of the road in front and to the side, road surface contour information, and information on changes in occlusion of the target ahead. The roadside cooperative information is preferably construction guidance information, roadside edge information, and temporary traffic restriction information. The vehicle operating status is preferably vehicle speed, displacement per unit time, and change in heading.

[0023] After data acquisition, the timestamps of each frame of the vehicle-mounted sensing information are used as a reference. Then, the data in the roadside cooperative information that is closest to each frame's time is mapped to this reference time. When there is a time interval between two adjacent roadside cooperative information messages, the positions corresponding to the preceding and following roadside cooperative information messages are compensated segment by segment based on the vehicle speed and heading changes within that time interval, so that they fall into the same ground reference position. The segment-by-segment compensation can be understood as: first, the preceding and following positions are longitudinally translated based on the vehicle's cumulative forward distance within that time interval; then, the edge orientation is rotated and corrected based on the heading changes, ultimately obtaining an alignment result that is at the same time and the same ground reference position as the vehicle-mounted sensing information, thus providing a unified input for subsequent extraction of boundary segments.

[0024] Based on the alignment results described above, the boundary segments are extracted. Specifically, the scope of the road surface to be processed is first defined based on roadside coordination information, and then edges with significant contour changes and stable positions after vehicle displacement compensation are searched in consecutive frames. Stable positions mean that the center position deviation of the same edge in three consecutive frames is no greater than 0.15 meters, and the change in extension direction is no greater than 10 degrees; edges meeting this condition are retained as candidate edges.

[0025] The starting and ending positions, extension direction, and spacing between each candidate edge and its adjacent candidate edge are compared. When the distance between the endpoints of two candidate edges is less than 1 / 4 of the vehicle width and the difference in extension direction is no more than 10 degrees, the two candidate edges are spliced ​​together as the same continuous boundary. For parts that cannot be directly spliced ​​but are consistent with the edge position indicated by the roadside cooperative information, the missing segments are filled in according to the edge direction given by the roadside cooperative information.

[0026] After the above processing, boundary segments are obtained that reflect the edges of construction guidance, temporary cones, or remaining road edges. These boundary segments serve as spatial references for subsequent extraction of slippery conditions, and slipperiness assessment is performed only within the road surface they define to eliminate interference from guardrails, vehicle reflections, and bright areas outside the road.

[0027] Using the boundary segment as a reference, the slippery signs are extracted. Specifically, within the road surface defined by the boundary segment, continuous strips are first divided along the vehicle's direction of travel. The length of each strip is preferably 2 meters, and the width is preferably the current visible width of the road surface.

[0028] For each strip, the average brightness, the degree of fluctuation in brightness distribution, and the number and continuous length of identifiable texture edges are statistically analyzed over three consecutive frames. The brightness fluctuation can be obtained by comparing the average brightness difference of the same strip in the current frame, the previous frame, and the two previous frames; the texture extension can be obtained by statistically analyzing the total length of continuously traceable road texture edges within the strip.

[0029] If a certain band exhibits a continuous increase in average brightness, a continuous decrease in the number of texture edges, and a continuous shortening in the total length of texture edges across three consecutive frames, and the outer edges of the bright areas in adjacent bands gradually expand with the direction of vehicle movement, then the corresponding area of ​​that band is identified as a candidate reflective area. Subsequently, the positional relationship between the candidate reflective areas and boundary segments is reviewed, retaining only candidate reflective areas located between or immediately inside boundary segments; bright areas that are not constrained by boundary segments and only appear on the outer side of the road or on the surface of suspended objects are not considered as criteria for determining slipperiness.

[0030] The reflected diffusion areas after verification are identified as slippery signs according to their distribution range and expansion trend over time. The distribution range is given by the total area of ​​the confirmed areas within a continuous strip, and the expansion trend is given by the direction and order in which the area of ​​the confirmed areas increases frame by frame in three consecutive frames.

[0031] Based on the slippery signs, the occlusion removal timing sequence is generated. Specifically, firstly, in the area where the slippery signs are located and in the adjacent areas, the road surface that is obscured by the vehicle in front, construction facilities or temporary obstacles is selected as the occluded area, and the background exposure process of the occluded area in consecutive frames is recorded frame by frame.

[0032] To avoid errors in exposure judgment caused by wet and slippery reflections, the occluded areas in consecutive frames are first compensated to the same ground reference position based on vehicle speed, displacement per unit time, and heading change. Then, the background change process from missing to complete is compared at this unified position.

[0033] The stable structures in the background can be residual marking edges, road joint edges, guardrail bottom edges, or construction cone bottom edges with a contour position deviation of no more than 0.1 meters in 5 consecutive frames.

[0034] Then, starting from the current frame, the first exposure time of the above stable structure is traced back frame by frame, and the exposure order from far to near or from inside to outside is determined according to its arrangement position in the longitudinal direction of the road, thereby forming an initial time node set.

[0035] If a time node in the initial time node set appears later than its adjacent position, or if the order of two time nodes contradicts the relationship between adjacent positions, then the time node is considered an abnormal jump node. For abnormal jump nodes, the intermediate time sequence of the two adjacent valid time nodes is used for correction, or the node is directly removed.

[0036] After continuous correction, the time node sequence of spatial continuous recovery is obtained, which is the occlusion release time sequence.

[0037] After obtaining the boundary segments, slippery signs, and the timing of the removal of obstructions, the dynamic passable area is pieced together.

[0038] First, project each boundary segment onto the same ground reference position and sort them according to the road extension direction. Then connect the segments that meet the requirements of endpoint spacing and direction consistency to form a continuous boundary zone.

[0039] The initial passable area is defined by a continuous boundary strip, and then divided into 2-meter segments along the direction of vehicle travel. For each segment, the coverage ratio of slippery signs is calculated, which is the ratio of the area of ​​slippery signs to the surface area of ​​the road in that segment. When the coverage ratio reaches or exceeds 1 / 3, the segment is shrunk inward by 0.3 meters on the side with the slippery signs; when the coverage ratio reaches or exceeds 1 / 2, the segment is shrunk inward by 0.5 meters on the side with the slippery signs.

[0040] After completing the slipperiness correction, the occlusion removal sequence is used to determine whether the segment has sufficient background exposure. If the segment only has scattered background exposure and has not yet formed a visible sequence of two or more consecutive stable structures, the segment is retained as an unconfirmed area and not included in the dynamically traversable area. If the segment has formed a continuous exposure sequence consistent with its spatial location, the segment is included in the dynamically traversable area. By performing the above processing on each segment in sequence and connecting the retained segments end to end, a dynamically traversable area that updates in real time according to changes in the current scene is obtained.

[0041] As the vehicle continues to move, the aforementioned dynamically passable area can be updated on a continuous frame-by-frame basis. That is, for each new frame of onboard perception information, corresponding roadside coordination information, and vehicle operating status input, time alignment, boundary segment updates, slippery sign updates, occlusion removal timing updates, and segment correction processing are repeatedly performed. For boundary segments determined in the previous time step, if the position stability condition is still met in the new time step, they are retained; if the position stability condition is no longer met for three consecutive frames, they are deleted from the current boundary segment.

[0042] For confirmed slippery signs, if the brightness fluctuations decrease and the number of texture edges recovers in three consecutive frames, then the distribution range is reduced; if the highlight diffusion continues to extend forward, then the distribution range is expanded in the direction of expansion.

[0043] For the occlusion removal sequence, the latest exposure sequence replaces the old sequence, while retaining valid time nodes consistent with the current spatial location. Through the above rolling update method, the dynamically passable area is always jointly defined by the current boundary segment, the current slippery sign, and the current occlusion removal sequence, thereby ensuring that the area corresponds to the road surface range that can be used by the vehicle to assist driving decisions at the current moment.

[0044] Based on the dynamically passable area, targets that are continuously approaching the dynamically passable area are identified, and a conflict propagation chain is generated according to the probability of entry and the timing of occlusion removal.

[0045] Using the spatial range of dynamically passable areas as a basic reference, targets in continuous frames of vehicle-mounted perception information are filtered.

[0046] Each target identified in each frame undergoes unified position correction at a ground reference position. Based on the vehicle's current speed and displacement per unit time, the target's position in adjacent frames is compensated in reverse to unify the target positions across frames to the same reference time. Subsequently, the shortest distance from each target to the boundary of the dynamically passable area is calculated. This distance is obtained by selecting the minimum value among the shortest distances from each point on the target's outer contour to the boundary of the dynamically passable area.

[0047] The shortest distances in three consecutive frames are sorted and compared. If the shortest distance of a target decreases sequentially in the three consecutive frames, the target is determined to be a target continuously approaching the dynamically passable area. If the distance changes fluctuate but the overall decrease exceeds 0.2 meters, it is also retained as an approaching target to avoid misjudgment due to local jitter. Through the above processing, a set of targets that meet the continuous approach condition is obtained.

[0048] Based on the distance change sequence and the occlusion release timing, the entry probability of approaching targets is constrained and ranked. For each approaching target, the change in the angle between its current movement direction and the boundary of the dynamically passable area is determined. If the angle decreases over two consecutive frames, it indicates that the target's movement direction is gradually pointing towards the interior of the dynamically passable area, and a higher entry probability is assigned; otherwise, a lower entry probability is assigned.

[0049] The locations of each target are mapped to the spatial position sequence in the occlusion release timeline, and the earliest time node in the direction in which the background is revealed is found. If the reveal time of a target's location is earlier than that of another target, it indicates that it has the spatial condition to enter the dynamically passable area earlier. By superimposing and sorting the direction of movement and the occlusion release time, an entry priority relationship is formed for each approaching target. The rule for determining the priority relationship is: prioritize targets with earlier reveal times and whose movement direction is closer to the interior of the dynamically passable area.

[0050] Construct the initial association sequence according to the entry priority relationship.

[0051] Arrange the sorted approximation targets in sequence. For two adjacent targets in the sequence, determine whether the path of the latter target passes through the area currently occupied by the former target or its possible occupied area.

[0052] The potential occupied area is determined as follows: Based on the target's current speed, its forward distance over the next two time intervals is calculated, and a fan-shaped coverage area is formed along its forward direction with this distance as the radius. If the forward direction of a subsequent target falls within this fan-shaped coverage area, then the entry of the subsequent target is considered to be affected by the spatial occupancy of the preceding target. When this condition is met, a sequential dependency relationship is established between the two targets, and the spatial position and temporal order corresponding to this dependency relationship are recorded. After processing all adjacent targets in sequence, an initial association sequence containing multiple dependencies is formed.

[0053] The initial association sequence is subjected to continuity correction to obtain a conflict transmission chain. Each pair of dependencies in the initial association sequence is examined one by one to determine whether its corresponding spatial location is within the effective range of the dynamically traversable area. If the spatial location corresponding to a dependency exceeds the boundary of the dynamically traversable area, or if its temporal sequence is inconsistent with the exposure order of the corresponding location in the occlusion release sequence, then the dependency is deleted. For the broken sequence that appears after deleting dependencies, it is reconnected based on the spatial proximity between adjacent targets, where spatial proximity is determined by comparing whether the shortest distance between targets is less than 0.5 meters. After the above correction and replenishment, a sequence reflecting the temporal and spatial constraints between multiple approaching targets is formed, and this sequence is the conflict transmission chain.

[0054] During continuous frame updates, the conflict propagation chain is updated synchronously with changes in the dynamic passable area and the timing of occlusion removal. For targets already in the conflict propagation chain, if they no longer meet the continuous approach condition in the following three frames, they are removed from the chain; for newly approaching targets, they are re-inserted into their corresponding positions according to the aforementioned entry priority relationship. Through continuous updates, the conflict propagation chain always reflects the actual entry relationships and mutual influence order between targets in the current scene, thus providing a basis for determining the subsequent deceleration and yielding intervals.

[0055] Please see Figure 3 As shown, based on the conflict transmission chain and slippery signs, the available passage range is compressed in segments along the dynamically passable area to obtain a passable window.

[0056] The dynamically passable area is segmented, and a spatial correspondence with the conflict transmission chain is established.

[0057] Based on the vehicle's current direction of travel, the dynamically passable area is divided longitudinally into several continuous segments. Each segment is preferably 2 to 3 meters long, and its width is the lateral width of the dynamically passable area at that location. After division, the spatial locations of each target in the conflict transmission chain are mapped to these segments, determining the segment number for each target. If at least one target from the conflict transmission chain exists within a segment, that segment is marked as an affected segment; otherwise, it is marked as an unaffected segment.

[0058] When multiple targets exist within the same segment, the target arrangement order is recorded according to the sequence in the conflict propagation chain, and this order is used as the basis for subsequent compression processing, thereby establishing a segment-by-segment association between the dynamically passable area and the conflict propagation chain.

[0059] For the affected sections, the process is carried out segment by segment inward according to the conflict propagation chain.

[0060] Starting with the affected section closest to the vehicle, targets are processed sequentially according to their order in the conflict propagation chain. For each target, its lateral occupancy range within the corresponding section is first determined. This lateral occupancy range is obtained by measuring the maximum lateral width of the target's outer contour within that section, and then extending 0.2 meters to each side as a safety margin.

[0061] The lateral occupancy area is subtracted from the original passage width of the segment to obtain the initial compression result of the segment. For subsequent targets within the same segment, the subtraction process continues based on the compression result of the previous target, that is, the lateral occupancy area of ​​the subsequent target is superimposed on the already compressed area to form a cumulative compression effect. When the cumulative compression width of a segment exceeds 2 / 3 of the original passage width, the segment is marked as a restricted segment, and only the portion of the remaining space with a continuous width greater than the vehicle width is retained as a candidate passage area.

[0062] Through the above segment-by-segment inward shifting process, the available passage range of the affected segment gradually shrinks as the target spatial distribution in the conflict transmission chain changes.

[0063] After the initial compression based on the conflict transmission chain is completed, slippery signs are introduced to correct the compression results of each segment.

[0064] For each affected section and adjacent sections, the distribution ratio of slippery signs within that section is calculated. This ratio represents the proportion of the slippery sign-covered area to the total area of ​​the section. When this ratio is less than 1 / 4, the passage boundary on the side containing the slippery sign is contracted inward by 0.2 meters; when the ratio is between 1 / 4 and 1 / 2, the boundary on that side is contracted inward by 0.3 meters; and when the ratio is greater than 1 / 2, the boundary on that side is contracted inward by 0.5 meters. During boundary contraction, priority is given to compressing the side that corresponds to the target in the conflict chain to avoid concentrating passage space on the higher-risk side.

[0065] For unaffected sections, if adjacent sections show signs of slipperiness and the distribution ratio exceeds 1 / 2, a 0.2-meter pre-compression treatment is also applied to the slippery side of the unaffected section to ensure a continuous transition between sections. In this way, the impact of slipperiness on vehicle traffic stability is incorporated into the traffic area compression process.

[0066] The remaining space of each segment, after being constrained by the conflict transmission chain and corrected for slippery conditions, is continuously spliced ​​together to form a yieldable passage window. Specifically, the lateral connectivity of the remaining passage areas of all segments is checked. When the remaining passage areas of adjacent segments overlap laterally, the overlapping portion is retained as a continuous passage path. When there is no lateral overlap between adjacent segments, but their boundary distance is less than 0.3 meters, the boundaries of the two segments are smoothly connected using a linear transition method to maintain the continuity of the passage path. For segment connections that do not meet the above conditions, the corresponding segments are removed from the passable paths. After segment-by-segment screening and splicing, a passage area that is spatially continuous, has a width that meets vehicle passage requirements, and avoids the main risk targets in the conflict transmission chain is formed. This passage area is the yieldable passage window.

[0067] As the vehicle continues to travel, the aforementioned yieldable passage windows are dynamically adjusted based on updates to the dynamic passable area, conflict transmission chain, and slippery signs. For a generated yieldable passage window, upon input of a new frame of data, it is first checked whether each segment is still within the updated dynamic passable area; if a segment has left the dynamic passable area, it is removed from the yieldable passage window. The affected segment calibration and segment-by-segment inward shifting process are re-executed based on the updated conflict transmission chain, and the compression degree of each segment is recalculated in conjunction with the new slippery sign distribution.

[0068] The continuous passage area is regenerated according to the above splicing rules, resulting in an updated yieldable passage window. This dynamic update process enables the yieldable passage window to reflect the effective space available for vehicles to yield and pass under complex road conditions in real time, providing a basis for the generation of subsequent control commands.

[0069] Based on the vehicle's current speed and the available passage window, determine the initial stable speed handover point and the subsequent return to takeover point.

[0070] The passable window is divided into continuous segments along the direction of vehicle travel, with each segment having a fixed length of 2 meters, and each segment is numbered. After obtaining the vehicle's current speed, this speed is converted into the distance traveled per unit time of 0.5 seconds. For example, when the vehicle speed is 10 meters per second, the corresponding distance traveled is 5 meters.

[0071] Starting from the initial segment of the transferable passage window, the first candidate position is determined by traversing the corresponding number of segments according to the travel distance. Then, the same method is used to increase the number of segments to form a sequence of multiple candidate positions.

[0072] For each candidate location, the remaining traffic width of its current segment and the two subsequent consecutive segments are extracted to form a three-segment width sequence, and the width difference between adjacent segments is calculated for each segment. When all three segments are not less than the vehicle width and the adjacent differences do not exceed 0.2 meters, the candidate location is marked as a location that meets the speed stability condition.

[0073] All candidate positions that meet the conditions are sorted according to their distance from the starting segment, and the position closest to the starting segment is selected as the initial selection point.

[0074] If multiple candidate locations are spaced less than 2 meters apart, the average width of their corresponding three segments is calculated. This involves adding the three widths together and dividing by 3, selecting the location with the larger average as the final, initial speed-stabilizing yield point. Simultaneously, the location is checked against one segment before and after it. If the width variation within these three segments still meets the condition of not exceeding 0.2 meters, the point is considered valid; otherwise, the next candidate location is considered. This process ensures that vehicles can maintain stable speeds in locations with minimal spatial changes after entering the yield window.

[0075] Starting from the section where the initial steady-speed handover point is located, the remaining passage width of at least five consecutive sections is extracted sequentially to form a width sequence, and the width increment is calculated for each adjacent section. If three consecutive sections meet the following conditions: the width of the later section is greater than the width of the earlier section, and the increase in width of each section is not less than 0.15 meters, then these three sections are marked as candidate recovery sections. At the same time, the previous section of the candidate recovery section is compared. If the width of the previous section is less than the width of the first section in the combination, then the combination is confirmed as an effective recovery starting point, thereby avoiding misjudging local fluctuations as recovery trends.

[0076] The spatial location of the candidate recovery segment is compared one by one with the corresponding locations of each target in the conflict transmission chain to determine whether the segment is still within the influence range of any target. The influence range is determined by extending the target's current position forward by the distance traveled over two future time intervals, forming a coverage area.

[0077] If a candidate recovery segment falls within the coverage area, it is considered still constrained by the target and is removed. If three consecutive segments do not fall within the coverage area of ​​any target, the combination of segments is marked as an unconstrained recovery segment. This process ensures that the recovery segment not only meets the condition of increased spatial width but is also no longer affected by the passage of preceding targets.

[0078] Following the segment numbering order, the search begins from the segment after the initial stable speed handover point, selecting the starting position of the first segment that meets the unconstrained recovery conditions as the re-takeover point. To improve stability, the width consistency of the next two consecutive segments at this position is checked again. If the width variation does not exceed 0.2 meters, the point is confirmed as a valid takeover point; if the variation exceeds 0.2 meters, the search continues to the next candidate recovery segment.

[0079] By following the steps described above, the timing of takeover can be determined at a location where the passage space is gradually restored and no longer constrained by the conflict transmission chain.

[0080] During continuous vehicle movement, the initial speed-stabilizing yield point and the subsequent takeover point are dynamically adjusted based on updates to the yieldable passage window. Whenever the segment width or number of segments within the yieldable passage window changes, the candidate position calculation, width sequence analysis, and conflict propagation chain screening process are re-executed. If the original point still meets the corresponding conditions in two consecutive updates, it remains unchanged; otherwise, it is replaced with the latest calculation result. This method ensures that the initial speed-stabilizing yield point and the subsequent takeover point always remain consistent with the current passage space state.

[0081] Based on the aforementioned first steady-speed handover point, then return to takeover point, and the aforementioned conflict transmission chain, at least one of the following assisted driving control commands is output: torque limiting braking, lateral constraint, and target yielding.

[0082] Using the initial stable speed handover point as the trigger reference, the timing for entering handover control is determined. The distance from the vehicle's current position to the initial stable speed handover point is calculated and converted into the corresponding travel time based on the vehicle's current speed. When this time is less than 1 second, torque-limiting and mild braking is initiated. The torque-limiting and mild braking is achieved by reducing the drive output in stages: reducing the current drive output by 10% every 0.2 seconds, while simultaneously applying an initial deceleration of 0.5 meters per square second, which is gradually increased by 0.1 meters per square second every 0.2 seconds, so that the vehicle speed smoothly decreases to a passing speed that matches the acceptable handover window.

[0083] Based on the order of targets in the conflict transmission chain, targets that need to be passed first are identified, and the estimated time for them to enter the yieldable passage window is calculated. If this time is earlier than the time it takes for the vehicle to pass through the corresponding section, the current deceleration state is maintained or the speed is further reduced. At the same time, based on the remaining passage width of each section within the yieldable passage window, the lateral position of the vehicle is restricted, controlling the lateral deviation of the vehicle to within half of the remaining width, and adjusting the deviation to the side away from the target by 0.05 meters every 0.1 seconds to avoid occupying the target's passage space.

[0084] When the vehicle's travel time to the return-to-takeoff point is less than 1.5 seconds, the braking intensity is gradually reduced, decreasing the current deceleration by 20% every 0.2 seconds, while simultaneously restoring drive output; at the same time, the lateral constraint range is relaxed, allowing the vehicle to gradually return to its normal driving trajectory. Throughout the process, control commands are continuously modified based on changes in the conflict transmission chain to ensure the continuity and stability of the yielding process.

[0085] When the conflict propagation chain continues to lengthen and the available passage window is smaller than a preset value, a minimum risk parking instruction is output.

[0086] The change in the number of targets in the conflict propagation chain is statistically analyzed over three consecutive frames. When the number of targets increases frame by frame and the new targets are located after existing targets in spatial position, forming a forward-extending sequence structure, the conflict propagation chain is determined to be in a continuously extending state.

[0087] At the same time, the distance between the newly added target and the previous target is recorded. When the distance is less than 1 meter, it is considered that there is a close relationship between the targets, thereby further confirming the trend of continuous occupation of the passage space by the conflict transmission chain.

[0088] Calculate the minimum remaining passage width of each segment in the current passable window and compare it with the vehicle width; at the same time, calculate the total length of consecutive passable segments.

[0089] When the minimum remaining passage width is less than the vehicle width plus a safety margin of 0.2 meters, or the length of the continuous passable section is less than the 2-second travel distance calculated at the current speed, the passable window will be determined to be less than a preset value. The preset value is jointly limited by the minimum passage width and the minimum continuous passage length.

[0090] Under the condition that the conflict transmission chain continues to extend and the yielding passage window is less than a preset value, a minimum-risk stopping command is output. Specifically, torque limiting control is first executed, reducing the drive output to 30% of its original level within 0.5 seconds. Then, a gradually increasing braking deceleration is applied, with an initial value of 1 m / s², increasing by 0.2 m / s² every 0.2 seconds until the vehicle decelerates to a standstill. During deceleration, the vehicle's lateral position is kept stable, and no lateral offset adjustments are performed to avoid lateral risks in confined spaces. Simultaneously, changes in the conflict transmission chain and the yielding passage window are continuously monitored. If the passage conditions return to a safe range during the stopping process, the stopping process is terminated and yielding passage control is re-entered. Through the above methods, minimum-risk stopping control is achieved under complex road conditions.

[0091] Example 2, please refer to Figure 2 As shown in this embodiment, the intelligent connected vehicle intelligent assisted driving system for complex road conditions includes: The perception fusion module acquires vehicle-mounted perception information, roadside cooperative information, and vehicle operating status, extracts boundary segments, slippery signs, and occlusion removal timing, and combines them to form a dynamic passable area. The conflict prediction module identifies targets that are continuously approaching the dynamically passable area based on the dynamic passable area, and generates a conflict propagation chain according to the probability of entry and the timing of the removal of obstruction. The passage window calculation module, based on the conflict transmission chain and slippery signs, segments and compresses the available passage range along the dynamically passable area to obtain a passable window; The yield decision module, in conjunction with the vehicle's current speed and the available yield window, determines the first steady-speed yield point and the subsequent return-to-take-over point. The driver assistance control module, based on the first speed stabilization and yield point, the second return to take over point and the conflict transmission chain, outputs at least one of the following driver assistance control commands: torque limiting and braking, lateral constraint, and target yield. The risk monitoring module outputs a minimum risk parking instruction when the conflict transmission chain continues to extend and the passable window is smaller than a preset value.

[0092] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for intelligent assisted driving of connected vehicles in complex road conditions, characterized in that, include: Acquire vehicle-mounted perception information, roadside cooperative information and vehicle operating status, extract boundary segments, slippery signs and occlusion removal time sequence, and piece them together to form a dynamic passable area; Based on the dynamically accessible area, targets that are continuously approaching the dynamically accessible area are identified, and a conflict propagation chain is generated according to the probability of entry and the timing of occlusion removal. Based on the conflict transmission chain and slippery signs, the available passage range is compressed in segments along the dynamically passable area to obtain a passable window. Based on the vehicle's current speed and the available passage window, determine the initial stable speed handover point and the subsequent return to takeover point; Based on the first stable speed handover point, the second return to takeover point and the conflict transmission chain, at least one of the following assisted driving control commands is output: torque limiting braking, lateral constraint and target yielding. When the conflict propagation chain continues to lengthen and the available passage window is smaller than a preset value, a minimum risk parking instruction is output.

2. The intelligent assisted driving method for intelligent connected vehicles in complex road conditions according to claim 1, characterized in that, The steps for acquiring vehicle-mounted perception information, roadside cooperative information, and vehicle operating status, and extracting boundary segments include: using the timestamps of each frame of the vehicle-mounted perception information as a reference, mapping the data with the closest time in the roadside cooperative information to the reference time; when there is a time interval between two adjacent roadside cooperative information, compensating the corresponding positions of the two roadside cooperative information according to the vehicle speed and heading changes within the time interval, so that they fall into the same ground reference position.

3. The intelligent assisted driving method for intelligent connected vehicles in complex road conditions according to claim 2, characterized in that, The steps for extracting boundary segments also include: after completing segment-by-segment compensation, retaining candidate edges in continuous frames whose center position deviation after vehicle displacement compensation is no greater than 0.15 meters and whose extension direction changes no greater than 10 degrees; when the endpoint spacing of adjacent candidate edges is less than 1 / 4 of the vehicle width and the difference in extension direction is no greater than 10 degrees, splicing adjacent candidate edges into the same continuous boundary; for parts that cannot be directly spliced ​​but are consistent with the edge position indicated by the roadside coordination information, filling in the missing segments according to the edge direction given by the roadside coordination information to obtain boundary segments.

4. The intelligent assisted driving method for intelligent connected vehicles in complex road conditions according to claim 1, characterized in that, The steps for extracting slippery signs include: dividing the road surface within the boundary segment into continuous strips along the vehicle's direction of travel; statistically analyzing the average brightness, brightness distribution fluctuation, number of identifiable texture edges, and continuous length of texture edges for each strip in three consecutive frames; when the same strip shows a continuous increase in average brightness, a continuous decrease in the number of texture edges, and a continuous shortening of continuous texture edge length in three consecutive frames, and the outer edge of the bright area in adjacent strips gradually expands along the vehicle's direction of travel, the corresponding area of ​​that strip is identified as a candidate reflective area; only candidate reflective areas located between or immediately inside the boundary segment are retained, and slippery signs are determined based on their distribution range and expansion trend over time.

5. The intelligent assisted driving method for intelligent connected vehicles in complex road conditions according to claim 1, characterized in that, The steps for generating the occlusion removal sequence and merging it to form a dynamically passable area include: In the area where the slippery signs are located and the adjacent areas, the road surface with occlusion is selected as the occluded area. After the occluded area in the consecutive frames is compensated to the same ground reference position, the process of the background from missing to complete is compared. The residual marking edge, road joint edge, guardrail bottom edge or construction cone bottom edge with a contour position deviation of no more than 0.1 meters in 5 consecutive frames are used as stable structures. The moment of their first appearance is traced back to form the occlusion removal sequence. Each boundary segment is projected onto the same ground reference position and connected to form a continuous boundary zone. The area enclosed by the continuous boundary zone is used as the initial passable range. The boundary is contracted according to the proportion of wet and slippery signs covered by each segment. Based on the time sequence of the removal of obstruction, only the segments that have formed a continuous exposure sequence are retained and combined to form a dynamic passable area.

6. The intelligent assisted driving method for intelligent connected vehicles in complex road conditions according to claim 1, characterized in that, The steps for generating a conflict propagation chain based on entry probability and occlusion removal timing include: A unified position correction is performed on each target in consecutive frames, and the shortest distance from each target to the boundary of the dynamically passable area is calculated. When the shortest distance decreases sequentially in three consecutive frames, or the overall decrease exceeds 0.2 meters, the target is identified as a continuously approaching target. Combining the change in the angle between the current movement direction of each target and the boundary of the dynamically passable area, as well as the exposure order of the corresponding spatial position in the occlusion release sequence, the continuously approaching targets are prioritized for entry. Then, it is determined whether the movement path of the next target passes through the area currently occupied by the previous target or the area that it may occupy in the next two time intervals. If so, a sequential dependency relationship is established between the two, and dependencies that exceed the boundary of the dynamically passable area or are inconsistent with the occlusion release sequence are deleted to obtain the conflict propagation chain.

7. The intelligent assisted driving method for intelligent connected vehicles in complex road conditions according to claim 1, characterized in that, Based on the conflict transmission chain and slippery signs, the available passage range is segmented and compressed along the dynamically passable area to obtain a transferable passage window, including: The dynamically passable area is divided into continuous segments along the direction of vehicle travel, and each segment is marked as an affected segment or an unaffected segment according to the spatial location of each target in the conflict transmission chain. For the affected sections, the lateral occupancy range of each target in the corresponding section is determined in the order of the targets in the conflict transmission chain. An additional 0.2 meters is added to each side of the target’s maximum lateral width as a safety margin, and then deducted from the original passage width one by one. Calculate the distribution ratio of slippery signs for each section, and continue to shrink the passage boundary inward on the side where the slippery signs are located according to the distribution ratio; Finally, the remaining passage areas in each section are checked for connectivity and continuously spliced ​​together to form a passable passage window.

8. The intelligent assisted driving method for intelligent connected vehicles in complex road conditions according to claim 7, characterized in that, Based on the vehicle's current speed and the available passage window, determine the initial stable speed handover point and the subsequent return-to-takeover point, including: The passable passage window is divided into continuous segments along the vehicle's direction of travel. The distance traveled per unit time is calculated based on the vehicle's current speed. Multiple candidate positions are marked sequentially starting from the initial segment of the passable passage window. When the remaining passage width of the segment where the candidate position is located and the two subsequent consecutive segments are not less than the vehicle width and the width difference between adjacent segments does not exceed 0.2 meters, the candidate position is determined as the position that meets the speed stabilization condition. The position closest to the initial segment is selected first as the speed stabilization handover point. Starting from the section where the first stable speed handover point is located, the remaining passage width of subsequent consecutive sections is extracted in sequence. When there are three consecutive sections whose width increases segment by segment and the increase in each segment is not less than 0.15 meters, and the three consecutive sections are not within the influence range of any target in the conflict transmission chain in the next two time intervals, the first position that meets the conditions is determined as the re-takeover point.

9. The intelligent assisted driving method for intelligent connected vehicles in complex road conditions according to claim 1, characterized in that, The steps for outputting a minimum-risk parking instruction when the conditions are met include: When the travel time from the vehicle's current position to the first stable speed yielding point is less than the preset time, torque limiting and slow braking are initiated. Between the first stable speed yielding point and the return to the takeover point, the target that needs to be passed first is determined according to the order of the targets in the conflict transmission chain, and the lateral position of the vehicle is constrained according to the remaining passage width of each section within the yieldable passage window, so as to output at least one of the following assisted driving control commands: torque limiting and slow braking, lateral constraint, and target yielding. When the number of targets in the collision propagation chain increases continuously over multiple consecutive frames, and the minimum remaining passage width of the passable window is less than the sum of the vehicle width and the preset safety margin, or the length of the continuous passable section is less than the preset travel distance calculated based on the current speed, a minimum risk stop command is output.

10. A smart connected vehicle intelligent assisted driving system for complex road conditions, used to implement the smart connected vehicle intelligent assisted driving method for complex road conditions as described in any one of claims 1-9, characterized in that, include: The perception fusion module acquires vehicle-mounted perception information, roadside cooperative information, and vehicle operating status, extracts boundary segments, slippery signs, and occlusion removal timing, and combines them to form a dynamic passable area. The conflict prediction module identifies targets that are continuously approaching the dynamically passable area based on the dynamic passable area, and generates a conflict propagation chain according to the probability of entry and the timing of the removal of obstruction. The passage window calculation module, based on the conflict transmission chain and slippery signs, segments and compresses the available passage range along the dynamically passable area to obtain a passable window; The yield decision module, in conjunction with the vehicle's current speed and the available yield window, determines the first steady-speed yield point and the subsequent return-to-take-over point. The driver assistance control module, based on the first speed stabilization and yield point, the second return to take over point and the conflict transmission chain, outputs at least one of the following driver assistance control commands: torque limiting and braking, lateral constraint, and target yield. The risk monitoring module outputs a minimum risk parking instruction when the conflict transmission chain continues to extend and the passable window is smaller than a preset value.

Images