Vehicle driving speed planning method and device, vehicle, medium and program product

Abstract

The invention provides a vehicle driving speed planning method and device, a vehicle, a medium and a program product, and belongs to the field of automatic driving. The vehicle driving speed planning method comprises the steps that all obstacles around a target vehicle are traversed, and the types of the obstacles are determined based on the relative position relation between the obstacles and a target vehicle driving lane; calculating a first target vehicle speed according to the obstacle type and updating a global target vehicle speed; acquiring an obstacle prediction track, and if the confidence coefficient of the prediction track meets a preset requirement, determining a collision risk point based on the prediction track and the target vehicle planning path; calculating a second target vehicle speed based on the collision risk point and updating the global target vehicle speed; and after the traversal is completed, outputting the minimum global target vehicle speed in the traversal process as the reference vehicle speed. According to the technical scheme provided by the invention, the risk assessment based on the lane relationship and the collision risk assessment based on the predicted trajectory are fused, and the driving safety and comfort of the automatic driving vehicle in a complex dynamic scene can be improved.

Description

Technical Field

[0001] This invention belongs to the field of autonomous driving technology, specifically relating to a vehicle driving speed planning method, device, vehicle, medium, and program product. Background Technology

[0002] Autonomous vehicles typically need to plan speed based on path planning, enabling them to make reasonable decisions when encountering vehicles, pedestrians, non-motorized vehicles, and roadside obstacles.

[0003] In existing technologies, common speed planning methods include graph search methods based on ST graphs and game-theoretic decision-making methods. ST graph-based solutions typically model obstacles on a time-distance plane and search for and optimize different speed trajectories using a cost function. These methods can provide feasible solutions when obstacle movement is relatively regular and behavior patterns are relatively stable. However, when pedestrians, non-motorized vehicles, or other vehicles suddenly encroach on the target vehicle's lane, or move irregularly at the lane edge, the feasible area on the ST graph can change drastically in a short period, easily leading to significant deceleration or even sudden braking in the planning results.

[0004] Game-theoretic speed planning methods typically treat the target vehicle and surrounding vehicles as multiple players in a game, predicting the behavior of each player and solving for the game equilibrium to obtain speed decisions. While theoretically capable of characterizing multi-vehicle interactions, these methods heavily rely on predictions of obstacle behavior or trajectories. When the prediction module faces obstacles with high motion uncertainty, such as pedestrians or non-motorized vehicles, the predicted trajectory may change significantly between different cycles, causing frequent fluctuations in speed planning results between adjacent cycles. This manifests as repeated acceleration and deceleration of the vehicle, making it difficult to maintain a stable speed. Summary of the Invention

[0005] The purpose of this disclosure is to provide a vehicle driving speed planning method, device, vehicle, medium, and program product that can improve the driving safety and comfort of vehicles in complex dynamic environments.

[0006] To achieve the above objectives, the technical solution provided in this disclosure is as follows:

[0007] In a first aspect, this disclosure provides a vehicle driving speed planning method, which includes:

[0008] The system iterates through all obstacles surrounding the target vehicle. For each obstacle: based on its relative position to the target vehicle's lane, the type of obstacle is determined; for each obstacle, a first target speed is calculated based on its type, and the global target speed of the target vehicle is updated based on the first target speed; for each obstacle, a predicted trajectory is obtained; if the confidence level of the predicted trajectory meets a preset requirement, a collision risk point between the obstacle and the target vehicle is determined based on the predicted trajectory and the planned path of the target vehicle; for each obstacle, a second target speed is calculated based on the collision risk point, and the global target speed of the target vehicle is updated based on the second target speed; after traversing all obstacles, the minimum global target speed among all the updated global target speeds during the traversal is used as the reference speed of the target vehicle.

[0009] In one or more embodiments, the type of obstacle is determined based on the relative positional relationship between the obstacle and the target vehicle's driving lane, including: based on the relationship between the obstacle's outline and the lane boundary of the target vehicle's driving lane, obstacles whose outlines are completely outside the lane boundary are identified as low-risk obstacles, obstacles whose outlines are partially inside the lane boundary are identified as intrusive obstacles, and obstacles whose outlines are completely inside the lane boundary are identified as lane-occupying obstacles; wherein, the low-risk obstacles are not involved in the calculation of the first target vehicle speed, the second target vehicle speed, or the update of the global target vehicle speed.

[0010] In one or more embodiments, when the obstacle is an intruding obstacle, the calculation of the first target vehicle speed includes: establishing a Frenet coordinate system based on the center reference line of the target vehicle's driving lane; calculating the intrusion distance of the intruding obstacle relative to the lane boundary in the lateral direction; dividing a preset longitudinal range in front of the target vehicle into a near distance range and a far distance range in the longitudinal direction, and determining the range where the intruding obstacle is located based on the longitudinal distance between the target vehicle and the intruding obstacle; when the intruding obstacle is in the near distance range, calculating the first target vehicle speed based on the intrusion distance, the target vehicle's current speed, and the maximum speed limit of the driving lane using a first intrusion coefficient; when the intruding obstacle is in the far distance range, calculating the first target vehicle speed based on the intrusion distance, the target vehicle's current speed, and the maximum speed limit of the driving lane using a second intrusion coefficient, wherein the speed limit of the second intrusion coefficient is less than that of the first intrusion coefficient.

[0011] In one or more embodiments, when the obstacle is a road-occupying obstacle, the calculation method of the first target vehicle speed includes: calculating the relative speed between the target vehicle and the road-occupying obstacle in the longitudinal direction of the driving lane; when the relative speed is greater than a preset relative speed threshold, taking the current speed of the target vehicle as the first target vehicle speed; when the relative speed is less than or equal to the preset relative speed threshold, calculating the safe distance of the target vehicle based on the preset maximum comfort deceleration and compensation distance of the target vehicle, and determining the first target vehicle speed based on the longitudinal distance between the target vehicle and the road-occupying obstacle and the safe distance.

[0012] In one or more embodiments, determining a first target speed based on the longitudinal distance between the target vehicle and the obstacle and the safe distance includes: when the longitudinal distance between the target vehicle and the obstacle is greater than the safe distance, using the current speed of the target vehicle as the first target speed of the obstacle; when the longitudinal distance between the target vehicle and the obstacle is less than or equal to the safe distance, using the longitudinal component speed of the obstacle as the first target speed.

[0013] In one or more embodiments, determining the collision risk point between the obstacle and the target vehicle based on the predicted trajectory and the planned path of the target vehicle includes: for each predicted point on the predicted trajectory, finding the path point with the shortest distance to the predicted point on the planned path of the target vehicle as the nearest path point corresponding to the predicted point; if the predicted point and its corresponding nearest path point meet preset risk judgment conditions, then the nearest path point is determined as a collision risk point; wherein, the risk judgment conditions include at least: the distance between the predicted point and its corresponding nearest path point is less than a first distance threshold; when the obstacle is at the predicted point and the target vehicle is at the corresponding nearest path point, the minimum distance between the outline of the obstacle and the outline of the target vehicle is less than a second distance threshold; the minimum distance between the nearest path point and the target vehicle is less than a third distance threshold; and the time difference between the estimated time for the obstacle to reach the predicted point and the estimated time for the target vehicle to reach the corresponding nearest path point falls within a preset range.

[0014] In one or more embodiments, calculating a second target vehicle speed based on the collision risk point includes: obtaining the path length from the target vehicle to the collision risk point, the current speed of the target vehicle, the estimated arrival time from the obstacle to the predicted point corresponding to the collision risk point, and a safety time margin; assuming that the target vehicle travels along the planned path with uniform deceleration, determining the expected arrival time of the target vehicle to the collision risk point, such that the difference between the expected arrival time and the estimated arrival time of the obstacle is greater than the safety time margin; and calculating the second target vehicle speed based on the path length and the current speed of the target vehicle.

[0015] Secondly, this disclosure provides a vehicle driving speed planning device, comprising: an obstacle determination module, a first calculation module, a collision risk determination module, a second calculation module, and an output module; the obstacle determination module is used to traverse all obstacles around a target vehicle, and for each obstacle, determine the type of the obstacle based on the relative positional relationship between the obstacle and the target vehicle's driving lane; the first calculation module is used to calculate a first target vehicle speed based on the type of the obstacle, and update the global target vehicle speed of the target vehicle based on the first target vehicle speed; the collision risk determination module is used to obtain the predicted trajectory of the obstacle, and if the confidence level of the predicted trajectory meets a preset requirement, determine the collision risk point between the obstacle and the target vehicle based on the predicted trajectory and the planned path of the target vehicle; the second calculation module is used to calculate a second target vehicle speed based on the collision risk point, and update the global target vehicle speed of the target vehicle based on the second target vehicle speed; the output module is used to output the minimum global target vehicle speed among all the global target vehicle speeds updated during the traversal as the reference speed of the target vehicle after traversing all obstacles.

[0016] Thirdly, this disclosure provides an autonomous vehicle, including a memory, a processor, and a program stored in the memory and executable on the processor, wherein the processor executes the program to implement the vehicle driving speed planning method as described above.

[0017] Fourthly, this disclosure provides a machine-readable medium carrying executable instructions, which, when executed by a processor, are used to implement the vehicle driving speed planning method as described above.

[0018] Fifthly, this disclosure provides a computer program product, which includes computer instructions that, when executed by a processor, implement the vehicle driving speed planning method as described above.

[0019] The vehicle driving speed planning method, device, vehicle, readable medium, and program product disclosed herein integrate lane relationship-based risk assessment with collision risk assessment based on predicted trajectory to form a unified constraint on the global target vehicle speed. This enables speed planning to simultaneously consider lane intrusion risk and collision interaction risk. By updating the target vehicle speed of each obstacle with the current global target vehicle speed and outputting the minimum global target vehicle speed as a reference speed, it ensures that the strictest safety constraints are followed in the case of multiple obstacles, thereby improving the driving safety and comfort of autonomous vehicles in complex dynamic scenarios. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments recorded in this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart of a vehicle driving speed planning method in one embodiment of the present disclosure;

[0022] Figure 2 This is an execution logic diagram of a vehicle driving speed planning method in one embodiment of the present disclosure;

[0023] Figure 3 This is a schematic diagram of a risk assessment based on lane relationships in one embodiment of this disclosure;

[0024] Figure 4 This is a schematic diagram of a collision risk assessment in one embodiment of the present disclosure;

[0025] Figure 5 This is a schematic diagram of a vehicle driving speed planning device in one embodiment of the present disclosure;

[0026] Figure 6 This is a schematic diagram of an autonomous vehicle according to one embodiment of the present disclosure. Detailed Implementation

[0027] To enable those skilled in the art to better understand the technical solutions in this disclosure, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this disclosure.

[0028] In the real-world operating environment of autonomous vehicles, speed planning must satisfy traffic safety while also considering comfort and driving efficiency. Especially in complex and dynamic road scenarios, target vehicles often encounter traffic participants with significant behavioral uncertainty, such as vehicles, pedestrians, and other moving objects that suddenly approach or cut into the vehicle's lane. These objects may change their relative position to the lane in a short period of time, and are also susceptible to occlusion, limited visibility, or interactive influences, exhibiting unpredictable and unstable movement trends. This creates an inherent contradiction between proactive prevention and avoiding overly conservative speed planning.

[0029] To address the aforementioned technical challenges, one approach tends to complete risk assessment and speed decisions within a single modeling framework, primarily relying on spatiotemporal graph search or unified cost optimization to generate speed curves. This type of method performs well when obstacle behavior is relatively predictable, but when sudden incursions, edge-hugging maneuvers, or lateral intrusions occur, risk constraints change drastically within a short period. Speed ​​planning can easily lead to abrupt deceleration or even sudden braking, resulting in decreased comfort and inconsistencies with human driving intuition. Another approach emphasizes interactive inference, often relying on predictions of future obstacle behavior for decision-making. When predictions are unstable due to occlusion or random behavior, speed decisions fluctuate with the prediction output, manifesting as frequent acceleration and deceleration adjustments. This leads to unstable control, inconsistent experience, and even decision-making instability when multiple objectives coexist.

[0030] Based on the above understanding, this disclosure proposes a speed planning approach for real traffic uncertainties. Its core is to decompose the risk constraints in speed planning into multiple levels with different credible sources and applicable boundaries, and form a stable and safe speed reference for speed planning through a unified fusion mechanism.

[0031] The basic principle of this approach is to first establish a risk hierarchy by assessing the relationship between obstacles and lanes in the environment, thereby forming a fundamental constraint on the target vehicle speed. This enables the vehicle to make timely preventative speed adjustments when high-impact situations such as sudden intrusions or lane occupancy occur. Building upon this foundation, for obstacles that provide reliable future trend information, a risk assessment based on the relationship between the predicted trajectory and the vehicle's intended movement is introduced to analyze the risk intensity at potential conflict locations, thus forming a further speed constraint. In this way, this disclosure integrates speed planning mechanisms from different perspectives, avoiding both sudden deceleration due to a lack of preventative measures and speed fluctuations caused by over-reliance on unstable predictions.

[0032] Please refer to Figure 1 and Figure 2 As shown, a vehicle driving speed planning method in one embodiment of this disclosure specifically includes the following steps:

[0033] S101: Traverse all obstacles around the target vehicle. For each obstacle, determine the type of obstacle based on the relative position of the obstacle to the target vehicle's driving lane.

[0034] In step S101, the perception and prediction results of the autonomous driving system can be used as input, including the state information of the target vehicle, obstacle information in the environment, and map information. Based on the environmental perception results, a set of obstacles to be processed is obtained, and each obstacle around the target vehicle is traversed to establish the object set and risk priority basis for subsequent speed constraint calculations.

[0035] The obstacles surrounding the target vehicle can be moving objects such as vehicles, pedestrians, and robots. Specifically, an obstacle list can be generated by the onboard perception system, where each obstacle can correspond to state information with a timestamp and outline information. In order to ensure that the subsequent speed constraints are consistent with the road constraints currently being traveled by the target vehicle, step S101 uses the target vehicle's driving lane as a determination benchmark and determines the type of obstacle by the relative positional relationship between the obstacle and the target vehicle's driving lane.

[0036] In one feasible implementation, lane center reference lines and lane boundary information can be used to establish a coordinate mapping relationship based on the center reference lines, allowing the outline of an obstacle to be projected or mapped onto a lane reference system consistent with the target vehicle's driving lane. Based on this reference system, the relative positional relationship between the obstacle and the target vehicle's driving lane can be characterized by the spatial relationship between the obstacle outline and the lane boundary of the target vehicle's driving lane.

[0037] Specifically, the target vehicle's driving lane can be defined by the left and right lane boundaries. The obstacle profile can be represented by its outer contour on the ground plane; this outer contour can be a polygon, a circumscribed rectangle, or a closed curve obtained by fitting several key points. The type of obstacle is determined by analyzing the positional relationship between the obstacle profile and the lane boundary. The obstacle type is used to distinguish between obstacles that directly affect the target vehicle's current driving lane and those that have a weaker impact, allowing for different speed limit logic to be applied to different types when calculating the first target speed.

[0038] In one exemplary embodiment, determining the type of obstacle based on the relative positional relationship between the obstacle and the target vehicle's driving lane includes: determining obstacles whose outlines are completely outside the lane boundary as low-risk obstacles, obstacles whose outlines are partially inside the lane boundary as intrusion obstacles, and obstacles whose outlines are completely inside the lane boundary as lane-occupying obstacles, based on the relationship between the obstacle's outline and the lane boundary of the target vehicle's driving lane; wherein, the low-risk obstacles are not involved in the calculation of the first target vehicle speed, the second target vehicle speed, or the update of the global target vehicle speed.

[0039] By utilizing the spatial relationship between the outline of obstacles and the lane boundary of the target vehicle's driving lane, obstacles are distinguished into low-risk obstacles, intruding obstacles, and lane-occupying obstacles. This provides a prerequisite for the differentiated calculation of the initial target vehicle speed. Furthermore, by removing low-risk obstacles from the global target vehicle speed update chain, excessive interference from irrelevant targets is avoided in speed planning. The target vehicle's driving lane can be determined by road reference lines, lane centerlines, and corresponding lane boundary descriptions. Lane boundaries can be generated from map data or road perception results. The lane boundaries define the range of lanes that the target vehicle can travel in, and subsequent judgments are all based on this range.

[0040] In one feasible implementation, the lane boundary can be fitted as a straight line, polyline, or curve using a series of discrete sampling points, and mapped together with the obstacle contour to a unified coordinate system for relationship determination. This unified coordinate system can be a lane reference coordinate system based on road reference lines, or it can be a vehicle local coordinate system or a global geographic coordinate system.

[0041] In specific assessments, the lane area defined by the lane boundary can be considered a closed or strip-shaped region. In-point testing can be performed on each boundary point of the obstacle profile, or the intersection / merger relationship between the obstacle profile and the lane area can be calculated. When all points of the obstacle profile are outside the lane area and there is no overlap between the obstacle profile and the lane area, the obstacle is identified as a low-risk obstacle. When some points of the obstacle profile are within the lane area, and there is overlap between the obstacle profile and the lane area, but the obstacle profile does not completely fall within the lane area, the obstacle is identified as an intruding obstacle. When the entire obstacle profile is contained within the lane area and is completely within the lane area, the obstacle is identified as a lane-occupying obstacle. To enhance safety, the obstacle profile can be expanded to more conservatively identify intruding and lane-occupying obstacles in the presence of perception or positioning errors, thereby improving the reliability of risk assessment.

[0042] Through the aforementioned type-based determination mechanism, risk stratification of environmental targets can be completed in the initial stage of speed planning, enabling subsequent speed constraint calculations to have clear applicable objects and differentiation. Low-risk obstacles are excluded from the global target vehicle speed update, which can reduce speed fluctuations caused by irrelevant targets and avoid frequent deceleration of target vehicles or unnecessary reduction in traffic efficiency due to oversensitivity to targets outside the lane. Intruding obstacles and lane-occupying obstacles are identified separately, enabling target vehicles to establish a smoother speed response to lane intrusion or lane occupancy risks, thereby improving ride comfort while ensuring safety, and making the update of the global target vehicle speed more in line with the intuitive judgment of human drivers regarding lane risks.

[0043] S102: For each obstacle, calculate a first target vehicle speed based on the type of the obstacle, and update the global target vehicle speed based on the first target vehicle speed.

[0044] Step S102 is used to, after the obstacle type is determined, convert the risk relationship between the obstacle and the target vehicle's driving lane into a first target speed for constrained speed planning for each obstacle, and integrate the first target speed into the target vehicle's global target speed according to a preset update rule, so that even if the obstacle prediction trajectory has not been intervened or the prediction confidence is insufficient, it is still possible to form a preventive speed limit for the target vehicle based on the real-time relationship between lane occupation and intrusion.

[0045] Different types of obstacles affect the speed planning of target vehicles in different ways. Low-risk obstacles do not participate in speed constraints outside the lane boundary. Intruding obstacles indicate that the obstacle has a lateral (perpendicular to the lane extension direction) intrusion tendency into the target vehicle's driving lane. The speed can be reduced in advance according to the degree of intrusion and the longitudinal (lane extension direction) proximity of the obstacle to reduce the probability of conflict. Obstacles occupying the lane indicate that the obstacle has occupied the target vehicle's driving lane and may pose a risk of rear-end collision or forced stop. The speed limit can be constrained according to the longitudinal distance and relative speed relationship with the target vehicle, so that the target vehicle can enter a controllable following or deceleration state earlier.

[0046] In one feasible implementation, when the obstacle type is an intruding obstacle, the first target vehicle speed can be designed to decrease monotonically with the degree of intrusion. The degree of intrusion can be characterized by the overlap depth or intrusion distance between the intruding obstacle profile and the lane boundary of the target vehicle's driving lane, and the speed limit intensity can also be determined by combining the longitudinal distance between the intruding obstacle and the target vehicle. When the obstacle type is a lane-occupying obstacle, the calculation of the first target vehicle speed can be modeled around the longitudinal rear-end collision risk. Since the lane-occupying obstacle is already located in the target vehicle's driving lane, the target vehicle needs to determine whether there is a risk of needing to brake in advance based on the longitudinal distance and longitudinal relative speed with the lane-occupying obstacle.

[0047] In an exemplary embodiment, when the obstacle is an intruding obstacle, the calculation method for the first target vehicle speed includes: establishing a Frenet coordinate system based on the center reference line of the target vehicle's driving lane; calculating the intrusion distance of the intruding obstacle relative to the lane boundary in the lateral direction; dividing a preset longitudinal range in front of the target vehicle into a near distance range and a far distance range in the longitudinal direction, and determining the range where the intruding obstacle is located based on the longitudinal distance between the target vehicle and the intruding obstacle; when the intruding obstacle is in the near distance range, calculating the first target vehicle speed based on the intrusion distance, the target vehicle's current speed, and the maximum speed limit of the driving lane using a first intrusion coefficient; when the intruding obstacle is in the far distance range, calculating the first target vehicle speed based on the intrusion distance, the target vehicle's current speed, and the maximum speed limit of the driving lane using a second intrusion coefficient, wherein the speed limit of the second intrusion coefficient is less than that of the first intrusion coefficient.

[0048] Please refer to Figure 3 As shown, the Frenet coordinate system uses the center reference line of the target vehicle's driving lane as the reference for the vertical axis, making the target vehicle (such as...) the reference line for the vertical axis. Figure 3 (in the context of vehicles) and intruding obstacles (such as...) Figure 3 The relative relationship between intruding vehicles 1 and 2 in the longitudinal and lateral directions of the lane can be characterized by the same reference frame, where the longitudinal distance is used to characterize the proximity of the intruding obstacle and the target vehicle in the direction of lane extension, and the lateral distance is used to characterize the degree of intrusion of the intruding obstacle into the lane in which the target vehicle is traveling.

[0049] In the horizontal direction ( Figure 3 In the L-direction, the intrusion intensity is characterized by calculating the intrusion distance of the intruding obstacle relative to the lane boundary. The intrusion distance can be understood as the depth to which the intruding obstacle's outline enters the lane boundary's defined area in the lateral direction. It can be calculated based on the overlap between the intruding obstacle's outline and the lane boundary, for example, by calculating the lateral difference between the innermost lateral boundary point of the intruding obstacle's outline and the adjacent lane boundary, or by calculating the overlap width between the intruding obstacle's outline and the lane boundary's defined area. A larger intrusion distance indicates that the intruding obstacle penetrates deeper into the target vehicle's lane in the lateral direction, resulting in a higher probability of potential conflict and requiring stricter constraints on the target vehicle's speed planning.

[0050] Specifically, for any intruding obstacle, the intrusion distance is calculated. The following formula can be used:

[0051]

[0052] This refers to the distance from the left lane boundary to the lane center reference line. This refers to the distance from the right lane boundary to the lane center reference line. It refers to the shortest distance from the obstacle to the center reference line of the lane (which can be obtained by mapping several sampling points on the obstacle outline to the Frenet coordinate system and taking the minimum value of the absolute value of the lateral coordinate).

[0053] In the vertical direction ( Figure 3 (In the S-direction), a preset longitudinal range in front of the target vehicle can be divided into a near-distance range and a far-distance range. The longitudinal distance between the target vehicle and the intruding obstacle can be calculated, and the range in which the intruding obstacle is located can be determined based on this longitudinal distance. The preset longitudinal range can be understood as the effective forward risk zone that speed planning focuses on, and its upper limit can be determined based on the target vehicle's current speed, road scene, and system configuration.

[0054] Divide the preset vertical range into close-range intervals (e.g.) Figure 3 middle / 2 distance interval) and long distance interval (such as Figure 3 middle / 2~ The purpose of the distance range is to differentiate between situations with different risk time margins: when the intruding obstacle is in the close range, the target vehicle has less time to respond, and the intrusion is more likely to escalate into a conflict in a short time, thus requiring a stronger speed limit; when the intruding obstacle is in the far range, the target vehicle has more time to smoothly adjust its speed in advance, and the response to the intrusion risk can be more gentle to avoid unnecessary excessive deceleration. The longitudinal distance can be calculated based on the projection points of the target vehicle and the intruding obstacle on the lane center reference line in the Frenet coordinate system (e.g., ...). Figure 3 The distance can be obtained by the longitudinal coordinate difference of projection points 1 and 2 in the image, or by using the longitudinal distance between the boundary of the intrusion obstacle profile that is closest to the target vehicle in the longitudinal direction and the front boundary of the target vehicle, so as to achieve a more conservative distance estimation.

[0055] Based on the aforementioned lateral intrusion distance and longitudinal zone division, for intruding obstacles in the near-distance zone, the first target speed is determined by the intrusion distance, the target vehicle's current speed, and the maximum speed limit of the driving lane, with the speed limit intensity reflected by a first intrusion coefficient. This design ensures that the first target speed will not exceed the maximum speed limit of the driving lane and can be reduced more significantly as the intrusion distance increases, thus forming a stronger preventative deceleration. For intruding obstacles in the far-distance zone, the first target speed is also determined by the intrusion distance, the target vehicle's current speed, and the maximum speed limit of the driving lane, but a second intrusion coefficient reflects a more moderate speed limit intensity. This allows the risk of long-distance intrusion to guide the target vehicle to slow down earlier in a gentler manner, reducing abrupt changes in the speed curve. The speed limit intensity of the second intrusion coefficient is less than that of the first intrusion coefficient. Under the same intrusion distance conditions, the attenuation of the first target speed is smaller in the far-distance zone, which can avoid excessive deceleration at a greater distance, leading to decreased traffic efficiency or abnormal following behavior.

[0056] Specifically, the first target vehicle speed is calculated based on the longitudinal distance between the intruding obstacle and the target vehicle, taking into account the longitudinal distance in front of the target vehicle. The first target vehicle speed for an intruding obstacle within / 2 is calculated using the following formula:

[0057]

[0058] For longitudinal distance in / 2~ The speed of the first target vehicle relative to the intruding obstacle is calculated using the following formula:

[0059]

[0060] in, The target vehicle's current speed. The first invasive coefficient, This is the second intrusion factor. The specific values ​​of these intrusion factors can be set according to the width of the driving lane and the current speed of the target vehicle. This represents the maximum speed limit allowed in the driving lane.

[0061] In an exemplary embodiment, when the obstacle is a road-occupying obstacle, the calculation method of the first target vehicle speed includes: calculating the relative speed between the target vehicle and the road-occupying obstacle in the longitudinal direction of the driving lane; when the relative speed is greater than a preset relative speed threshold, taking the current speed of the target vehicle as the first target vehicle speed; when the relative speed is less than or equal to the preset relative speed threshold, calculating the safe distance of the target vehicle based on the preset maximum comfort deceleration and compensation distance of the target vehicle, and determining the first target vehicle speed based on the longitudinal distance between the target vehicle and the road-occupying obstacle and the safe distance.

[0062] In one feasible implementation, the velocity component of the obstacle can be obtained by projecting the velocity vector output by the sensing module longitudinally. The target vehicle's current speed can also be projected in the same direction to obtain the longitudinal relative velocity. The relative velocity reflects the trend of the longitudinal distance between the target vehicle and the obstacle over time: when the relative velocity is negative or small, it indicates that the obstacle is not moving away from the target vehicle longitudinally, and the target vehicle is gradually approaching, potentially leading to a rear-end collision; when the relative velocity is large and positive, it indicates that the obstacle is moving away from the target vehicle longitudinally, the longitudinal distance tends to increase, and the risk of a rear-end collision is low.

[0063] Based on the aforementioned relative speed, the calculation logic sets a preset relative speed threshold to distinguish between situations requiring preventative speed limits and those not requiring them. When the relative speed exceeds the preset threshold, the obstacle has a sufficient speed advantage over the target vehicle. The longitudinal distance between the target vehicle and the obstacle will gradually increase or at least not decrease rapidly. In this case, using the target vehicle's current speed as the primary target speed avoids unnecessary speed suppression, making speed planning closer to a human driver's intuitive judgment of the vehicle ahead being faster and the distance increasing, thus avoiding the decrease in traffic efficiency caused by slowing down. The selection of the preset relative speed threshold can be related to the target vehicle's current speed, road speed limits, and system safety strategies. For example, a higher threshold can be set at high speeds to avoid oversensitivity to small fluctuations in relative speed, while a lower threshold can be set in low-speed congestion to ensure following stability.

[0064] When the relative speed is less than or equal to a preset relative speed threshold, and the obstacle does not show a significant tendency to increase its distance, the target vehicle needs further evaluation to determine whether it can safely decelerate under comfortable braking conditions to avoid a rear-end collision. In this case, the safe distance required for the target vehicle to achieve comfortable deceleration is calculated based on the target vehicle's maximum permissible comfortable deceleration and the compensation distance. The maximum comfortable deceleration can be understood as the upper limit of braking deceleration that the target vehicle can continuously apply within the range allowed by ride comfort. This upper limit of braking deceleration can be predetermined through road tests or calibration data. The compensation distance is used to cover additional safety margins caused by uncertainties such as perception errors, control lag, and changes in road adhesion.

[0065] The safe distance can be calculated using kinematic relationships, namely, the braking distance required for the target vehicle to converge its relative speed to a safe state through comfortable deceleration at the current relative speed, and then the compensation distance is added to form the final safe distance. Subsequently, a first target speed is determined based on the longitudinal distance between the target vehicle and the obstacle occupying the lane (this longitudinal distance can be calculated in the same way as the longitudinal distance between the target vehicle and the intruding obstacle mentioned above) and the safe distance, so that the first target speed can respond to the urgency of the risk.

[0066] Specifically, relative vehicle speed The following formula can be used for calculation:

[0067]

[0068] in, The absolute speed of the obstruction. The absolute speed of the target vehicle. These are the heading angles of the obstructing obstacle and the target vehicle relative to the lane center reference line, respectively.

[0069] In one exemplary embodiment, determining a first target vehicle speed based on the longitudinal distance between the target vehicle and the obstacle occupying the road and the safe distance includes: when the longitudinal distance between the target vehicle and the obstacle occupying the road is greater than the safe distance, using the current speed of the target vehicle as the first target vehicle speed of the obstacle occupying the road; when the longitudinal distance between the target vehicle and the obstacle occupying the road is less than or equal to the safe distance, using the longitudinal component speed of the obstacle occupying the road as the first target vehicle speed.

[0070] When the longitudinal distance between the target vehicle and the obstacle is greater than the safe distance, it indicates that the target vehicle still has sufficient space to decelerate or follow under comfortable deceleration conditions. Therefore, the initial target speed can be maintained at the target vehicle's current speed, thus avoiding excessive deceleration prematurely when the risk is not imminent. When the longitudinal distance between the target vehicle and the obstacle is less than or equal to the safe distance, it indicates that maintaining the current speed may require applying braking exceeding the comfort threshold within a shorter distance to avoid a rear-end collision. In this case, the initial target speed should be limited to a more conservative value, such as not exceeding the speed component of the obstacle along its direction of travel, thereby prompting the target vehicle to reduce its speed earlier and enter a stable following position, reducing the probability of sudden braking or collision.

[0071] Specifically, to prevent the target vehicle from experiencing sudden braking or other situations, assume that the maximum permissible comfortable braking deceleration of the target vehicle is... So at the current relative vehicle speed The target vehicle decelerates at this speed. Safety distance required for uniform deceleration braking As shown in the following formula, where To supplement the distance:

[0072]

[0073] Then, based on the current longitudinal distance between the target vehicle and the obstacle blocking the road. Make a judgment:

[0074] like The first target vehicle speed is taken as the current speed of the target vehicle. This indicates that preventative speed planning is not currently necessary; if The first target vehicle speed is selected as follows: .

[0075] Since multiple intruding and lane-occupying obstacles often coexist around a target vehicle, and different obstacles pose varying degrees of risk to the target vehicle, imposing speed limits only on a single obstacle while ignoring others can easily lead to insufficient speed planning response to high-risk targets. Conversely, simply applying speed limits to all obstacles may be overly conservative and cause unnecessary deceleration. Therefore, a step-by-step convergence approach can be adopted, starting with minimizing the speed by type, then minimizing across different types, and finally minimizing the speed globally. This ensures that the final target speed output covers the most stringent safety constraints under the current lane relationship.

[0076] Specifically, based on the various types of obstacles mentioned above, the first target vehicle speed based on the lane relationships in the Frenet coordinate system can be obtained as follows:

[0077]

[0078] in, This can be understood as the minimum first target vehicle speed of each of the intruding obstacles that meet the close-range conditions. This can be understood as the minimum first target vehicle speed of each of the intruding obstacles that meet the long-distance interval conditions. This can be understood as the minimum initial target speed of each of the obstacles occupying the road.

[0079] After completing the above calculations, update the global target vehicle speed in step S102 according to the following formula:

[0080]

[0081] in, The current global target vehicle speed is maintained. This means that the global target vehicle speed always takes the smaller value when traversing obstacles or when different constraint modules output speed suggestions, so as to ensure that the global target vehicle speed simultaneously meets all the risk constraints that have been considered.

[0082] S103: For each obstacle, obtain the predicted trajectory of the obstacle. If the confidence level of the predicted trajectory meets the preset requirements, then based on the predicted trajectory and the planned path of the target vehicle, determine the collision risk point between the obstacle and the target vehicle.

[0083] Step S103 is the second-level risk verification step. After completing the first target vehicle speed constraint based on the lane relationship in the Frenet coordinate system, this step introduces the predicted trajectory of the obstacle for each obstacle to further identify potential collision risks, thereby supplementing the shortcomings of lane relationship risk assessment at a level that is closer to real spatial interaction.

[0084] In this step, for each obstacle, the predicted trajectory of the obstacle is first obtained. The predicted trajectory can be understood as the possible movement path of the obstacle within a preset time range in the future. The predicted trajectory can be output by the perception prediction module, and can be a single most probable trajectory or a set of multiple candidate trajectories. Each candidate trajectory can be composed of a series of prediction points, which can include information such as position, speed, heading, and timestamp. Since the prediction results may be significantly unstable when there is occlusion, interaction uncertainty, or large sensor noise, step S103 does not directly use all predicted trajectories for speed constraints. Instead, it introduces the confidence level of the predicted trajectory as a gating condition, so that only the predicted trajectory whose confidence level meets the preset requirements enters the subsequent collision risk point determination process. When the confidence level of the predicted trajectory does not meet the preset requirements, the method directly returns to step S101 to traverse the next obstacle, thereby avoiding unnecessary interference from low-confidence predictions on speed planning.

[0085] In one feasible implementation, the confidence level of the predicted trajectory can be characterized by the probability value, distribution entropy, trajectory consistency score, or historical continuity index provided by the prediction module. For example, when the prediction module outputs multiple candidate trajectories, the probability of the trajectory with the highest probability can be used as the confidence level, or the concentration of the probability distribution of the candidate trajectories can be used as the confidence level, with a higher concentration of probability indicating a more certain prediction. When the prediction module outputs a single trajectory, the stability of the trajectory within a continuous planning period can be used for evaluation. For example, if changes in trajectory shape, keypoint drift, or speed prediction are within a threshold, the confidence level is considered high. The preset requirement can be a fixed threshold or a threshold that is dynamically adjusted according to the scenario. A fixed threshold is suitable for scenarios with relatively stable system configuration, while a dynamic threshold is suitable for scenarios with significant changes in occlusion, road structure complexity, or traffic density. For example, the preset requirement can be increased at intersections and in densely populated pedestrian areas to suppress misjudgments, while the preset requirement can be appropriately reduced on highway straight roads to improve advance prediction.

[0086] When the confidence level of the predicted trajectory meets the preset requirements, step S103 determines the collision risk points between the obstacle and the target vehicle based on the predicted trajectory and the planned path of the target vehicle. The planned path of the target vehicle can be understood as the future driving trajectory or reference path point sequence of the target vehicle output by the path planning module. The planned path usually contains a series of path points with arc length parameters and is accompanied by time or speed planning information.

[0087] Collision risk points characterize locations in the future time domain where obstacles and target vehicles may come into close contact or overlap in space, serving as crucial inputs for subsequent calculations of the second target vehicle speed. To determine collision risk points, a correspondence can be established between each predicted point on the predicted trajectory and the path points on the planned path. For example, for each predicted point on the obstacle's predicted trajectory, the nearest path point on the target vehicle's planned path is found as a corresponding candidate risk path point, and this candidate risk path point is then jointly determined spatially and temporally in a Cartesian coordinate system.

[0088] In one exemplary embodiment, determining the collision risk point between the obstacle and the target vehicle based on the predicted trajectory and the planned path of the target vehicle includes: for each predicted point on the predicted trajectory, finding the path point with the shortest distance to the predicted point on the planned path of the target vehicle as the nearest path point corresponding to the predicted point; if the predicted point and its corresponding nearest path point meet the preset risk judgment conditions, then the nearest path point is determined as the collision risk point.

[0089] The risk determination conditions include at least the following: the distance between the predicted point and its corresponding nearest path point is less than a first distance threshold; when an obstacle is at the predicted point and the target vehicle is at its corresponding nearest path point, the minimum distance between the outline of the obstacle and the outline of the target vehicle is less than a second distance threshold; the minimum distance between the nearest path point and the target vehicle is less than a third distance threshold; and the time difference between the estimated time for the obstacle to reach the predicted point and the estimated time for the target vehicle to reach its corresponding nearest path point falls within a preset range.

[0090] By transforming the potential spatiotemporal conflict between the predicted obstacle trajectory and the planned path of the target vehicle into a small number of critical path points (i.e. collision risk points) that can be directly used to calculate the second target vehicle speed, we can avoid missing real collision risks in scenarios such as unstructured areas, intersection merging, or lateral crossings by relying solely on the relationship between obstacles and lanes.

[0091] This mechanism uses each predicted point on the predicted trajectory as a starting point. It finds the shortest path point on the target vehicle's planned path that is closest to the predicted point, ensuring that each predicted point finds a candidate location on the target vehicle's planned path where it is most likely to intersect or make close contact. Finding the shortest path point can be implemented in a Cartesian coordinate system. For example, the Euclidean distance between each path point in the planned path and the predicted point can be calculated, and the path point corresponding to the minimum value can be selected as the shortest path point. To reduce computational load, the planned path can be trimmed based on the predicted point's timestamp or arc length range, and then the shortest path point can be found in the trimmed local path point set. Alternatively, a spatial index structure can be used to accelerate the nearest neighbor search to meet real-time requirements.

[0092] After obtaining the nearest path point, it is not directly used as a collision risk point. Instead, it is further screened through preset risk judgment conditions. Only when the predicted point and its corresponding nearest path point simultaneously satisfy spatial proximity, contour proximity, path validity and temporal consistency is the nearest path point determined as a collision risk point.

[0093] The first distance threshold is used to constrain the spatial proximity between the predicted point and the nearest path point. It reflects the geometric deviation between the predicted position of the obstacle and the planned path of the target vehicle. When the distance is less than the first distance threshold, it means that the future position of the obstacle is close enough to the planned path of the target vehicle and has the geometric basis for intersection. The selection of the first distance threshold can be related to the road width, the speed of the target vehicle, and the sampling of the planned path. It can be taken as a larger value in high-speed scenarios to cover a larger braking and deviation tolerance range, and a smaller value in low-speed fine maneuvering scenarios to reduce false alarms.

[0094] The second distance threshold is used to constrain the minimum distance between the obstacle profile and the target vehicle profile. It emphasizes that risk assessment not only focuses on point-to-point geometric proximity, but also on whether there is close contact or potential overlap under the condition of actual vehicle size. The profile here can be represented by a circumscribed rectangular profile or a polygonal profile. The minimum distance can be calculated using the shortest boundary distance between the profiles, or it can be equivalently replaced by judging whether there is overlap after profile expansion.

[0095] The third distance threshold is used to constrain the distance between the nearest path point and the target vehicle. Its function is to limit the collision risk point to fall within the target vehicle's future effective driving range, avoiding misjudging path points that are too far from the target vehicle on the planned path as risk points requiring immediate speed limits. The distance between the nearest path point and the target vehicle can be calculated along the arc length of the planned path, or it can be calculated using the Euclidean distance from the target vehicle's current position to the nearest path point, or the projected distance along the driving direction. It can also be dynamically adjusted with vehicle speed to balance advance capability and efficiency.

[0096] Please refer to Figure 4 As shown, the condition that the expected arrival time difference falls within a preset range is used to introduce temporal consistency, ensuring that risk points are not only spatially close but also potentially conflicting in time. When the target vehicle (e.g. Figure 4 (The vehicle) relative to an obstacle (such as...) Figure 4 (The car in the middle of the risk) arrived late seconds or earlier If so, the risk point is identified as a collision risk point, and a targeted target vehicle speed calculation is required in the subsequent process to determine if a collision risk exists.

[0097] Among these, the corresponding time thresholds before and after the design can be selected based on the expected system security sensitivity. , These represent the time it takes for the target vehicle and the obstacle to reach their respective risk points, respectively. The estimated time for the obstacle to reach its predicted point is also mentioned. The estimated time for the target vehicle to reach the nearest path point can be obtained from the timestamps in the predicted trajectory. It can be obtained from speed planning or time parameters attached to waypoints, or it can be calculated based on the path length under the assumption that the target vehicle is traveling at the current speed or the current target speed.

[0098] By converging the complex spatiotemporal intersection problem into a candidate risk set at the pathpoint level, the computability and real-time performance of risk identification can be improved. By jointly determining the first distance threshold, the second distance threshold, the third distance threshold, and the time difference window, and simultaneously constraining spatial proximity, geometric size influence, risk urgency, and temporal consistency, false alarms caused by pure spatial proximity can be effectively reduced, and false alarms caused by only considering temporal proximity while ignoring spatial accessibility can be avoided, thereby improving the accuracy of collision risk point identification.

[0099] S104: For each obstacle, calculate the second target vehicle speed based on the collision risk point, and update the global target vehicle speed based on the second target vehicle speed.

[0100] Step S104 is used to further quantify the spatiotemporal conflict risk reflected by the collision risk point of each obstacle into a second target vehicle speed that can be directly used for speed planning, based on the collision risk point already determined in step S103. The second target vehicle speed is then integrated into the global target vehicle speed of the target vehicle, thereby providing more targeted speed limit constraints for scenarios with high confidence in predicted trajectory and effective conflict windows, in addition to lane relationship risk assessment.

[0101] Since the collision risk point has simultaneously satisfied spatial proximity, contour proximity, path validity and temporal consistency, the core task of step S104 is no longer to determine whether there is a risk, but to adjust the speed to create sufficient time margin between the target vehicle and the obstacle at the collision risk point, so that the target vehicle avoids arriving at the obstacle at the same time or having contour overlap with it near the collision risk point.

[0102] In one feasible implementation, the calculation of the second target vehicle speed can primarily use the path length from the target vehicle to the collision risk point, the target vehicle's current speed, and the estimated arrival time of the obstacle at the predicted point corresponding to the collision risk point as inputs, while introducing a safety time margin as the conflict avoidance objective. The path length from the target vehicle to the collision risk point can be obtained by accumulating the arc length along the planned path of the target vehicle, or it can be approximated locally as a combination of Euclidean distance and curvature correction to improve real-time calculation efficiency.

[0103] The estimated arrival time of an obstacle can be given by the timestamp of the predicted trajectory or calculated from the velocity profile output by the prediction module. The safety time margin is used to constrain the arrival time difference between the target vehicle and the obstacle at the collision risk point, so that the two form a staggered passage relationship near the risk point. The value of the safety time margin can be set according to the system safety strategy and can be adjusted differently according to the obstacle type.

[0104] In one exemplary embodiment, calculating a second target vehicle speed based on the collision risk point includes: obtaining the path length from the target vehicle to the collision risk point, the current speed of the target vehicle, the estimated arrival time from the obstacle to the predicted point corresponding to the collision risk point, and a safety time margin; assuming the target vehicle travels along the planned path with uniform deceleration, determining the expected arrival time of the target vehicle to the collision risk point, such that the difference between the expected arrival time and the estimated arrival time of the obstacle is greater than the safety time margin; calculating an expected deceleration based on the path length, the current speed of the target vehicle, and the expected arrival time, and calculating the second target vehicle speed based on the expected deceleration.

[0105] The implementation logic for calculating the second target vehicle speed based on collision risk points is to take the sufficient staggered time at the collision risk points as the speed limit target, and then use this staggered time constraint to inversely deduce the upper speed limit that the target vehicle needs to meet.

[0106] In one feasible implementation, the path length from the target vehicle to the collision risk point, the target vehicle's current speed, the estimated arrival time from the obstacle to the predicted point corresponding to the collision risk point, and the safety time margin are first obtained. The path length can be obtained by accumulating arc lengths along the planned path of the target vehicle. That is, starting from the path point corresponding to the current position of the target vehicle, the segment lengths between adjacent path points are successively added along the planned path until the path point corresponding to the collision risk point is reached, so as to obtain the travel distance from the target vehicle to the collision risk point. To improve real-time performance, the arc length parameter difference can be directly used as the path length when the sampling density of the planned path is high, or the path length can be approximated by Euclidean distance and a curvature correction term can be superimposed when the local curvature is small.

[0107] The target vehicle's current speed can be obtained from chassis feedback or the state estimation module, and can be further converted into a velocity component along the tangential direction of the planned path to ensure consistency with the definition of path length. The estimated arrival time of obstacles is usually provided by the predicted trajectory. The predicted points in the predicted trajectory contain timestamps or relative time series. The predicted point corresponding to the collision risk point can be determined based on the correspondence between the predicted point and the nearest path point in step S103, thereby obtaining the estimated arrival time of that predicted point. The safety time margin is used to define the minimum time offset that the system wants to achieve at the risk point. The selection of the safety time margin can be related to the obstacle type, road environment, and system safety strategy. For example, a larger safety time margin can be set for pedestrians, and a relatively smaller safety time margin can be set for vehicles traveling in the same direction, to balance safety and traffic efficiency.

[0108] Subsequently, assuming the target vehicle travels along the planned path with uniform deceleration, the expected arrival time of the target vehicle at the collision risk point is determined, ensuring that the difference between the expected arrival time and the estimated arrival time of the obstacle is greater than the safety time margin. The assumption of uniform deceleration simplifies the complex speed planning problem into an analytical kinematic inverse problem, making the second target vehicle speed easier to calculate.

[0109] The expected arrival time can be set conservatively, for example, by setting the expected arrival time to the obstacle's estimated arrival time plus a safety time margin and then adding a system response delay margin, thus ensuring that the target vehicle arrives significantly later than the obstacle at the collision risk point; or a window selection method can be used, selecting the smallest expected arrival time from all feasible arrival times that meet the safety time margin, in order to minimize unnecessary excessive deceleration.

[0110] After determining the expected arrival time, the expected deceleration is calculated based on the path length, the target vehicle's current speed, and the expected arrival time. The second target speed is then calculated based on this expected deceleration. The expected deceleration can be calculated using uniform deceleration kinematics, for example, by inversely calculating the deceleration given a travel time and distance, enabling the target vehicle to complete the planned path length within the expected arrival time.

[0111] To ensure the feasibility of the results, the expected deceleration can be checked for consistency with vehicle dynamics constraints and comfort constraints. For example, the expected deceleration should not exceed the maximum comfort deceleration threshold and should not exceed the upper limit of the physical deceleration allowed by the braking system. When the expected deceleration obtained by reverse calculation exceeds the comfort threshold, it can be corrected by increasing the expected arrival time or adopting a segmented deceleration strategy to make the deceleration process corresponding to the second target speed smoother. The second target speed can be calculated from the relationship between the target vehicle's current speed, expected deceleration, and expected arrival time. It is used to express the upper limit of speed or expected speed level that the target vehicle should follow under the current risk constraints, and is subsequently integrated with other constraints through a global target speed update mechanism.

[0112] Specifically, assuming we want the time difference between the target vehicle and the obstacle reaching the collision risk point after the speed limit is reached. Greater than the safety time margin The following speed limit design can be implemented:

[0113] Assume the initial speed of the target vehicle is The speed at which the collision risk point is reached along the planned path is The estimated arrival time is Now, we hope the expected arrival time after the speed limit is... It needs to meet the following requirements: .

[0114] Due to initial velocity Since the velocity remains constant, we can assume that the target vehicle is undergoing uniform deceleration. The desired second target velocity can be calculated using the following formula. :

[0115]

[0116]

[0117] in, The path length from the target vehicle to the collision risk point. The desired deceleration.

[0118] In cases where there are multiple collision risk points, the corresponding second target vehicle speed needs to be calculated separately for each collision risk point. And select the minimum value of the second target vehicle speed from them. And in step S104, update the global target vehicle speed up to the current step:

[0119]

[0120] in, The global target vehicle speed in step S104, The global target vehicle speed in step S102.

[0121] The aforementioned second target speed calculation mechanism realizes the mapping from spatial risk identification to temporal safety constraints, enabling speed planning to quantitatively limit speed with time margin as the target, and improving the coverage of interactive risks.

[0122] S105: After traversing all obstacles, use the minimum global target vehicle speed among all the global target vehicle speeds updated during the traversal as the reference speed of the target vehicle.

[0123] Step S105 is used to complete the convergence and output in the obstacle traversal dimension. Its core is to unify the first target vehicle speed and the second target vehicle speed calculated for each obstacle into a global target vehicle speed that can represent the most stringent safety constraints at the current moment. After traversing all obstacles, the global target vehicle speed is output as the reference vehicle speed of the target vehicle to the subsequent speed planning module.

[0124] Since multiple obstacles may exist around the target vehicle at the same time, and the sources and urgency of risks corresponding to different obstacles are different, if speed planning is based solely on the speed limit result of a single obstacle, it is easy to overlook other obstacles with higher risks. If the speed limit results of multiple obstacles are simply averaged or weighted, it may also mask the most stringent constraint and lead to insufficient safety margin. Therefore, step S105 adopts a fusion strategy of updating the global target vehicle speed one obstacle at a time and finally taking the minimum value. This ensures that the reference vehicle speed is always subject to the common constraints of all identified risk constraints, and is dominated by the most stringent constraint, thereby guaranteeing overall safety in scenarios with multiple obstacles.

[0125] In one feasible implementation, the global target vehicle speed can be initialized before obstacle traversal begins as the smaller of the maximum speed limit allowed by the driving lane or the current speed of the target vehicle, in order to provide an upper bound for subsequent convergence updates. Subsequently, when traversing each obstacle, a first target vehicle speed is calculated based on the obstacle type and lane relationship, and a second target vehicle speed is further calculated if the predicted trajectory confidence meets preset requirements and a collision risk point is identified.

[0126] The first target speed and the second target speed can represent the speed limit under two different coordinate systems and different risk modeling methods. The first target speed emphasizes the preventive speed limit related to lane intrusion and lane occupation, while the second target speed emphasizes the temporal margin speed limit for collision risk points. Since both are used to form the speed limit constraint, in step S105, after each obstacle is processed, the first target speed and the second target speed corresponding to that obstacle are compared with the current global target speed, and the smaller value is selected as the new global target speed.

[0127] This update process can be understood as a monotonically non-increasing convergence process of the global target vehicle speed. That is, as obstacle traversal progresses, the global target vehicle speed can only remain constant or decrease, thus avoiding the unexpected relaxation of the global speed constraint due to traversal order or local risk assessment errors. The updated global target vehicle speed is then used for speed planning of subsequent obstacles, allowing subsequent calculations to be performed under a more conservative speed benchmark.

[0128] Specifically, after all obstacles have been traversed, the target vehicle speed calculated based on each obstacle is used to select the minimum speed value as the reference speed for the target vehicle. ,Right now

[0129] .

[0130] By employing a minimum fusion strategy, the global target vehicle speed can cover the speed limit constraints imposed by all obstacles, ensuring overall safety in speed planning even in multi-objective conflict scenarios. Through obstacle-by-obstacle updates and monotonically converged, the global target vehicle speed is insensitive to the obstacle traversal order, resulting in a more stable and predictable output that facilitates control module execution. Simultaneously, the global target vehicle speed serves as a reference speed, providing a unified constraint for subsequent speed planning. This helps reduce the mutual interference between the outputs of different risk modules, lowering the probability of speed fluctuations and sudden braking, thereby enhancing the driving safety and comfort of autonomous vehicles in complex dynamic environments.

[0131] Please refer to Figure 5 As shown, based on the same inventive concept as the aforementioned vehicle driving speed planning method, this disclosure provides a vehicle driving speed planning device 500, which includes: an obstacle determination module 501, a first calculation module 502, a collision risk determination module 503, a second calculation module 504, and an output module 505.

[0132] The obstacle determination module 501 is used to traverse all obstacles around the target vehicle. For each obstacle, the type of obstacle is determined based on the relative position of the obstacle and the target vehicle's driving lane. The first calculation module 502 is used to calculate a first target speed based on the type of obstacle and update the global target speed of the target vehicle based on the first target speed. The collision risk determination module 503 is used to obtain the predicted trajectory of the obstacle. If the confidence level of the predicted trajectory meets a preset requirement, the collision risk point between the obstacle and the target vehicle is determined based on the predicted trajectory and the planned path of the target vehicle. The second calculation module 504 is used to calculate a second target speed based on the collision risk point and update the global target speed of the target vehicle based on the second target speed. The output module 505 is used to output the minimum global target speed among all the global target speeds updated during the traversal as the reference speed of the target vehicle after traversing all obstacles.

[0133] Please refer to Figure 6 As shown, this disclosure also provides an autonomous driving vehicle 600, which includes at least one processor 601, a memory 602 (e.g., non-volatile memory), a main memory 603, and a communication interface 604, and the at least one processor 601, memory 602, main memory 603, and communication interface 604 are connected together via an internal bus 605. The at least one processor 601 is used to invoke at least one program instruction stored or encoded in the memory 602 to cause the at least one processor 601 to perform various operations and functions of the vehicle driving speed planning method described in the various embodiments of this specification.

[0134] In the embodiments of this specification, the autonomous vehicle 600 may be configured with a functional terminal to carry the above-mentioned hardware structure. The terminal includes, but is not limited to: personal computer, server computer, workstation, desktop computer, laptop computer, notebook computer, mobile electronic device, smartphone, tablet computer, cellular phone, personal digital assistant (PDA), handheld device, messaging device, wearable electronic device, consumer electronic device, etc.

[0135] This disclosure also provides a machine-readable medium carrying executable instructions, which, when executed by a processor, can be used to implement various operations and functions of the vehicle driving speed planning method described in the various embodiments of this specification.

[0136] The machine-readable medium in this disclosure can be a machine-readable signal medium or a machine-readable storage medium, or any combination thereof. A machine-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a machine-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a machine-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0137] In this disclosure, the machine-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying machine-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The machine-readable signal medium may also be any machine-readable medium other than a machine-readable storage medium, capable of transmitting, propagating, or transmitting a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the machine-readable medium may be transmitted using any suitable medium, including but not limited to: wireless, wireline, optical fiber, RF, etc., or any suitable combination thereof.

[0138] This disclosure also provides a computer program product, which includes computer instructions that, when executed by a processor, can implement the above-described vehicle driving speed planning method.

[0139] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, systems, or program products. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a program product embodied on one or more storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing program code.

[0140] It will be apparent to those skilled in the art that this disclosure is not limited to the details of the exemplary embodiments described above, and that this disclosure can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of this disclosure is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within this disclosure. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0141] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A vehicle driving speed planning method characterized by comprising: The method comprises the following steps: traversing each obstacle around the target vehicle, and for each obstacle: determining the type of the obstacle based on the relative position relationship between the obstacle and the driving lane of the target vehicle; calculating a first target vehicle speed based on the type of the obstacle, and updating the global target vehicle speed of the target vehicle based on the first target vehicle speed; obtaining a predicted trajectory of the obstacle, and if the confidence degree of the predicted trajectory meets a preset requirement, determining a collision risk point between the obstacle and the target vehicle based on the predicted trajectory and the planned path of the target vehicle; calculating a second target vehicle speed based on the collision risk point, and updating the global target vehicle speed of the target vehicle based on the second target vehicle speed; after traversing all obstacles, taking the minimum global target vehicle speed obtained in the traversal process as the reference speed of the target vehicle.

2. The vehicle driving speed planning method according to claim 1, characterized by, determining the type of the obstacle based on the relative position relationship between the obstacle and the driving lane of the target vehicle comprises: based on the relationship between the contour of the obstacle and the lane boundary of the driving lane of the target vehicle, determining the obstacle whose contour is completely outside the lane boundary as a low-risk obstacle, determining the obstacle whose contour is partially inside the lane boundary as an intrusion obstacle, and determining the obstacle whose contour is completely inside the lane boundary as an occupying obstacle; wherein the low-risk obstacle does not participate in the calculation of the first target vehicle speed, the second target vehicle speed and the update of the global target vehicle speed.

3. The vehicle driving speed planning method according to claim 2, characterized by, When the obstacle is an intrusion obstacle, the calculation method of the first target vehicle speed comprises: establishing a Frenet coordinate system based on the center reference line of the driving lane of the target vehicle; in the lateral direction, calculating the intrusion distance of the intrusion obstacle relative to the lane boundary; in the longitudinal direction, dividing a preset longitudinal range in front of the target vehicle into a near distance interval and a far distance interval, and determining the interval in which the intrusion obstacle is located based on the longitudinal distance between the target vehicle and the intrusion obstacle; when the intrusion obstacle is in the near distance interval, calculating the first target vehicle speed through a first intrusion coefficient based on the intrusion distance, the current speed of the target vehicle and the maximum speed limit of the driving lane; when the intrusion obstacle is in the far distance interval, calculating the first target vehicle speed through a second intrusion coefficient based on the intrusion distance, the current speed of the target vehicle and the maximum speed limit of the driving lane, and the speed limit degree of the second intrusion coefficient is smaller than that of the first intrusion coefficient.

4. The vehicle driving speed planning method according to claim 2, characterized by, When the obstacle is an occupying obstacle, the calculation method of the first target vehicle speed comprises: calculating the relative speed of the target vehicle and the occupying obstacle in the longitudinal direction of the driving lane; when the relative speed is greater than a preset relative speed threshold, taking the current speed of the target vehicle as the first target vehicle speed; when the relative speed is less than or equal to the preset relative speed threshold, calculating the safety distance of the target vehicle based on the preset maximum comfortable deceleration and the compensation distance of the target vehicle, and determining the first target vehicle speed based on the longitudinal distance between the target vehicle and the occupying obstacle and the safety distance.

5. The vehicle driving speed planning method according to claim 4, characterized by, determining the first target vehicle speed based on the longitudinal distance between the target vehicle and the occupying obstacle and the safety distance comprises: when the longitudinal distance between the target vehicle and the occupying obstacle is greater than the safety distance, taking the current speed of the target vehicle as the first target speed of the occupying obstacle; when the longitudinal distance between the target vehicle and the occupying obstacle is less than or equal to the safety distance, taking the component speed of the occupying obstacle in the longitudinal direction as the first target speed.

6. The vehicle driving speed planning method according to claim 1, characterized by, based on the predicted trajectory and the planned path of the target vehicle, determining a collision risk point of the obstacle and the target vehicle, comprising: for each predicted point on the predicted trajectory, finding a path point on the planned path of the target vehicle closest to the predicted point as a nearest path point corresponding to the predicted point; if the predicted point and the nearest path point corresponding thereto satisfy a preset risk determination condition, determining the nearest path point as the collision risk point; wherein the risk determination condition at least comprises: the distance between the predicted point and the nearest path point corresponding thereto is less than a first distance threshold; in the case that the obstacle is at the predicted point and the target vehicle is at the nearest path point corresponding thereto, the minimum distance between the obstacle contour and the target vehicle contour is less than a second distance threshold; the minimum distance between the nearest path point and the target vehicle is less than a third distance threshold; the time difference between the predicted time for the obstacle to reach the predicted point and the predicted time for the target vehicle to reach the nearest path point corresponding thereto falls within a preset range.

7. The vehicle driving speed planning method according to claim 1, characterized by, based on the collision risk point, calculating a second target speed, comprising: obtaining the path length of the target vehicle to the collision risk point, the current speed of the target vehicle, the predicted arrival time of the obstacle to the predicted point corresponding to the collision risk point, and a safety time margin; assuming that the target vehicle travels along the planned path at a uniform deceleration, determining the expected arrival time of the target vehicle to the collision risk point, so that the difference between the expected arrival time and the predicted arrival time of the obstacle is greater than the safety time margin; based on the path length, the current speed of the target vehicle, and the expected arrival time, calculating the second target speed.

8. A vehicle driving speed planning device characterized by comprising: comprising: an obstacle determination module, configured to traverse each obstacle around the target vehicle, and for each obstacle, based on the relative positional relationship between the obstacle and the driving lane of the target vehicle, determine the type of the obstacle; a first calculation module, configured to calculate the first target speed based on the type of the obstacle, and update the global target speed of the target vehicle based on the first target speed; a collision risk determination module, configured to obtain the predicted trajectory of the obstacle, and if the confidence of the predicted trajectory satisfies a preset requirement, based on the predicted trajectory and the planned path of the target vehicle, determine the collision risk point of the obstacle and the target vehicle; a second calculation module, configured to calculate the second target speed based on the collision risk point, and update the global target speed of the target vehicle based on the second target speed; an output module, configured to output the minimum global target speed among all the global target speeds updated during the traversal process as the reference speed of the target vehicle after traversing all the obstacles.

9. An autonomous vehicle comprising a memory, a processor, and a program stored on the memory and executable on the processor, wherein the program comprises instructions for: The processor implements the vehicle driving speed planning method according to any one of claims 1-7 when executing the program. The processor implements the vehicle driving speed planning method according to any one of claims 1-7 when executing the program.

10. A machine readable medium / computer program product, characterized in that, The machine readable medium carries executable instructions, which when executed by a processor, implement the vehicle driving speed planning method according to any one of claims 1-7; and / or The computer program product comprises computer instructions, which when executed by a processor, implement the vehicle driving speed planning method according to any one of claims 1-7.

Images