Lane changing decision method and device, electronic equipment and storage medium

Abstract

The present disclosure relates to the technical field of unmanned driving, and discloses a lane changing decision method and device, electronic equipment and a storage medium. The method comprises: for each candidate lane, determining first key obstacle information affecting lane changing of a self vehicle to a candidate lane based on a predicted driving track; and determining second key obstacle information affecting lane changing of the self vehicle to the candidate lane based on a predicted driving track of a reference obstacle of a preset intention; retaining obstacle information closest in interaction distance based on a principle of closest interaction distance, and the obstacle information closest in interaction distance of different candidate lanes forming a first feature obstacle information set, determining a second feature obstacle information set based on a passable distance of the self vehicle in a current lane and a lane changing distance required for lane changing to a candidate lane; and determining a target lane for lane changing according to the feature obstacle information set in combination with a utility function. The present disclosure improves the rationality and flexibility of lane changing decision.

Description

Technical Field

[0001] This disclosure relates to the field of autonomous driving technology, and in particular to a lane-changing decision-making method, apparatus, electronic device, and storage medium. Background Technology

[0002] In recent years, with the development of science and technology and the rise of artificial intelligence, autonomous driving technology has become a hot research topic. Among them, lane-changing decision-making addresses the problem of how a vehicle can choose the most appropriate time and method to change lanes based on traffic conditions and the target route during driving, in order to ensure high-efficiency passage.

[0003] Currently, mainstream lane-changing decision-making methods include rule-based and experience-based methods, and machine learning-based methods. These methods aim to develop accurate and reliable algorithms that enable autonomous driving systems to make correct and safe lane-changing decisions in complex traffic environments. The most mature and widely used method is the rule-based and experience-based method. This approach uses predefined rules and experiential knowledge to formulate lane-changing decisions. It utilizes expert experience and prior knowledge such as road traffic rules, encoding them into a series of rules or logic to guide the autonomous driving system in lane-changing. These rules can be based on perception data such as distance, speed, and traffic signals, as well as prior knowledge such as traffic rules and traffic flow.

[0004] However, rule-based and experience-based lane-changing decision-making methods for autonomous driving have the advantages of simplicity, strong interpretability, and the ability to adjust to specific road conditions and rules. However, in the real world, traffic environments are often complex, changeable, and uncertain. Rule-based and experience-based methods struggle to handle complex and dynamic traffic environments, lacking flexibility and adaptability. Their effectiveness is limited when considering interactions with surrounding traffic participants in complex traffic scenarios. Summary of the Invention

[0005] To address or at least partially address the aforementioned technical problems, embodiments of this disclosure provide a lane-changing decision-making method, apparatus, electronic device, and storage medium, thereby improving the rationality and flexibility of lane-changing decisions.

[0006] In a first aspect, embodiments of this disclosure provide a lane-changing decision-making method, the method comprising:

[0007] For each candidate lane, information on a first key obstacle affecting a vehicle's lane change to a candidate lane is determined based on the predicted driving trajectory, and the information on the first key obstacle corresponding to different candidate lanes forms a first set of candidate obstacle information; and information on a second key obstacle affecting a vehicle's lane change to a candidate lane is determined based on the predicted driving trajectory of a reference obstacle with a preset intent, and the information on the second key obstacle corresponding to different candidate lanes forms a second set of candidate obstacle information.

[0008] The first set of candidate obstacle information is compared with the second set of candidate obstacle information. For a candidate lane, the obstacle information with the closest interaction distance is retained based on the principle of closest interaction distance. The obstacle information with the closest interaction distance of different candidate lanes forms the first feature obstacle information set.

[0009] The second set of feature obstacle information is determined from the first set of feature obstacle information based on the vehicle's passable distance in the current lane and the lane-changing distance required to change lanes to a candidate lane;

[0010] The target lane for lane changing is determined based on the first set of feature obstacle information and the second set of feature obstacle information, combined with a utility function.

[0011] Secondly, embodiments of this disclosure also provide a lane-changing decision-making device, the device comprising:

[0012] The first determining module is used to determine, for each candidate lane, first key obstacle information affecting the vehicle's lane change to a candidate lane based on the predicted driving trajectory, and the first key obstacle information corresponding to different candidate lanes forms a first candidate obstacle information set; and to determine second key obstacle information affecting the vehicle's lane change to a candidate lane based on the predicted driving trajectory of a reference obstacle with a preset intention, and the second key obstacle information corresponding to different candidate lanes forms a second candidate obstacle information set.

[0013] The second determining module is used to compare the first set of candidate obstacle information with the second set of candidate obstacle information. For a candidate lane, the obstacle information with the closest interaction distance is retained based on the principle of closest interaction distance. The obstacle information with the closest interaction distance of different candidate lanes forms the first feature obstacle information set.

[0014] The third determining module is used to determine the second set of feature obstacle information from the first set of feature obstacle information based on the passable distance of the vehicle in the current lane and the lane-changing distance required to change lanes to a candidate lane;

[0015] The fourth determining module is used to determine the target lane for lane changing based on the first set of feature obstacle information and the second set of feature obstacle information combined with a utility function.

[0016] Thirdly, embodiments of this disclosure also provide an electronic device, the electronic device comprising: one or more processors; a storage device for storing one or more programs; and when the one or more programs are executed by the one or more processors, causing the one or more processors to implement the lane-changing decision method as described above.

[0017] Fourthly, embodiments of this disclosure also provide a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the lane-changing decision method as described above.

[0018] The lane-changing decision-making method provided in this disclosure optimizes the extraction method of obstacles in candidate lanes by considering the interaction information of the predicted driving trajectory of the vehicle and other traffic participants. Based on the utility function method, it comprehensively considers factors such as traffic efficiency, traffic safety and traffic capacity to evaluate lane characteristics, and finally decides on the target lane for lane changing, thereby improving the rationality, flexibility and safety of lane-changing decision-making. Attached Figure Description

[0019] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.

[0020] Figure 1 This is a flowchart of a lane-changing decision method according to an embodiment of the present disclosure;

[0021] Figure 2 This is a schematic diagram of an obstacle with different driving intentions in an embodiment of this disclosure;

[0022] Figure 3 This is a schematic diagram of the technical route of a lane-changing decision-making method in an embodiment of this disclosure;

[0023] Figure 4 This is a schematic diagram of a process for extracting lane feature obstacles based on predicted trajectory interaction in an embodiment of this disclosure;

[0024] Figure 5 This is a schematic diagram of the structure of a lane-changing decision device according to an embodiment of the present disclosure.

[0025] Figure 6 This is a schematic diagram of the structure of an electronic device according to an embodiment of this disclosure. Detailed Implementation

[0026] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0027] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0028] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0029] Figure 1 This is a flowchart illustrating a lane-changing decision-making method according to an embodiment of this disclosure. The method can be executed by a lane-changing decision-making device, which can be implemented in software and / or hardware and can be configured in an electronic device. Figure 1 As shown, the method may specifically include the following steps:

[0030] S110. For each candidate lane, determine the first key obstacle information that affects the vehicle's lane change to a candidate lane based on the predicted driving trajectory, and the first key obstacle information corresponding to different candidate lanes forms a first candidate obstacle information set; and determine the second key obstacle information that affects the vehicle's lane change to a candidate lane based on the predicted driving trajectory of a reference obstacle with a preset intention, and the second key obstacle information corresponding to different candidate lanes forms a second candidate obstacle information set.

[0031] Among them, the candidate lane refers to the target lane that the vehicle can change to. For example, if a road has four lanes in the same direction, the current lane of the vehicle is one of them, and the other three lanes are all candidate lanes.

[0032] The predicted trajectory of a vehicle refers to its trajectory within a future time window (e.g., 6 seconds from the current moment). The predicted trajectory of an obstacle refers to its trajectory within a future time window.

[0033] In general, the predicted driving trajectory is obtained through a specific algorithm based on information such as the current speed, acceleration, and position of obstacles or the vehicle itself. Specifically, autonomous vehicles use sensors (such as cameras, LiDAR, and radar) to perceive the surrounding environment and track the movement of other road users (i.e., obstacles). These sensors provide information such as the position, speed, and direction of obstacles around the vehicle. The perception module fuses data from different sensors to obtain more accurate and complete position and motion information of the obstacles. Subsequently, the prediction module generates predicted driving trajectories for these obstacles over a future period based on the position and motion information of the obstacles, combined with map information and constraints such as traffic regulations. In general, these predicted driving trajectories can be calculated based on historical data, motion models, and environmental conditions, and this application embodiment does not limit this.

[0034] For example, determining the first key obstacle information affecting a vehicle's lane change to a candidate lane based on the predicted driving trajectory includes:

[0035] Within a preset time window, based on the predicted trajectory of an obstacle within the vehicle's perception range and the vehicle's expected lane-changing trajectory when changing lanes to a candidate lane, it is determined whether there is a collision risk between the obstacle and the vehicle. If so, the obstacle is identified as a first critical obstacle, and its information is recorded. This first critical obstacle information includes at least the collision distance, collision time, and collision position of the vehicle when the obstacle collides with it. For multiple first critical obstacles, the first critical obstacles whose collision positions are at the rear of the vehicle are eliminated, resulting in one or more remaining first critical obstacles. The first critical obstacle with the smallest corresponding collision distance among the remaining one or more first critical obstacles is identified as the final first critical obstacle. In summary, the autonomous vehicle (i.e., the vehicle itself) interacts with the predicted trajectories of other traffic participants on each candidate lane to ultimately obtain a first critical obstacle that needs to be considered in lane-changing decisions, and extracts it as lane feature information for that candidate lane.

[0036] In some specific implementations, for each candidate lane, a time window is used to observe the predicted trajectory of the autonomous vehicle and the predicted trajectories of other traffic participants. The time window can be a fixed-length period, such as the next 6 seconds. Within this time window, the autonomous vehicle analyzes the changes in the predicted trajectories of itself and other traffic participants in the candidate lane to determine if there is a potential collision risk. For example, starting from the current moment, at the same sampling time interval, such as 0.2 seconds, the expected trajectory of the autonomous vehicle (i.e., the predicted trajectory) and the predicted trajectory of the obstacle are sampled separately, resulting in two discrete sampling point trajectories. Each sampling point records information such as its position, speed, and heading angle in the vehicle's local coordinate system at the corresponding sampling time. At each sampling time t, based on the length, width, and heading angle information of the autonomous vehicle and the obstacle recorded on the two sampling point trajectories, two polygonal boxes representing the areas occupied by each at sampling time t can be generated in the coordinate system. Using this method, at all sampling times within the time window, the GJK collision detection algorithm is called to determine whether two polygonal boxes intersect. This indicates whether a collision has occurred between the vehicle and the obstacle at each sampling time t. If there is an intersection, it means the vehicle and the obstacle have collided; otherwise, it means they have not collided. If a collision is determined to have occurred, the location of the collision (front, body, or rear of the vehicle), the collision distance d_collision, and the collision time t_collision are recorded.

[0037] After the interactive processing of the predicted driving trajectory of the vehicle and the predicted driving trajectories of surrounding traffic participants, obstacles with potential collision risks to the vehicle and related collision information in each candidate lane can be obtained. Based on the different collision positions and distances with the vehicle, the main obstacles affecting lane characteristics in each candidate lane are further distinguished. For obstacles whose collision position is at the rear of the vehicle, it means that the vehicle was in a forward position at the time of the collision. According to the description of liability determination in this scenario in relevant traffic regulations, the primary responsible party is the vehicle that rear-ended the other vehicle while occupying the right-of-way. Therefore, when making lane-changing decisions, only obstacles whose collision positions are at the front and sides of the vehicle are considered, and the obstacle corresponding to the minimum collision distance d_collision in each candidate lane is recorded. That is, the obstacle with the closest collision distance is used as the set of candidate obstacles affecting lane characteristics (i.e., the first set of candidate obstacle information), which can be denoted as obs_list1 = {On|n = 1, 2, ..., N}, where N is the number of candidate lanes.

[0038] After the interactive processing of the predicted driving trajectory of the vehicle and the predicted driving trajectories of surrounding traffic participants, obstacles with potential collision risks to the vehicle in each candidate lane can be obtained, and the one with the closest collision distance is selected as the first candidate obstacle for the corresponding candidate lane (i.e., the final first critical obstacle). However, other obstacles without potential collision risks do not necessarily mean that they have no impact on the vehicle's lane-changing movement. For example, when there are other obstacle vehicles merging into the current lane (the target lane to which the vehicle is changing lanes) in front of the vehicle, even if they do not collide with the vehicle within the time window, i.e., there is no potential collision risk, they still need to be considered as the second critical obstacle for the current lane because they affect the vehicle's travel range within the current lane. See attached... Figure 2 As shown, for the candidate lane `this_lane`, obstacles can be categorized into five intent types based on the location of the traffic participant: a. stationary obstacles; b. obstacles whose current position is not within the candidate lane `this_lane` and whose predicted intent is a CUT IN type obstacle; c. obstacles whose current position is within the candidate lane `this_lane` and whose predicted intent is a LANE KEEPING type obstacle; d. obstacles whose current position is within the candidate lane `this_lane` and whose predicted intent is a TO LANE type obstacle; e. obstacles whose current position is within the candidate lane `this_lane` and whose predicted intent is a LEAVE LANE type obstacle. Obstacles with predicted intent types a, c, and d have already been processed through collision detection in the previous steps. Obstacles that do not pose a collision risk with the vehicle's predicted trajectory can be ignored in terms of their impact on lane-changing decisions within the candidate lane. Therefore, this step primarily focuses on obstacles with predicted intent types b and e. The main focus is on determining whether the predicted driving trajectory under the corresponding intention of the obstacle affects the vehicle. The specific steps are as follows: For the vehicle, a heuristic speed search method is adopted to fix the expected driving trajectory in each candidate lane with a constant acceleration, for example, 0.5 m / s². 2The vehicle accelerates to a candidate lane (denoted as the target lane). Based on a preset speed curve, the time t_enter required for the vehicle to enter the target lane along the desired trajectory is calculated. Sampling time points on the predicted trajectory of obstacles after t_enter are taken. If t_enter is greater than the time window of the predicted trajectory, the last sampling time point on the predicted trajectory is selected. The relative position of the obstacle to the target lane at the selected sampling time point is calculated. If the obstacle's centroid is within the target lane's range, it is considered to still affect the vehicle's movement when entering the target lane. The relative distance d_enter between the sampling point of the predicted trajectory and the vehicle when the obstacle first enters the target lane is recorded. The obstacle with the smallest d_enter is recorded for each candidate lane, forming a second set of candidate obstacle information, denoted as obs_list2 = {On|n = 1, 2, ..., N}, where N is the number of candidate lanes.

[0039] In general, the reference obstacles with the preset intent include obstacles whose current position is not in the candidate lane and whose intent is to enter the candidate lane, and obstacles whose current position is in the candidate lane and whose intent is to exit the candidate lane.

[0040] Correspondingly, the second key obstacle information affecting the vehicle's lane change to a candidate lane is determined based on the predicted driving trajectory of the reference obstacle with a preset intent, including:

[0041] Based on the predicted driving trajectory of reference obstacles, the reference positions of each reference obstacle when interacting with the vehicle are determined. The reference obstacle with the smallest relative distance between its reference position and the vehicle is identified as the second key obstacle, and this relative distance is recorded as the interaction distance when interacting with the vehicle, thus obtaining the second key obstacle information. Here, "interaction between the obstacle and the vehicle" means that both the obstacle and the vehicle are traveling in the same lane or are about to travel in the same lane, and the vehicle's movement may affect the obstacle. The interaction distance refers to the relative distance between the obstacle and the vehicle at a given moment.

[0042] Furthermore, before determining the second key obstacle information affecting the vehicle's lane change to a candidate lane based on the predicted driving trajectory of the reference obstacle with a preset intent, the process also includes:

[0043] The reference time required for the vehicle to enter the candidate lane is predicted when the vehicle is traveling according to the desired lane-changing trajectory and the preset speed.

[0044] Based on the predicted driving trajectory of a reference obstacle, the reference position of the reference obstacle at the time point corresponding to the reference time is determined. If the reference position of the reference obstacle is within a candidate lane, the relative distance between the reference obstacle and the vehicle at the time point corresponding to the reference time is determined as the interaction distance when the reference obstacle interacts with the vehicle. Compared to the existing technology that directly calculates the time to collision (TTC) or passable distance on each candidate lane in real time using the current speed and distance information of the obstacle to select feature obstacles, this application embodiment optimizes the lane feature obstacle selection method through a collision detection method based on the time dimension. This fully considers the impact of obstacles on the vehicle's lane-changing behavior and has the advantage of improving lane-changing safety. In addition, the predicted intent type of the obstacle is considered, and obstacles that affect the vehicle's lane-changing decision within a certain time window are simultaneously considered and filtered. Therefore, the lane-changing decision method provided by this application embodiment can fully consider the potential impact of surrounding traffic participants on the vehicle's lane-changing decision and improve the safety of lane-changing behavior.

[0045] S120. The first set of candidate obstacle information is compared with the second set of candidate obstacle information. For a candidate lane, the obstacle information with the closest interaction distance is retained based on the principle of closest interaction distance. The obstacle information with the closest interaction distance of different candidate lanes forms the first feature obstacle information set.

[0046] Specifically, the obstacles in each candidate lane from the obstacle set obs_list2 (i.e., the second set of candidate obstacle information) and obs_list1 (i.e., the first set of candidate obstacle information) obtained in the above steps are compared. The characteristic obstacles for a candidate lane are retained based on the principle of the closest interaction distance to the vehicle. The obstacle with the closest interaction distance for each candidate lane (the closest interaction distance obstacle may come from obs_list1 or obs_list2) is retained, which is the first set of characteristic obstacle information.

[0047]

[0048] S130. Determine a second set of feature obstacle information from the first set of feature obstacle information based on the passable distance of the vehicle in the current lane and the lane-changing distance required to change lanes to a candidate lane.

[0049] To ensure decision stability, after obtaining the first set of obstacle information for candidate lanes, obstacles in the candidate lanes can be filtered based on their motion states. In obstacle filtering, existing technologies typically calculate the time to collision (TTC) or passable distance in real time based on the obstacle's current speed and distance information. However, when the predicted trajectory of the vehicle changes relative to other traffic participants, the change in collision detection results leads to changes in the selection of target obstacles, which can easily result in errors in lane information judgment and thus unreasonable lane-changing decisions. To address this problem, this application's embodiment optimizes the stability of target obstacle selection within candidate lanes by considering the passable distance of the vehicle's current lane and the lane-changing safety distance of the candidate lanes, eliminating the phenomenon of unreasonable decisions caused by jumps in collision detection results.

[0050] In one embodiment, determining the second set of feature obstacle information from the first set of feature obstacle information based on the vehicle's passable distance in its current lane and the lane-changing distance required to change lanes to a candidate lane includes:

[0051] The obstacles in the first set of feature obstacle information that pose a collision risk to the vehicle and are located in front of the vehicle in the candidate lane are identified as candidate obstacles.

[0052] Determine the permissible distance of the vehicle in its current lane and the lane-changing distance required to change lanes to the candidate lane;

[0053] If the passable distance is greater than the sum of the lane-changing distance and the safe following distance, then the candidate obstacle is identified as the second feature obstacle. The safe following distance is determined based on the following model and the motion information of the candidate obstacle. The second feature obstacles corresponding to different candidate lanes constitute the second feature obstacle information set.

[0054] Specifically, for the obstacle O of candidate lane n n O n If a vehicle is in front of another vehicle and there is a collision risk with it, it is recorded as risk_obj in the obs_list. n More specifically, risk_obj nThis records obstacles that are currently in candidate lane n, ahead of the vehicle, and have a potential collision risk with the vehicle at historical moments. For these obstacles, the vehicle still needs to maintain a safe distance from the obstacle when changing lanes, provided there is no risk of collision. The specific steps are as follows: Calculate the passable distance of the vehicle in the current lane. Specifically, based on the trajectory collision information, consider the collision distances between the vehicle and dynamic / static obstacles and virtual obstacles (such as intersections, dead ends, merge intersections, etc.) in the current lane, and calculate the farthest distance d_reach (i.e., the passable distance) that can be reached while maintaining the vehicle speed in the current lane. Calculate the lane-changing distance to the target lane (i.e., the candidate lane to which the vehicle changes lanes). Specifically, based on the vehicle's lane-changing trajectory in the target lane, the lane-changing distance d_lc when the vehicle completes the lane change and enters the target lane can be obtained. When d_reach > (d_lc + d_stop) (where d_stop is the safe following and stopping distance of the vehicle before the obstacle, calculated based on the IDM following model, the speed, acceleration, and position of the obstacle), it means that the passable distance of the current lane satisfies the condition that the vehicle can continue to drive in the current lane until it passes the obstacle before considering changing lanes to the target lane. While ensuring passability, the influence of the lane-changing trajectory and the distance to the obstacle needs to be considered additionally. In this case, risk_obj needs to be retained. n The retained risk_obj n The second feature obstacle information set risk_obj_list is formed.

[0055] S140. Determine the target lane for lane changing based on the first set of feature obstacle information and the second set of feature obstacle information, combined with a utility function.

[0056] After the interactive processing based on the predicted driving trajectory in steps S110-S130, a first feature obstacle information set obs_list and a second feature obstacle information set risk_obj_list are obtained. The first feature obstacles in the first feature obstacle information set obs_list represent feature obstacles within each candidate lane that directly affect lane-changing decisions, while the second feature obstacles in the second feature obstacle information set risk_obj_list represent feature obstacles within each candidate lane that indirectly affect lane-changing decisions. However, directly using the relative motion information between the vehicle and obstacles for lane-changing decisions often only considers the comparison of travel speeds between different candidate lanes, failing to comprehensively evaluate the performance of each candidate lane under factors such as sustainable traffic capacity, traffic efficiency, and traffic safety. Therefore, this application optimizes the above-mentioned problem by adopting a lane feature evaluation method based on a utility function, comprehensively evaluating the traffic efficiency, traffic safety, and sustainable traffic capacity among candidate lanes. The specific method is as follows:

[0057] Based on the information of each first feature obstacle in the first feature obstacle information set, determine the traffic efficiency utility of the corresponding candidate lane; based on the information of each second feature obstacle in the second feature obstacle information set, determine the traffic safety utility of the corresponding candidate lane; based on the travel distance of the vehicle in a candidate lane, determine the sustainable traffic utility of the corresponding candidate lane; determine the overall utility of the corresponding candidate lane based on the traffic efficiency utility, traffic safety utility, and sustainable traffic utility; determine the gain of the corresponding candidate lane based on the overall utility and the overall utility of the currently selected lane; and determine the candidate lane with a gain greater than the gain threshold and the largest gain as the target lane for lane changing decision.

[0058] Furthermore, determining the traffic efficiency utility of the corresponding candidate lane based on the information of each first feature obstacle in the first feature obstacle information set includes:

[0059] The traffic efficiency of a candidate lane is determined based on the collision time and collision distance when the first feature obstacle corresponding to a candidate lane in the first feature obstacle information set collides with the vehicle.

[0060] The traffic efficiency utility is determined based on the following first formula:

[0061] u p (n)=w ttc ×u ttc +w dist ×u dist ;

[0062] Among them, u p (n) represents the traffic efficiency utility of candidate lane n, u ttc and u dist These represent the utility value obtained by normalizing the collision time when the first feature obstacle in the first feature obstacle information set corresponds to the candidate lane n, and the utility value obtained by normalizing the collision distance when it collides with the vehicle; w ttc and w dist These represent the preset weights, w and w respectively. ttc +w dist =1. Optionally, when the first characteristic obstacle of candidate lane n is a static obstacle, w dist The value can be 0.7. When the first characteristic obstacle of candidate lane n is a dynamic obstacle, w dist The possible value is 0.2.

[0063] There are several ways to perform normalization. For example, one normalization method can be represented by the following function:

[0064]

[0065] UPPER represents the maximum value set, and LOWER represents the minimum value set.

[0066] Therefore, the above u ttc and u dist They can be represented as follows:

[0067] u ttc =sigmoid(ttc,TTC_UPPER,TTC_LOWER)

[0068] u dist =sigmoid(dist,DIST_UPPER,DIST_LOWER)

[0069] Where ttc represents the collision time when the first feature obstacle corresponding to candidate lane n collides with the vehicle, dist represents the collision distance when the first feature obstacle corresponding to candidate lane n collides with the vehicle, and TTC_UPPER, TTC_LOWER, DIST_UPPER, and DIST_LOWER represent the upper and lower bounds of the normalization process, which are set values.

[0070] The step of determining the traffic safety effectiveness of the corresponding candidate lane based on the information of each second feature obstacle in the second feature obstacle information set includes:

[0071] Determine the information of the second feature obstacle corresponding to a candidate lane from the second feature obstacle information set;

[0072] The normalized data representing the interaction distance when interacting with the vehicle in the information of the second feature obstacle is determined as the traffic safety utility of a candidate lane.

[0073] Traffic safety utility measures the proximity of a vehicle to an obstacle while following a lane-changing trajectory. The traffic safety utility u of candidate lane n is... s (n) can be represented as:

[0074] u s (n)=sigmoid(safe_dist, SAFE_UPPER, SAFE_LOWER)

[0075] Where, safe_dist represents the interaction distance between the second feature obstacle corresponding to candidate lane n and the vehicle, and SAFE_UPPER and SAFE_LOWER represent the upper and lower bound values ​​during normalization processing, respectively, which are set values.

[0076] Sustainable travel utility measures the maximum distance a vehicle can travel when changing lanes in candidate lane n, without deviating from the global path or changing lanes. This typically occurs when the permitted directions of travel on the road are inconsistent, such as when the global path is for left turns while the lane is for straight travel. Sustainable travel distance can be represented by `continue_dist`, and its utility is `u`. c (n) can be represented as:

[0077] u c (n)=sigmoid(continue_dist, CONTINUE_UPPER,

[0078] CONTINUE_LOWER and continue_dist represent the maximum distance a vehicle can travel when changing lanes in candidate lane n, without deviating from the global path or changing lanes. CONTINUE_UPPER and CONTINUE_LOWER represent the upper and lower bounds during normalization, respectively, and are set values.

[0079] In summary: the determination of the sustainable traffic utility of a candidate lane based on the travel distance of a vehicle within a candidate lane includes:

[0080] The normalized data of the distance traveled by a vehicle within a candidate lane is used to determine the sustainable traffic utility of a candidate lane.

[0081] The determination of the overall utility of the corresponding candidate lane based on the traffic efficiency utility, traffic safety utility, and sustainable traffic utility includes:

[0082] The overall utility of a candidate lane is obtained by weighted summing of its traffic efficiency utility, traffic safety utility, and sustainable traffic utility. Specifically, it can be expressed as: u(n) = w p ×u p (n)+w s ×u s (n)+w c ×u c (n).

[0083] Among them, w p w s and w c These represent the weights of traffic efficiency utility, traffic safety utility, and sustainable traffic utility when evaluating the overall utility of a lane. These weights can be determined as follows: when the vehicle is traveling in a lane on the globally planned path, w... c =0, w p =w s=0.5; When the vehicle is traveling in a lane of the globally planned path, and the passable distance between adjacent lanes is less than a set threshold (for example, this threshold can be 200m), w c =0.5, w p =w s =0.25; when the vehicle is not traveling in the lane of the globally planned path, w c =0.2, w p =w s =0.4.

[0084] Following the above processing method, the overall utility of each candidate lane can be obtained, and further gain evaluation can be performed on each candidate lane. Specifically, for example, when evaluating the gain of candidate lane n, the gain is compared with that of the currently selected lane. Let u(n) and u(selected) represent the overall utility of candidate lane n and the currently selected lane, respectively. Then, the gain R(n) of candidate lane n is calculated as follows:

[0085] R(n)=compare_sigmoid(u(n)-u(selected), COMPARE_UPPER, COMPARE_LOWER);

[0086] The meaning of compare_sigmoid(*, a, b) is as follows:

[0087] If * ≥ 0, compare_sigmoid = compare_sigmoid(*, a, b);

[0088] If * < 0, compare_sigmoid = —compare_sigmoid(—*, a, b).

[0089] In general, determining the gain of the corresponding candidate lane based on the overall utility and the overall utility of the currently selected lane includes:

[0090] Determine the difference between the overall utility of a candidate lane and the overall utility of the currently selected lane;

[0091] The normalized data of the difference is determined as the gain of a candidate lane.

[0092] After obtaining the gain of each candidate vehicle, the candidate lane with the largest gain that is greater than the gain threshold is determined as the target lane for lane changing decision.

[0093] The lane-changing decision-making method provided in this embodiment optimizes the extraction of feature obstacles within candidate lanes by considering the interaction information of the predicted driving trajectories of the vehicle and other traffic participants. Furthermore, it comprehensively considers factors such as traffic efficiency, traffic safety, and traffic capacity based on a utility function method to evaluate lane features, thereby improving the rationality and safety of lane-changing decisions. Specifically: First, when selecting lane feature obstacles, the selection method is optimized based on a collision detection method in the time dimension, which can fully consider the impact of obstacles on the vehicle's lane-changing behavior, thus improving lane-changing safety. In addition, the predicted intent type of obstacles is considered, and obstacles affecting the vehicle's lane-changing decision within a certain time window are simultaneously considered and screened. Therefore, the lane-changing decision-making method provided in this application can fully consider the potential impact of surrounding traffic participants on the vehicle's lane-changing decision, improving the safety of lane-changing behavior. Secondly, to address the issue of changes in the selection of feature obstacles due to changes in the relative positions of the vehicle and other traffic participants, which can lead to errors in lane information judgment, this application optimizes the stability of feature obstacle selection within the lane by considering the passable distance of the currently selected lane and the lane-changing safety distance of the candidate lanes, thus eliminating the irrational decision-making caused by sudden changes in results. Finally, this application proposes using a utility function-based method for lane feature evaluation, comprehensively measuring the utility gain of each candidate lane under factors such as sustainable traffic capacity, traffic efficiency, and traffic safety, avoiding the irrational and unstable decision-making caused by a single indicator, and improving the rationality of lane-changing decisions.

[0094] Based on the above embodiments, and generally referring to the following: Figure 3 The diagram illustrates a technical roadmap for a lane-changing decision-making method, specifically including: lane feature obstacle prediction based on predicted trajectory interaction – lane feature evaluation based on a utility function method – outputting the lane-changing decision result. The inputs include road information, vehicle trajectory information, and obstacle prediction information; or, more specifically, map information and prediction information from other surrounding traffic participants. The map information defines the range of lanes that the autonomous vehicle can choose when making lane-changing decisions, and the prediction information from other traffic participants is used to interact with the vehicle's trajectory to extract lane feature obstacles. This can be summarized into two steps: lane feature obstacle extraction based on predicted trajectory interaction and lane selection based on a utility function method, finally outputting the lane-changing decision result.

[0095] Further, see references such as Figure 4The flowchart shown illustrates a process for extracting lane feature obstacles based on predicted trajectory interaction. Specifically, it includes: within a time window, performing collision detection on each pair of sampling points on two trajectories and recording collision location, distance, and other information; filtering out obstacles whose collision locations are on the vehicle body and front, and recording the obstacle with the closest collision distance as a candidate obstacle set affecting lane features; for obstacles with predicted intent of CUT_IN and LEAVE type, calculating the relative position of the obstacle to the target lane at the time the vehicle body enters the target lane; obstacles whose centroid is located within the lane are considered to still affect the lane feature when the vehicle enters the target lane, and calculating the relative distance corresponding to the point on each trajectory that first meets the condition; for each lane, recording the obstacle with the smallest relative distance and adding it to the obstacle candidate set, and retaining the target lane feature obstacle according to the principle of closest interaction distance with the vehicle; comparing the passable distance of the currently selected lane with the lane-changing safety distance of the candidate target lane, and recording dangerous obstacles that require a certain safety distance when changing lanes in the absence of collision.

[0096] Figure 5 This is a schematic diagram of the structure of a lane-changing decision-making device according to an embodiment of this disclosure. Figure 2 As shown: The device includes: a first determining module 510, a second determining module 520, a third determining module 530 and a fourth determining module 540.

[0097] The system comprises the following modules: a first determining module 510, for each candidate lane, determining first key obstacle information affecting the vehicle's lane change to a candidate lane based on the predicted driving trajectory, wherein the first key obstacle information corresponding to different candidate lanes forms a first candidate obstacle information set; and determining second key obstacle information affecting the vehicle's lane change to a candidate lane based on the predicted driving trajectory of a reference obstacle with a preset intent, wherein the second key obstacle information corresponding to different candidate lanes forms a second candidate obstacle information set; a second determining module 520, comparing the first candidate obstacle information set with the second candidate obstacle information set, retaining the obstacle information with the closest interaction distance for a candidate lane based on the principle of closest interaction distance, wherein the obstacle information with the closest interaction distance for different candidate lanes forms a first characteristic obstacle information set; a third determining module 530, determining a second characteristic obstacle information set from the first characteristic obstacle information set based on the vehicle's passable distance in its current lane and the lane change distance required to change lanes to a candidate lane; and a fourth determining module 540, determining a target lane for lane change based on the first characteristic obstacle information set and the second characteristic obstacle information set combined with a utility function.

[0098] Furthermore, the first determining module 510 includes a first critical obstacle determining unit, specifically used for: within a preset time window, determining whether there is a collision risk between the obstacle and the vehicle based on the predicted driving trajectory of an obstacle within the vehicle's perception range and the expected lane-changing trajectory when the vehicle changes lanes to the candidate lane; if so, determining the obstacle as a first critical obstacle and recording the first critical obstacle information, the first critical obstacle information including at least the collision distance, collision time, and collision position of the vehicle when the obstacle collides with the vehicle; for multiple first critical obstacles, eliminating the first critical obstacles whose corresponding collision positions are at the rear of the vehicle to obtain one or more remaining first critical obstacles; and determining the first critical obstacle with the smallest corresponding collision distance among the remaining one or more first critical obstacles as the corresponding final first critical obstacle.

[0099] Furthermore, the preset intended reference obstacles include obstacles whose current position is not within the candidate lane and whose intention is to enter the candidate lane, and obstacles whose current position is within the candidate lane and whose intention is to exit the candidate lane. Correspondingly, the first determining module 510 includes a second key obstacle determining unit, specifically used for: determining the reference position of each reference obstacle when interacting with the vehicle based on the predicted driving trajectory of the reference obstacles; determining the reference obstacle with the smallest relative distance between its reference position and the vehicle as the second key obstacle, and recording the relative distance as the interaction distance when interacting with the vehicle, thereby obtaining the second key obstacle information.

[0100] Furthermore, it also includes a fifth determining module, used to predict the reference time required for the vehicle to enter the candidate lane when the vehicle travels according to the desired lane-changing trajectory and a preset speed before determining the second key obstacle information affecting the vehicle's lane change to a candidate lane based on the predicted driving trajectory of the reference obstacle with the preset intention; to determine the reference position of the reference obstacle at the time point corresponding to the reference time based on the predicted driving trajectory of the reference obstacle; if the reference position of the reference obstacle is within the candidate lane, the relative distance between the reference obstacle and the vehicle at the time point corresponding to the reference time is determined as the interaction distance when the reference obstacle and the vehicle interact.

[0101] Furthermore, the third determining module 530 is specifically used to: determine obstacles in the first set of characteristic obstacle information that pose a collision risk to the vehicle and are located in front of the vehicle in the candidate lane as candidate obstacles; determine the passable distance of the vehicle in the current lane and the lane-changing distance required to change lanes to the candidate lane; if the passable distance is greater than the sum of the lane-changing distance and the safe following distance, then determine the candidate obstacle as a second characteristic obstacle, wherein the safe following distance is determined based on the following model and the motion information of the candidate obstacle, and the second characteristic obstacles corresponding to different candidate lanes constitute the second set of characteristic obstacle information.

[0102] Furthermore, the fourth determining module 540 specifically includes: a first determining unit, used to determine the traffic efficiency utility of the corresponding candidate lane based on the information of each first feature obstacle in the first feature obstacle information set; a second determining unit, used to determine the traffic safety utility of the corresponding candidate lane based on the information of each second feature obstacle in the second feature obstacle information set; a third determining unit, used to determine the sustainable traffic utility of the corresponding candidate lane based on the travel distance of the vehicle in a candidate lane; a fourth determining unit, used to determine the overall utility of the corresponding candidate lane based on the traffic efficiency utility, traffic safety utility, and sustainable traffic utility; a fifth determining unit, used to determine the gain of the corresponding candidate lane based on the overall utility and the overall utility of the currently selected lane; and a sixth determining unit, used to determine the candidate lane with a gain greater than the gain threshold and the largest as the target lane for lane changing decisions.

[0103] Furthermore, the first determining unit is specifically used to determine the traffic efficiency of a candidate lane based on the collision time and collision distance when the first feature obstacle corresponding to a candidate lane in the first feature obstacle information set collides with the vehicle.

[0104] Specifically, the traffic efficiency utility is determined based on the following first formula:

[0105] u p (n)=w ttc ×u ttc +w dist ×u dist ;

[0106] Among them, u p (n) represents the traffic efficiency utility of candidate lane n, u ttc and u dist These represent the utility value obtained by normalizing the collision time when the first feature obstacle in the first feature obstacle information set corresponds to the candidate lane n, and the utility value obtained by normalizing the collision distance when it collides with the vehicle; w ttc and wdist These represent the preset weights, w and w respectively. ttc +w dist =1.

[0107] Furthermore, the second determining unit is specifically used to: determine the information of the second feature obstacle corresponding to a candidate lane from the second feature obstacle information set; and determine the normalized data representing the interaction distance when interacting with the vehicle in the information of the second feature obstacle as the traffic safety effect of a candidate lane.

[0108] Furthermore, the third determining unit is specifically used to: determine the normalized data of the travel distance of the vehicle in a candidate lane as the sustainable traffic utility of a candidate lane.

[0109] Furthermore, the fourth determining unit is specifically used to: perform a weighted summation of the traffic efficiency utility, traffic safety utility, and sustainable traffic utility of a candidate lane to obtain the overall utility of the candidate lane.

[0110] Furthermore, the fifth determining unit is specifically used to: determine the difference between the overall utility of a candidate lane and the overall utility of the currently selected lane; and determine the normalized data of the difference as the gain of a candidate lane.

[0111] The lane-changing decision-making device provided in this embodiment can execute the steps in the lane-changing decision-making method provided in this embodiment and can obtain the same beneficial effects, which will not be repeated here.

[0112] Figure 6 This is a schematic diagram of the structure of an electronic device according to an embodiment of this disclosure. See below for details. Figure 6 It shows a schematic diagram of a structure suitable for implementing the electronic device 500 in the embodiments of this disclosure. Figure 6 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0113] like Figure 6 As shown, the electronic device 500 may include a processing device (e.g., a central processing unit, a graphics processor, etc.) 501, which can perform various appropriate actions and processes to implement the methods of the embodiments described herein, based on a program stored in a read-only memory (ROM) 502 or a program loaded from a storage device 508 into a random access memory (RAM) 503. The RAM 503 also stores various programs and data required for the operation of the electronic device 500. The processing device 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0114] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts, thereby implementing the lane-changing decision method as described above. In such embodiments, the computer program can be downloaded and installed from a network via communication device 509, or installed from storage device 508, or installed from ROM 502. When the computer program is executed by processing device 501, it performs the functions defined in the methods of embodiments of this disclosure.

[0115] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-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 computer-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 computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0116] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to: for each candidate lane, determine first key obstacle information affecting the vehicle's lane change to a candidate lane based on a predicted driving trajectory, wherein the first key obstacle information corresponding to different candidate lanes constitutes a first candidate obstacle information set; and, based on the predicted driving trajectory of a reference obstacle with a preset intent, determine second key obstacle information affecting the vehicle's lane change to a candidate lane, wherein the second key obstacle information corresponding to different candidate lanes constitutes a second candidate obstacle information set; compare the first candidate obstacle information set with the second candidate obstacle information set, and for a candidate lane, retain the obstacle information with the closest interaction distance based on the principle of closest interaction distance, wherein the obstacle information with the closest interaction distance for different candidate lanes constitutes a first characteristic obstacle information set; determine a second characteristic obstacle information set from the first characteristic obstacle information set based on the vehicle's passable distance in its current lane and the lane change distance required to change lanes to a candidate lane; and determine a target lane for lane change based on the first characteristic obstacle information set and the second characteristic obstacle information set combined with a utility function.

[0117] Optionally, when one or more of the above-described procedures are executed by the electronic device, the electronic device may also perform other steps described in the above embodiments.

[0118] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, 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 devices, magnetic storage devices, or any suitable combination of the foregoing.

[0119] Option 1: A lane-changing decision-making method, the method comprising:

[0120] For each candidate lane, information on a first key obstacle affecting a vehicle's lane change to a candidate lane is determined based on the predicted driving trajectory, and the information on the first key obstacle corresponding to different candidate lanes forms a first set of candidate obstacle information; and information on a second key obstacle affecting a vehicle's lane change to a candidate lane is determined based on the predicted driving trajectory of a reference obstacle with a preset intent, and the information on the second key obstacle corresponding to different candidate lanes forms a second set of candidate obstacle information.

[0121] The first set of candidate obstacle information is compared with the second set of candidate obstacle information. For a candidate lane, the obstacle information with the closest interaction distance is retained based on the principle of closest interaction distance. The obstacle information with the closest interaction distance of different candidate lanes forms the first feature obstacle information set.

[0122] The second set of feature obstacle information is determined from the first set of feature obstacle information based on the vehicle's passable distance in the current lane and the lane-changing distance required to change lanes to a candidate lane;

[0123] The target lane for lane changing is determined based on the first set of feature obstacle information and the second set of feature obstacle information, combined with a utility function.

[0124] Option 2, according to the method described in Option 1, the step of determining the first key obstacle information affecting the vehicle's lane change to a candidate lane based on the predicted driving trajectory includes:

[0125] Within a preset time window, based on the predicted driving trajectory of an obstacle within the vehicle's perception range and the expected lane-changing trajectory when the vehicle changes lanes to the candidate lane, it is determined whether there is a risk of collision between the obstacle and the vehicle. If so, the obstacle is identified as the first critical obstacle, and the information of the first critical obstacle is recorded. The information of the first critical obstacle includes at least the collision distance, collision time, and collision position of the vehicle when the obstacle collides with the vehicle.

[0126] For multiple first key obstacles, the first key obstacles whose corresponding collision positions are at the rear of the vehicle are eliminated to obtain one or more remaining first key obstacles;

[0127] The first critical obstacle with the smallest collision distance among the remaining one or more first critical obstacles is identified as the corresponding final first critical obstacle.

[0128] Option 3: According to the method described in Option 1, the reference obstacle with the preset intention includes an obstacle whose current position is not in the candidate lane and whose intention is to cut into the candidate lane, and an obstacle whose current position is in the candidate lane and whose intention is to drive out of the candidate lane.

[0129] Correspondingly, the second key obstacle information affecting the vehicle's lane change to a candidate lane is determined based on the predicted driving trajectory of the reference obstacle with a preset intent, including:

[0130] Based on the predicted driving trajectory of reference obstacles, determine the reference position of each reference obstacle when interacting with the vehicle;

[0131] The reference obstacle with the smallest relative distance between the reference position and the vehicle is identified as the second key obstacle, and the relative distance is recorded as the interaction distance when interacting with the vehicle to obtain the information of the second key obstacle.

[0132] Option 4, according to the method described in Option 1, before determining the second key obstacle information affecting the vehicle's lane change to a candidate lane based on the predicted driving trajectory of the reference obstacle with a preset intent, the method further includes:

[0133] The reference time required for the vehicle to enter the candidate lane is predicted when the vehicle is traveling according to the desired lane-changing trajectory and the preset speed.

[0134] Based on the predicted driving trajectory of a reference obstacle, the reference position of the reference obstacle at the time point corresponding to the reference time is determined. If the reference position of the reference obstacle is within the candidate lane, the relative distance between the reference obstacle and the vehicle at the time point corresponding to the reference time is determined as the interaction distance when the reference obstacle interacts with the vehicle.

[0135] Option 5: According to the method described in Option 1, determining the second set of feature obstacle information from the first set of feature obstacle information based on the vehicle's passable distance in the current lane and the lane-changing distance required to change lanes to a candidate lane includes:

[0136] The obstacles in the first set of feature obstacle information that pose a collision risk to the vehicle and are located in front of the vehicle in the candidate lane are identified as candidate obstacles.

[0137] Determine the permissible distance of the vehicle in its current lane and the lane-changing distance required to change lanes to the candidate lane;

[0138] If the passable distance is greater than the sum of the lane-changing distance and the safe following distance, then the candidate obstacle is identified as the second feature obstacle. The safe following distance is determined based on the following model and the motion information of the candidate obstacle. The second feature obstacles corresponding to different candidate lanes constitute the second feature obstacle information set.

[0139] Option 6: According to the method described in Option 1, the step of determining the target lane for lane changing based on the first set of feature obstacle information and the second set of feature obstacle information combined with a utility function includes:

[0140] The traffic efficiency utility of the corresponding candidate lane is determined based on the information of each first feature obstacle in the first feature obstacle information set;

[0141] The traffic safety effectiveness of the corresponding candidate lane is determined based on the information of each second feature obstacle in the second feature obstacle information set;

[0142] The sustainable traffic utility of a candidate lane is determined based on the distance traveled by the vehicle within a candidate lane.

[0143] The overall utility of the corresponding candidate lane is determined based on the traffic efficiency utility, traffic safety utility, and sustainable traffic utility.

[0144] The gain of the corresponding candidate lane is determined based on the overall utility and the overall utility of the currently selected lane.

[0145] The candidate lane with the largest gain exceeding the gain threshold is identified as the target lane for lane-changing decisions.

[0146] Solution 7: According to the method described in Solution 6, the step of determining the traffic efficiency utility of the corresponding candidate lane based on the information of each first feature obstacle in the first feature obstacle information set includes:

[0147] The traffic efficiency of a candidate lane is determined based on the collision time and collision distance when the first feature obstacle corresponding to a candidate lane in the first feature obstacle information set collides with the vehicle.

[0148] Option 8: The method described in Option 7,

[0149] The traffic efficiency utility is determined based on the following first formula:

[0150] u p (n)=w ttc ×u ttc +w dist ×u dist ;

[0151] Among them, u p (n) represents the traffic efficiency utility of candidate lane n, u ttc and u dist These represent the utility value obtained by normalizing the collision time when the first feature obstacle in the first feature obstacle information set corresponds to the candidate lane n, and the utility value obtained by normalizing the collision distance when it collides with the vehicle; w ttcand w dist These represent the preset weights, w and w respectively. ttc +w dist =1.

[0152] Option 9: According to the method described in Option 6, the step of determining the traffic safety effectiveness of the corresponding candidate lane based on the information of each second feature obstacle in the second feature obstacle information set includes:

[0153] Determine the information of the second feature obstacle corresponding to a candidate lane from the second feature obstacle information set;

[0154] The normalized data representing the interaction distance when interacting with the vehicle in the information of the second feature obstacle is determined as the traffic safety utility of a candidate lane.

[0155] Option 10: According to the method described in Option 6, determining the sustainable traffic utility of the corresponding candidate lane based on the travel distance of the vehicle within a candidate lane includes:

[0156] The normalized data of the distance traveled by a vehicle within a candidate lane is used to determine the sustainable traffic utility of a candidate lane.

[0157] Option 11: According to the method described in Option 6, determining the overall utility of the corresponding candidate lane based on the traffic efficiency utility, traffic safety utility, and sustainable traffic utility includes:

[0158] The overall utility of a candidate lane is obtained by weighted summing of its traffic efficiency utility, traffic safety utility, and sustainable traffic utility.

[0159] Option 12: According to the method described in Option 6, determining the gain of the corresponding candidate lane based on the overall utility and the overall utility of the currently selected lane includes:

[0160] Determine the difference between the overall utility of a candidate lane and the overall utility of the currently selected lane;

[0161] The normalized data of the difference is determined as the gain of a candidate lane.

[0162] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

Claims

1. A lane-changing decision-making method, characterized in that, The method includes: For each candidate lane, information on a first key obstacle affecting a vehicle's lane change to a candidate lane is determined based on the predicted driving trajectory. The information on the first key obstacle corresponding to different candidate lanes forms a first set of candidate obstacle information. Additionally, information on a second key obstacle affecting a vehicle's lane change to a candidate lane is determined based on the predicted driving trajectory of a reference obstacle with a preset intent. The information on the second key obstacle corresponding to different candidate lanes forms a second set of candidate obstacle information. The reference obstacle with the preset intent includes obstacles whose current position is not within the candidate lane and whose intent is to enter the candidate lane, and obstacles whose current position is within the candidate lane and whose intent is to exit the candidate lane. The first set of candidate obstacle information is compared with the second set of candidate obstacle information. For a candidate lane, the obstacle information with the closest interaction distance is retained based on the principle of closest interaction distance. The obstacle information with the closest interaction distance of different candidate lanes forms the first feature obstacle information set. The second set of feature obstacle information is determined from the first set of feature obstacle information based on the vehicle's passable distance in the current lane and the lane-changing distance required to change lanes to a candidate lane; The traffic efficiency utility of the corresponding candidate lane is determined based on the information of each first feature obstacle in the first feature obstacle information set; The traffic safety effectiveness of the corresponding candidate lane is determined based on the information of each second feature obstacle in the second feature obstacle information set; The sustainable traffic utility of a candidate lane is determined based on the distance traveled by the vehicle within a candidate lane. The overall utility of the corresponding candidate lane is determined based on the traffic efficiency utility, traffic safety utility, and sustainable traffic utility. The gain of the corresponding candidate lane is determined based on the overall utility and the overall utility of the currently selected lane. The candidate lane with the largest gain exceeding the gain threshold is identified as the target lane for lane-changing decisions.

2. The method according to claim 1, characterized in that, The determination of the first key obstacle information affecting the vehicle's lane change to a candidate lane based on the predicted driving trajectory includes: Within a preset time window, based on the predicted driving trajectory of an obstacle within the vehicle's perception range and the expected lane-changing trajectory when the vehicle changes lanes to the candidate lane, it is determined whether there is a risk of collision between the obstacle and the vehicle. If so, the obstacle is identified as the first critical obstacle, and the information of the first critical obstacle is recorded. The information of the first critical obstacle includes at least the collision distance, collision time, and collision position of the vehicle when the obstacle collides with the vehicle. For multiple first key obstacles, the first key obstacles whose corresponding collision positions are at the rear of the vehicle are eliminated to obtain one or more remaining first key obstacles; The first critical obstacle with the smallest collision distance among the remaining one or more first critical obstacles is identified as the corresponding final first critical obstacle.

3. The method according to claim 1, characterized in that, The predicted driving trajectory based on a reference obstacle with a preset intent determines the second key obstacle information affecting the vehicle's lane change to a candidate lane, including: Based on the predicted driving trajectory of reference obstacles, determine the reference position of each reference obstacle when interacting with the vehicle; The reference obstacle with the smallest relative distance between the reference position and the vehicle is identified as the second key obstacle, and the relative distance is recorded as the interaction distance when interacting with the vehicle to obtain the information of the second key obstacle.

4. The method according to claim 1, characterized in that, Before determining the second key obstacle information affecting lane change to a candidate lane based on the predicted driving trajectory of the reference obstacle with a preset intent, the method further includes: The reference time required for the vehicle to enter the candidate lane is predicted when the vehicle is traveling according to the desired lane-changing trajectory and the preset speed. Based on the predicted driving trajectory of a reference obstacle, the reference position of the reference obstacle at the time point corresponding to the reference time is determined. If the reference position of the reference obstacle is within the candidate lane, the relative distance between the reference obstacle and the vehicle at the time point corresponding to the reference time is determined as the interaction distance when the reference obstacle interacts with the vehicle.

5. The method according to claim 1, characterized in that, The process of determining a second set of feature obstacle information from the first set of feature obstacle information based on the vehicle's passable distance in its current lane and the lane-changing distance required to change lanes to a candidate lane includes: The obstacles in the first set of feature obstacle information that pose a collision risk to the vehicle and are located in front of the vehicle in the candidate lane are identified as candidate obstacles. Determine the permissible distance of the vehicle in its current lane and the lane-changing distance required to change lanes to the candidate lane; If the passable distance is greater than the sum of the lane-changing distance and the safe following distance, then the candidate obstacle is identified as the second feature obstacle. The safe following distance is determined based on the following model and the motion information of the candidate obstacle. The second feature obstacles corresponding to different candidate lanes constitute the second feature obstacle information set.

6. The method according to claim 1, characterized in that, The step of determining the traffic efficiency utility of the corresponding candidate lane based on the information of each first feature obstacle in the first feature obstacle information set includes: The traffic efficiency of a candidate lane is determined based on the collision time and collision distance when the first feature obstacle corresponding to a candidate lane in the first feature obstacle information set collides with the vehicle.

7. A lane-changing decision-making device, characterized in that, include: The first determining module is used to determine, for each candidate lane, the first key obstacle information affecting the vehicle's lane change to a candidate lane based on the predicted driving trajectory. The first key obstacle information corresponding to different candidate lanes constitutes a first candidate obstacle information set. In addition, based on the predicted driving trajectory of the reference obstacle with the preset intention, the second key obstacle information affecting the vehicle's lane change to a candidate lane is determined, and the second key obstacle information corresponding to different candidate lanes constitutes a second set of candidate obstacle information. The reference obstacle with the preset intention includes an obstacle whose current position is not in the candidate lane and whose intention is to cut into the candidate lane, and an obstacle whose current position is in the candidate lane and whose intention is to drive out of the candidate lane. The second determining module is used to compare the first set of candidate obstacle information with the second set of candidate obstacle information. For a candidate lane, the obstacle information with the closest interaction distance is retained based on the principle of closest interaction distance. The obstacle information with the closest interaction distance of different candidate lanes forms the first feature obstacle information set. The third determining module is used to determine the second set of feature obstacle information from the first set of feature obstacle information based on the passable distance of the vehicle in the current lane and the lane-changing distance required to change lanes to a candidate lane; The fourth determining module is used to determine the traffic efficiency utility of the corresponding candidate lane based on the information of each first feature obstacle in the first feature obstacle information set; The traffic safety effectiveness of the corresponding candidate lane is determined based on the information of each second feature obstacle in the second feature obstacle information set; The sustainable traffic utility of a candidate lane is determined based on the distance traveled by the vehicle within a candidate lane. The overall utility of the corresponding candidate lane is determined based on the traffic efficiency utility, traffic safety utility, and sustainable traffic utility. The gain of the corresponding candidate lane is determined based on the overall utility and the overall utility of the currently selected lane; the candidate lane with the largest gain that is greater than the gain threshold is determined as the target lane for the lane-changing decision.

8. An electronic device, characterized in that, The electronic device includes: One or more processors; Storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-6.

Images