Automatic driving lane change decision method, device, and storage medium

Abstract

The application relates to an automatic driving lane changing decision method and device and a storage medium, which comprises the following steps: acquiring vehicle position information and road information of a current navigation task, constructing five lanes, and taking the center line of each lane as the reference line of the lane; determining the behavior intention of the vehicle according to obstacle information, vehicle position information and road information of the current navigation task; determining whether to perform lane changing cost calculation according to the behavior intention of the vehicle, selecting the reference line of the lane with the minimum lane changing value as the target reference line; when the target reference line is the current lane, determining that the vehicle keeps the current lateral action; when the target reference line is the reference line of another lane, determining that lane changing is needed, calculating the space between any two obstacles of the target reference line according to the obstacle information of the lane corresponding to the target reference line, and determining the optimal space between the two obstacles as the lane changing space for lane changing; and the application can improve the decision efficiency and rationality of automatic driving lane changing.

Description

Technical Field

[0001] The present invention relates to the field of vehicle autonomous driving technology, and in particular to an autonomous driving lane change decision method, a device, and a storage medium thereof. Background Art

[0002] The current autonomous driving lane changing decision methods are divided into three categories: rule-based decision methods, AI reinforcement learning-based decision methods, and rule and AI fusion-based decision methods.

[0003] Among them, the rule-based decision-making method is the most widely used and is currently the mainstream method adopted in the industry. Its advantage is that it has a good space for behavioral interpretation. It can analyze the rationality of rule-making based on the performance of real vehicles or simulations, and adjust the rules as expected, so that the results of autonomous driving decisions meet expectations; but it also has the following disadvantages: it is strongly related to the flexibility of rule-making and decision-making. The more complex the rules, the higher the decision-making flexibility, but the difficulty of maintaining or adjusting the rules will increase; if the rules are simple, the decision-making flexibility will decrease, and it cannot effectively handle some special scenarios.

[0004] Among them, the decision-making method based on AI reinforcement learning is currently in the research stage and has few practical applications. Its advantage is that it only requires increasing the amount of training data to obtain better output results without designing complex rule devices; but it also has the following disadvantages: poor interpretability, the trained model cannot be adjusted in a targeted manner, and the erroneous results cannot be effectively improved. Therefore, it cannot be well applied to the field of autonomous driving with high safety requirements.

[0005] Among them, the decision-making method based on the fusion of rules and AI is a method derived from the combination of the above two methods. Its starting point is to combine the advantages of both to achieve better results; but in fact, the actual vehicle uses two sets of algorithms to work simultaneously, and then adds an arbitration module to determine the output results of the two. However, the arbitration module is based on rules and does not have a real arbitration effect. In addition, the simultaneous operation of the two sets of algorithms greatly increases the consumption of on-board computing power.

[0006] In summary, the current automatic driving lane changing decision methods all have shortcomings, so it is necessary to study an automatic driving lane changing decision method that can achieve better lane changing effects, so as to improve the efficiency and rationality of automatic driving lane changing decisions. Summary of the Invention

[0007] The purpose of the present invention is to propose an automatic driving lane change decision method and its device and storage medium to improve the decision efficiency and rationality of automatic driving lane change.

[0008] To achieve the above objectives, a first aspect of the present invention provides a method for autonomous driving lane change decision making, the method comprising:

[0009] Obtain the vehicle's location information and the road information of the current navigation task;

[0010] Taking the lane where the vehicle is located as the current lane, generate five lanes: the left adjacent lane, the left lane of the left adjacent lane, the right adjacent lane, and the right lane of the right adjacent lane, and the reference line of each lane is its center line;

[0011] Obtaining obstacle information detected by the vehicle perception module, and determining the vehicle's behavioral intention based on the obstacle information, the vehicle's position information, and the road information of the current navigation task;

[0012] Determining whether to perform lane change cost calculation based on the vehicle's behavioral intention; if so, calculating the lane change cost for each lane's reference line based on a preset cost function, and selecting the reference line of the lane with the smallest lane change cost as the target reference line; if not, determining that the vehicle maintains its current lateral action;

[0013] When the target reference line is the reference line of the current lane, it is determined that the vehicle maintains the current lateral action;

[0014] When the target reference line is the reference line of any one of the left adjacent lane, the left vehicle lane of the left adjacent lane, the right adjacent lane, and the right lane of the right adjacent lane, it is determined that a lane change is required, obstacle information of the lane corresponding to the target reference line is obtained, the space between any two obstacles in the lane corresponding to the target reference line is calculated based on the obstacle information of the lane corresponding to the target reference line, the space between the two optimal obstacles is selected as the lane change space, and the lane change is performed based on the lane change space.

[0015] Preferably, the vehicle's behavioral intention includes at least one of roadside starting, side parking, abandoning lane change, forced lane change, meeting, obstacle avoidance, lane change and overtaking, and lane deviation;

[0016] The determining of the vehicle's behavioral intention based on the obstacle information, the vehicle's position information, and the road information of the current navigation task includes:

[0017] If the vehicle is in the starting state and the distance between the vehicle's origin and the reference line of the current lane is greater than the preset roadside starting distance threshold, the vehicle's behavioral intention is determined to be a roadside start;

[0018] If the number of lane changes in the current lane is 0 and the distance between the vehicle and the end point of the navigation task is less than the preset pullover distance threshold, the vehicle's behavior intention is determined to be pullover parking;

[0019] If the vehicle is currently in the process of changing lanes and there is a risk of collision with an obstacle in the target lane, the vehicle's behavioral intention is determined to be to abandon the lane change;

[0020] If the distance between the vehicle and the end of the current lane is less than the total lane-changing distance after subtracting the length of the front solid line and the length of the current road intersection, it is determined that the vehicle's behavior intention is forced lane-changing;

[0021] If the vehicle is in the leftmost lane and the distance between the vehicle and the center line of the current lane is within a preset first threshold range, it is determined that the vehicle's behavior intention is meeting intention;

[0022] If the vehicle's navigation task planning route has a collision risk with the stationary obstacle in front, it is determined that the vehicle's behavior intention is obstacle avoidance;

[0023] If there is a moving obstacle in front of the current lane that limits the vehicle's driving speed, and the vehicle's speed meets the preset condition, it is determined that the vehicle's behavior intention is lane-changing and overtaking;

[0024] If there is an obstacle in front of the vehicle, and the distance between the obstacle boundary and the center of the current lane is within a preset second threshold range, it is determined that the vehicle's behavior intention is lane-internal deviation.

[0025] Preferably, the vehicle behavior intention is determined according to the obstacle information, the vehicle position information, and the road information of the current navigation task, comprising:

[0026] The moving route of the obstacle is determined according to the obstacle information, and the obstacles existing on each lane are determined according to the moving route; when the moving route of any obstacle is located on any lane, it is determined that the one lane exists the one obstacle.

[0027] Preferably, the vehicle behavior intention is determined according to the obstacle information, the vehicle position information, and the road information of the current navigation task, comprising:

[0028] The target obstacle with a collision risk with the vehicle is determined according to the obstacle information, and the vehicle behavior intention is determined according to the obstacle information of the target obstacle, the vehicle position information, and the road information of the current navigation task.

[0029] Preferably, the determination of whether to perform lane-changing cost calculation according to the vehicle behavior intention comprises:

[0030] When any of the vehicle behavior intentions of road side starting, side parking, giving up lane-changing, forced lane-changing, meeting, obstacle avoidance, lane-changing and overtaking, and lane-internal deviation is established, it is determined to perform lane-changing cost calculation;

[0031] When none of the vehicle behavior intentions of road side starting, side parking, giving up lane-changing, forced lane-changing, meeting, obstacle avoidance, lane-changing and overtaking, and lane-internal deviation is established, it is determined not to perform lane-changing cost calculation.

[0032] Preferably, the lane change cost value of the reference line of each lane includes at least one of a global navigation cost Global_cost, a distance cost Dis_cost, a speed cost Speed_cost, a lane change cost Lc_cost, a stability cost Stable_cost, a safety cost Safety_cost, and a side lane cost Side_cost.

[0033] Preferably, wherein:

[0034] The global navigation cost Global_cost is equivalent to the number of lane changes required for the vehicle to reach the end of the navigation task;

[0035] The distance cost Dis_cost is related to the distance between the vehicle and the end of the current lane. The closer the distance between the vehicle and the end of the current lane, the greater the corresponding distance cost Dis_cost.

[0036] Speed ​​cost = C*Speed_cost+B*A*V_max / v_ego; where V_max is the speed limit of the current lane, v_ego is the vehicle speed, and A, B, and C are preset coefficients.

[0037] The lane change cost Lc_cost represents the number of lane changes from the current lane to the target lane;

[0038] Stability cost Stable_cost = (T_stable – t_do_lc / T_stable) + dis_line + delta_angle; where T_stable is the preset time threshold, t_do_lc is the time after the vehicle completes a lane change, and when t_do_lc>T_stable, T_stable –

[0039] t_do_lc / T_stable = 0; dis_line is the lateral distance error between the vehicle origin and the target lane reference line, delta_angle is the tangent error between the vehicle heading and the target lane reference line;

[0040] When there is a collision risk in the target lane, the safety cost Safety_cost = MAX_COST1; when there is no collision risk in the target lane, the safety cost Safety_cost = 0; MAX_COST1 is the preset first-generation value;

[0041] When the current lane, the left adjacent lane and the right adjacent lane are all impassable, the side costs Side_cost of the current lane, the left adjacent lane and the right adjacent lane are 0, and the side costs Side_cost of the left vehicle lane of the left adjacent lane and the right lane of the right adjacent lane are MAX_COST2; when any one of the current lane, the left adjacent lane and the right adjacent lane is impassable, the side cost Side_cost of that lane is MAX_COST2, and the side costs Side_cost of the left vehicle lane of the left adjacent lane and the right lane of the right adjacent lane are 0; wherein MAX_COST2 is the preset second-generation value.

[0042] Preferably, the calculating, based on the obstacle information of the lane corresponding to the target reference line, the space between any two obstacles in the lane corresponding to the target reference line includes:

[0043] Calculate the speed V_start of the obstacle that the vehicle needs to overtake, the speed V_end of the obstacle that the vehicle needs to follow, the distance GAP-Min_s from the vehicle to the obstacle that needs to be overtaken, and the distance GAP-Max_s from the vehicle to the obstacle that needs to be followed based on the obstacle information of the lane corresponding to the target reference line;

[0044] When GAP-Max_s-GAP-Min_s <D_gap_0,判定对应的两个障碍物之间的空间不可用;其中,D_gap_0为预设的第一空间阈值;

[0045] When D_gap_0 <GAP-Max_s-GAP-Min_s<D_gap_1,且满足V_start> E*V_end, determines that the space between the corresponding two obstacles is unusable; where D_gap_0 is the preset second space threshold, and E is the preset coefficient;

[0046] When D_gap_0 <GAP-Max_s-GAP-Min_s<D_gap_1,且不满足V_start> E*V_end, determines whether the space between the corresponding two obstacles is available;

[0047] When D_gap_1 <GAP-Max_s-GAP-Min_s时,判定对应的两个障碍物之间的空间可用。

[0048] A second aspect of the present invention provides an automatic driving lane change decision device, configured to implement the automatic driving lane change decision method described in the first aspect. The device comprises:

[0049] Information acquisition module, used to obtain the vehicle's location information and the road information of the current navigation task;

[0050] The lane generation module is used to generate five lanes, namely, the left adjacent lane, the left vehicle lane of the left adjacent lane, the right adjacent lane, and the right lane of the right adjacent lane, with the reference line of each lane being its center line;

[0051] an intention determination module, configured to obtain obstacle information detected by the vehicle perception module and determine the vehicle's behavioral intention based on the obstacle information, the vehicle's position information, and the road information of the current navigation task;

[0052] a cost calculation module, configured to determine whether to perform lane change cost calculation based on the vehicle's behavioral intention; if so, calculate the lane change cost for each lane's reference line according to a preset cost function, and select the reference line of the lane with the smallest lane change cost as the target reference line; if not, determine that the vehicle maintains its current lateral action; and

[0053] a lane keeping module for determining that the vehicle maintains its current lateral action when the target reference line is the reference line of the current lane; and a lane changing calculation module for determining that a lane change is required when the target reference line is the reference line of any one of the left adjacent lane, the left vehicle lane of the left adjacent lane, the right adjacent lane, and the right lane of the right adjacent lane, obtaining obstacle information of the lane corresponding to the target reference line, calculating the space between any two obstacles in the lane corresponding to the target reference line based on the obstacle information of the lane corresponding to the target reference line, selecting the optimal space between the two obstacles as the lane changing space, and performing lane changing based on the lane changing space.

[0054] A third aspect of the present invention provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the automatic driving lane change decision method as described in the first aspect above is implemented.

[0055] The present invention has at least the following beneficial effects:

[0056] The embodiment of the present invention is a rule-based lane-changing decision. Compared with the traditional rule-based lane-changing decision, the present invention can divide the vehicle's actions into multiple intentions according to the actual scenario, and can cover L4 Robotaxi application scenarios; the autonomous driving lateral decision is divided into three parts: behavior intention determination, reference line cost calculation, and lane change space GAP calculation. The rule hierarchy is clear, which can cope with all working conditions and can be flexibly adjusted; at the same time, the embodiment of the present invention is based on the improvement of the five-lane model, which can avoid the problem of the three-lane model being unable to move forward after being congested during merging, thereby ensuring that the autonomous driving vehicle can smoothly reach the end point from the starting point of the navigation task, and can more effectively improve the efficiency and rationality of decision-making.

[0057] Other features and advantages of the present invention will be set forth in the description that follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required in the embodiments or the description of the prior art based on the preset signature authentication algorithm. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0059] Figure 1 The present invention is a flowchart of an automatic driving lane change decision method according to an embodiment of the present invention.

[0060] Figure 2 Schematic diagram of a five-lane model in one embodiment of the present invention.

[0061] Figure 3 Schematic diagram of lane-changing space in one embodiment of the present invention.

[0062] Figure 4 The figure is a flowchart of behavior intention determination in one embodiment of the present invention.

[0063] Figure 5 Schematic diagram of the framework of an automatic driving lane changing device in one embodiment of the present invention. DETAILED DESCRIPTION

[0064] The detailed description of the accompanying drawings is intended as an explanation of the presently preferred embodiments of the present invention and is not intended to represent the only form in which the present invention can be implemented. It should be understood that the same or equivalent functions can be accomplished by different embodiments that are intended to be included in the spirit and scope of the present invention.

[0065] One embodiment of the present invention provides an automatic driving lane change decision method, which is applied to an automatic driving vehicle, see Figure 1 The method of this embodiment includes the following steps:

[0066] Step S1: Obtain the vehicle's location information and the road information of the current navigation task;

[0067] Specifically, for an autonomous vehicle, the basic process of autonomous driving includes the navigation and positioning system positioning the vehicle, determining the starting and ending points, and performing navigation planning based on the starting and ending points to obtain a navigation task. The navigation task includes the driving route, starting and ending points, and lane changing operations during driving. The vehicle's position information and the road information of the current navigation task can be obtained from the navigation and positioning system of the autonomous vehicle.

[0068] Step S2: Taking the lane where the vehicle is located as the current lane, generate five lanes: the left adjacent lane, the left lane of the left adjacent lane, the right adjacent lane, and the right lane of the right adjacent lane, and the reference line of each lane is its center line;

[0069] Specifically, this embodiment proposes to construct a five-lane model to judge the behavior intention of the vehicle. The five-lane model includes the current lane, the left adjacent lane, the left lane of the left adjacent lane, the right adjacent lane, and the right lane of the right adjacent lane, as described in the steps. Figure 2 As shown, Figure 2 The dotted line in the figure is the reference line;

[0070] Step S3: Obtain obstacle information detected by the vehicle perception module, and determine the vehicle's behavioral intention based on the obstacle information, the vehicle's position information, and the road information of the current navigation task;

[0071] Specifically, to achieve autonomous driving, an autonomous driving vehicle is equipped with an on-board perception module. The on-board perception module may be, for example, an image perception module, a radar module, or other sensors, as well as an information processing module that identifies and processes the environmental images and radar signals detected by the image perception module and the radar module to obtain obstacle information, road information, and other information. In this embodiment, obstacle information can be obtained from the on-board perception module. As for how the on-board perception module specifically obtains obstacle information, the algorithms of different autonomous driving vehicles may vary, but they can all be applied in combination with the method of this embodiment. Therefore, the algorithm for obtaining obstacle information is not described in detail here. The obstacle information includes, but is not limited to, information such as obstacle speed, obstacle position, and obstacle movement direction.

[0072] Furthermore, in this embodiment, the actions of the vehicle can be divided into multiple vehicle behavioral intentions according to the actual scenario, such as roadside starting, side parking, abandoning lane change, forced lane change, meeting, obstacle avoidance, lane change overtaking, lane deviation, etc., which may require lane change. It can be understood that the automatic driving of the vehicle simulates human driving, and the driving process is mainly based on obstacles on the road and the driving route given by the navigation task. Therefore, the operation currently required by the vehicle, that is, the vehicle behavioral intention, can be determined based on the obstacle information, the vehicle position information, and the road information of the current navigation task.

[0073] Step S4: Determine whether to perform lane change cost calculation based on the vehicle's behavioral intention. If so, calculate the lane change cost for each lane reference line based on a preset cost function, and select the reference line of the lane with the smallest lane change cost as the target reference line. If not, determine that the vehicle maintains its current lateral action.

[0074] Specifically, the lane changing cost refers to the difficulty or risk of the vehicle changing from the current lane to the target lane, and the smaller the cost value is, the lower the difficulty or risk is; the cost value can be calculated by setting a cost function; when the intentions of road side starting, parking by the roadside, giving up lane changing, forced lane changing, meeting, obstacle avoidance, lane changing and overtaking, and lane deviation do not exist, it can also be understood that the behavior intention of the vehicle at this time is no lane changing intention, and the lane changing cost is not calculated;

[0075] Step S5, when the target reference line is the reference line of the current lane, determining that the vehicle keeps the current lateral action;

[0076] Specifically, when the target reference line is the reference line of the current lane, it means that there is no need to change lanes, the vehicle is already driving on the lane of the target reference line, and only the current action of the vehicle needs to be kept;

[0077] It should be noted that, since the vehicle behavior intentions of road side starting, parking by the roadside, obstacle avoidance, and lane deviation do not necessarily require the vehicle to change lanes, the cost value calculation of the reference line triggered by the vehicle behavior intentions of road side starting, parking by the roadside, obstacle avoidance, and lane deviation is mostly the cost value of the current lane being the smallest;

[0078] Step S6, when the target reference line is the reference line of any one of the left adjacent lane, the left lane of the left adjacent lane, the right adjacent lane, and the right lane of the right adjacent lane, it is determined that lane changing is needed, the obstacle information of the lane corresponding to the target reference line is obtained, the space between any two obstacles of the lane corresponding to the target reference line is calculated according to the obstacle information of the lane corresponding to the target reference line, one of the optimal spaces between the two obstacles is selected as the lane changing space, and lane changing is performed according to the lane changing space;

[0079] Specifically, the lane corresponding to the target reference line is taken as the target lane for lane changing, and when lane changing is performed, a suitable lane changing space GAP needs to be found. The lane changing space GAP, for example Figure 3 is shown in Figure 3 GAP1, GAP2, and GAP3 are shown in the figure, and the space between the two obstacles specifically refers to the obstacle that the vehicle needs to overtake and the obstacle that the vehicle needs to follow after successful lane changing. If the lane changing space GAP is too small, there is a risk of collision, and lane changing is not allowed. Only when the lane changing space GAP meets the lane changing requirement can lane changing be performed. If there are multiple lane changing spaces GAP that meet the lane changing requirement, one of the optimal ones can be selected, for example, the larger the space is, the safer it is.

[0080] In some embodiments, the vehicle behavior intention includes at least one of road side starting, parking by the roadside, giving up lane changing, forced lane changing, meeting, obstacle avoidance, lane changing and overtaking, and lane deviation.

[0081] wherein, the road information of the current navigation task includes a global navigation distance D_m of the ego vehicle to the end of the task, a total number of lane changes N_m required for the ego vehicle to reach the end of the current navigation task in the current lane, a number of lane changes N_r required for the ego vehicle in the current lane on the current road segment, a distance D_r of the ego vehicle to the end of the current lane, a distance D_j of the ego vehicle to the intersection of the current road segment, and a distance D_s of the ego vehicle to the front solid line, and of course, the global navigation distance D_m of the ego vehicle to the end of the task, the total number of lane changes N_m required for the ego vehicle to reach the end of the current navigation task in the current lane, the number of lane changes N_r required for the ego vehicle in the current lane on the current road segment, the distance D_r of the ego vehicle to the end of the current lane, the distance D_j of the ego vehicle to the intersection of the current road segment, and the distance D_s of the ego vehicle to the front solid line;

[0082] wherein, in the step S3, the ego vehicle behavior intention is determined according to the obstacle information, the ego vehicle position information, and the road information of the current navigation task, including:

[0083] (3.1) if the ego vehicle is in a starting state and the distance between the origin of the ego vehicle and the reference line of the current lane is greater than a preset roadside starting distance threshold, the ego vehicle behavior intention is determined to be roadside starting;

[0084] Specifically, the ego vehicle being in a starting state means that the ego vehicle has just started, and the roadside starting distance threshold in the embodiment is preferably but not limited to 1 meter; the origin of the ego vehicle refers to the origin of the vehicle coordinate system, which can be the mass center position of the vehicle;

[0085] (3.2) if the number of lane changes of the ego vehicle in the current lane is 0 and the distance of the ego vehicle to the end of the navigation task is less than a preset parking distance threshold, the ego vehicle behavior intention is determined to be parking by the roadside;

[0086] (3.3) if the ego vehicle is currently in a lane changing process and there is a collision risk with the obstacle on the target lane, the ego vehicle behavior intention is determined to be giving up lane changing;

[0087] Specifically, the collision risk determination formula is as follows:

[0088] The collision time ttc=(D_obj+D_safe) / |(v_obj-v_ego)|, wherein D_obj is the distance of the ego vehicle to the obstacle, D_safe is a preset safety distance threshold, v_obj is the speed of the obstacle, and v_ego is the speed of the ego vehicle;

[0089] When the collision time ttc is less than a preset collision threshold TTC, i.e. ttc

[0090] (3.4) If the distance from the vehicle to the end of the current lane minus the length of the solid line ahead and the length of the current road section intersection is less than the total lane change distance, the vehicle's intention is determined to be a forced lane change;

[0091] Specifically, the default lane change distance for the vehicle is set to D_lc, which is a calibration value and can be initially set to 400m. Then, when the distance from the vehicle to the end of the current lane minus the length of the solid line ahead and the length of the intersection is less than the total lane change distance, the forced lane change intention is established; that is, D_r–D_j–D_ls <D_lc*N_r,其中,D_j为路口长度,D_ls为前方实线长度;

[0092] (3.5) If the vehicle is in the leftmost lane and the distance D_l_re_obj between the oncoming vehicle in the left oncoming lane and the centerline of the vehicle's current lane is within a preset first threshold, the vehicle's behavior intention is determined to be a meeting intention;

[0093] Specifically, D_l_re_min <D_l_re_obj<D_l_re_max;其中,D_l_re_min和D_l_re_max为距离阈值,其均为标定值,确定本车行为意图为会车意图;

[0094] (3.6) If there is a risk of collision between the vehicle's planned navigation route and a stationary obstacle ahead, the vehicle's behavioral intention is determined to be obstacle avoidance.

[0095] Specifically, the discrete planned route points are expanded according to the vehicle's shape and tangent direction to form a vehicle driving space. If this space overlaps with the boundary of a stationary obstacle, the planned route is determined to be blocked and the vehicle needs to go around the obstacle. This means that the vehicle's behavioral intention is determined to be obstacle avoidance.

[0096] (3.7) If there is a moving obstacle in front of the current lane that limits the vehicle's speed, and the vehicle's speed meets the preset conditions, then the vehicle's intention is to change lanes and overtake;

[0097] Specifically, a moving obstacle that limits the vehicle's speed refers to a slow-moving obstacle. The slow-moving vehicle ahead affects the vehicle's speed, and in this case, the vehicle's intention is determined to be a lane-changing overtaking.

[0098] The vehicle speed meets the preset conditions as follows:

[0099] The current lane speed limit is V_max. When the vehicle speed is less than the speed limit threshold, that is, v_ego <A*V_max,认为此时速度受影响,对应的速度代价Speed_cost进行比例累加;

[0100] Speed_cost = C * Speed_cost + B * A * V_max / v_ego;

[0101] When the speed cost Speed_cost is greater than a threshold, i.e., Speed_cost > T_speedcost, it is determined that the lane changing and overtaking intention is established,

[0102] wherein A is a speed limit threshold coefficient, which is a calibrated value, and an initial value can be set as 0.7; C and B are speed cost accumulation proportion coefficients, B + C = 1, the greater B is, the faster the speed cost accumulates, and the overtaking and lane changing intention is triggered more quickly; T_speedcost is a speed cost threshold, which is a calibrated value, and an initial value can be set as 1;

[0103] (3.8) If there is an obstacle in front of the ego vehicle, and the distance D_l_obj from the boundary of the obstacle to the center of the current lane is within a preset second threshold range, it is determined that the behavior intention of the ego vehicle is lane-keeping within offset;

[0104] Specifically, when D_l_obj is within the preset second threshold range, i.e., D_l_obj_min < D_l_obj < D_l_obj_max, the lane-keeping within offset intention is triggered, wherein D_l_re_min and D_l_re_max are preset distance thresholds, which are calibrated values, D_l_obj_min should be greater than half the vehicle width, and D_l_obj_max plus half the lane width should be less than the vehicle width.

[0105] In some embodiments, the ego vehicle behavior intention is determined according to the obstacle information, the ego vehicle position information, and the road information of the current navigation task, comprising:

[0106] The moving route of the obstacle is determined according to the obstacle information, and the obstacle existing on each lane is determined according to the moving route; wherein when the moving route of any obstacle is located on any lane, it is determined that the one lane has the one obstacle.

[0107] Specifically, it should be noted that, as can be seen from the intention determination rules described above, it is necessary to know the obstacles on each lane in order to determine the behavior intention of the vehicle. Therefore, the above method is proposed in this embodiment to determine the obstacles on each lane; specifically, it should be understood that the movement route of the obstacle can be predicted based on the speed and movement direction of the obstacle. This part can be implemented by the on-board perception module of the autonomous driving system. This part belongs to the obstacle avoidance function in the autonomous driving technology and is not described in detail here; the obstacles on each lane are determined based on the movement route of the obstacle; if the movement route of an obstacle is located on a certain lane, it is determined that the obstacle exists on the lane, that is, if the movement route of an obstacle is located on multiple lanes, it is determined that the obstacle exists on multiple lanes.

[0108] In some embodiments, determining the vehicle's behavioral intention based on the obstacle information, the vehicle's position information, and the road information of the current navigation task includes:

[0109] A target obstacle with a collision risk with the vehicle is determined based on the obstacle information, and a behavior intention of the vehicle is determined based on the obstacle information of the target obstacle, the vehicle position information, and the road information of the current navigation task.

[0110] Specifically, in this embodiment, obstacles that do not pose a collision risk with the vehicle are screened and eliminated, and only obstacles that pose a collision risk (i.e., the target obstacles mentioned above) are further calculated, thereby reducing the amount of calculation; wherein, the target obstacles that pose a collision risk with the vehicle are determined based on the obstacle information, including whether there is a collision risk between the vehicle and the obstacle when the vehicle is traveling in the current lane, and whether there is a collision risk between the obstacle and the vehicle when the vehicle is changing lanes.

[0111] In some embodiments, in step S4, Figure 4 As shown, determining whether to calculate the lane change cost based on the vehicle's behavioral intention includes:

[0112] When any of the vehicle's behavioral intentions, including roadside starting, pulling over, abandoning lane change, forced lane change, meeting, obstacle avoidance, lane change and overtaking, or lane deviation, is established, the lane change cost calculation is determined;

[0113] When the vehicle's behavioral intentions of starting from the roadside, pulling over, giving up lane change, forced lane change, meeting, going around obstacles, changing lanes to overtake, and shifting within the lane are not established, the lane change cost calculation is determined not to be performed.

[0114] Specifically, in this embodiment, whether the intentions of starting from the roadside, parking by the side of the road, giving up lane change, forced lane change, meeting, bypassing obstacles, lane change and overtaking, and lane deviation are established are judged in turn. If none of them are established, it is determined that the vehicle has no intention to change lanes, and there is no need to calculate the lane change cost. The current lateral action is maintained. If any of the intentions is established, the lane change cost calculation is performed.

[0115] In some embodiments, the lane change cost value of the reference line of each lane includes at least one of a global navigation cost Global_cost, a distance cost Dis_cost, a speed cost Speed_cost, a lane change cost Lc_cost, a stability cost Stable_cost, a safety cost Safety_cost, and a side lane cost Side_cost;

[0116] The preset lane-changing cost function Total_cost is, for example:

[0117] Total_cost=Global_cos+Dis_cost+Speed_cost+Lc_cost+Stable_cost+Safety_cost+Side_cost;

[0118] The preset lane-changing cost function Total_cost is, for example:

[0119] Total_cost=Global_cos+Dis_cost+Speed_cost+Lc_cost+Stable_cost+Safety_cost;

[0120] In some embodiments, wherein:

[0121] The global navigation cost Global_cost is equivalent to the number of lane changes required for the vehicle to reach the end of the navigation task; that is, Global_cost = N_m;

[0122] The distance cost Dis_cost is related to the distance from the vehicle to the end of the current lane. The closer the distance from the vehicle to the end of the current lane, the greater the corresponding distance cost Dis_cost. Its calculation formula is as follows: Dis_cost = D_l * N_r / D_r;

[0123] Speed ​​cost = C*Speed_cost+B*A*V_max / v_ego; where V_max is the speed limit of the current lane, v_ego is the vehicle speed, and A, B, and C are preset coefficients.

[0124] The lane change cost Lc_cost represents the number of lane changes from the current lane to the target lane. Its function is to make the vehicle tend to change to the lane in the direction of the mission endpoint. When changing to a lane with a lower total lane change count, Lc_cost = -N_lc; otherwise, Lc_cost = N_lc. N_lc is the difference between the total lane change counts of the target lane and the current lane.

[0125] The stability cost Stable_cost takes into account the control deviation after the vehicle changes lanes. Therefore, a cost value is added to suppress continuous lane changes. The factors considered are time, angle deviation, lateral deviation, etc. The calculation formula is as follows:

[0126] Stable_cost = (T_stable – t_do_lc / T_stable) + dis_line + delta_angle; where T_stable is the preset time threshold, which is a calibration value and can be initially set to 3s. t_do_lc is the time it takes for the vehicle to complete a lane change. When t_do_lc > T_stable, the time cost of the first term in the above formula is ignored. dis_line is the lateral distance error between the vehicle's origin and the target lane reference line. delta_angle is the tangent error between the vehicle's heading and the target lane reference line.

[0127] Safety cost Safety_cost considers whether the target lane is safe. When the target lane has a collision risk, the safety cost Safety_cost = MAX_COST1. When the target lane has no collision risk, the safety cost Safety_cost = 0. MAX_COST1 is the preset first-generation value.

[0128] When the current lane, the left adjacent lane and the right adjacent lane are all impassable, the side costs Side_cost of the current lane, the left adjacent lane and the right adjacent lane are 0, and the side costs Side_cost of the left vehicle lane of the left adjacent lane and the right lane of the right adjacent lane are MAX_COST2; when any one of the current lane, the left adjacent lane and the right adjacent lane is impassable, the side cost Side_cost of that lane is MAX_COST2, and the side costs Side_cost of the left vehicle lane of the left adjacent lane and the right lane of the right adjacent lane are 0; wherein MAX_COST2 is the preset second-generation value.

[0129] Specifically, after the cost value calculation is completed, the cost values ​​of the reference lines of the five lanes are sorted, and the reference line with the smallest cost value is selected as the target reference line.

[0130] In some embodiments, the space between any two obstacles in the lane corresponding to the target reference line is calculated according to the obstacle information of the lane corresponding to the target reference line, comprising:

[0131] The speed V_start of the obstacle that the vehicle needs to overtake, the speed V_end of the obstacle that the vehicle needs to follow, the distance GAP-Min_s from the vehicle to the obstacle that needs to be overtaken, and the distance GAP-Max_s from the vehicle to the obstacle that needs to be followed are calculated according to the obstacle information of the lane corresponding to the target reference line.

[0132] When GAP-Max_s-GAP-Min_s

[0133] When D_gap_0

[0134] When D_gap_1

[0135] Specifically, when there are multiple obstacles on the reference line, multiple space GAPs will be formed, for example Figure 3 As shown in the figure, each space GAP is calculated, and the nearest available space GAP is selected for lane changing.

[0136] Referring to Figure 5 Another embodiment of the present application proposes an automatic driving lane changing decision device, which can be used to realize the automatic driving lane changing decision method as described in the above embodiment. The device of the present embodiment comprises the following functional modules:

[0137] The information acquisition module 1 is used to acquire the vehicle position information and the road information of the current navigation task.

[0138] Lane generation module 2 is used to generate five lanes, namely, the left adjacent lane, the left vehicle lane of the left adjacent lane, the right adjacent lane, and the right lane of the right adjacent lane, with the reference line of each lane being its center line;

[0139] Intention determination module 3, used to obtain obstacle information detected by the vehicle perception module, and determine the vehicle's behavioral intention based on the obstacle information, the vehicle's position information, and the road information of the current navigation task;

[0140] a cost calculation module 4 for determining whether to perform lane change cost calculation based on the vehicle's behavioral intention; if so, calculating the lane change cost for each lane reference line according to a preset cost function, and selecting the lane reference line with the smallest lane change cost as the target reference line; if not, determining that the vehicle maintains its current lateral action; and

[0141] The lane keeping module 5 is used to determine that the vehicle maintains the current lateral action when the target reference line is the reference line of the current lane; and the lane changing calculation module is used to determine that a lane change is required when the target reference line is the reference line of any one of the left adjacent lane, the left vehicle lane of the left adjacent lane, the right adjacent lane, and the right lane of the right adjacent lane, obtain obstacle information of the lane corresponding to the target reference line, calculate the space between any two obstacles in the lane corresponding to the target reference line based on the obstacle information of the lane corresponding to the target reference line, select the optimal space between the two obstacles as the lane changing space, and perform the lane changing based on the lane changing space.

[0142] It should be noted that the automatic driving lane changing device described in this embodiment corresponds to the automatic driving lane changing decision method described in the above embodiment. Therefore, the parts of the automatic driving lane changing device described in this embodiment that are not described in detail can be obtained by referring to the contents of the automatic driving lane changing decision method described in the above embodiment, so they will not be repeated here.

[0143] Furthermore, if the apparatus of the above embodiment is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0144] Another embodiment of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the automatic driving lane change decision method as described in the above embodiment.

[0145] Specifically, the computer-readable storage medium may include: any entity or recording medium that can carry the computer program instructions, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0146] Another embodiment of the present invention provides an electronic device, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the automatic driving lane change decision method as described in the above embodiment is implemented.

[0147] Wherein, electronic equipment can also include the bus connecting different components (including memory and processor).Memory can include computer-readable medium in the form of volatile memory, such as random access memory (RAM) and / or cache memory.Memory can also include at least one program product, and this program product has one group (such as at least one) program module, and these program modules are configured to perform the function of each embodiment of the present application.Electronic equipment can also communicate with one or more external devices (such as keyboard, pointing device, display etc.), can also communicate with one or more devices that enable users to interact with this electronic equipment, and / or communicate with any device (such as network card) that enables this electronic equipment to communicate with one or more other computing devices, this communication can be carried out through input / output (I / O) interface, and electronic equipment can also communicate with one or more networks (such as local area network (LAN), wide area network (WAN) and / or public network, such as the Internet) through network adapter.

[0148] It can be seen from the description of the above embodiments that the embodiments of the present invention have the following advantages:

[0149] The embodiment of the present invention is a rule-based lane-changing decision. Compared with the traditional rule-based lane-changing decision, the present invention can divide the vehicle's actions into multiple intentions according to the actual scenario, and can cover L4 Robotaxi application scenarios; the autonomous driving lateral decision is divided into three parts: behavior intention determination, reference line cost calculation, and lane change space GAP calculation. The rule hierarchy is clear, which can cope with all working conditions and can be flexibly adjusted; at the same time, the embodiment of the present invention is based on the improvement of the five-lane model, which can avoid the problem of the three-lane model being unable to move forward after being congested during merging, thereby ensuring that the autonomous driving vehicle can smoothly reach the end point from the starting point of the navigation task, and can more effectively improve the efficiency and rationality of decision-making.

[0150] Having described various embodiments of the application, it is to be understood that the above description is meant not to limit and not to encompass all of the possible embodiments covered by the claims. Many modifications and variations of this application can be apparent to those of ordinary skill in the art without departing from the spirit and scope of the described embodiments. It is intended that the scope of the application should only be limited by the appended claims.

Claims

1. A lane-changing decision method for an autonomous driving vehicle, characterized in that: The method comprises: Obtain the vehicle's location information and the road information of the current navigation task; Taking the lane where the vehicle is located as the current lane, generate five lanes: the left adjacent lane, the left lane of the left adjacent lane, the right adjacent lane, and the right lane of the right adjacent lane, and the reference line of each lane is its center line; Obtaining obstacle information detected by the vehicle perception module, and determining the vehicle's behavioral intention based on the obstacle information, the vehicle's position information, and the road information of the current navigation task; Determining whether to perform lane change cost calculation based on the vehicle's behavioral intention; if so, calculating the lane change cost for each lane's reference line based on a preset cost function, and selecting the reference line of the lane with the smallest lane change cost as the target reference line; if not, determining that the vehicle maintains its current lateral action; When the target reference line is the reference line of the current lane, it is determined that the vehicle maintains the current lateral action; When the target reference line is the reference line of any one of the left adjacent lane, the left vehicle lane of the left adjacent lane, the right adjacent lane, and the right lane of the right adjacent lane, it is determined that a lane change is required, obstacle information of the lane corresponding to the target reference line is obtained, the space between any two obstacles in the lane corresponding to the target reference line is calculated based on the obstacle information of the lane corresponding to the target reference line, the space between the two optimal obstacles is selected as the lane change space, and the lane change is performed based on the lane change space.

2. The automatic driving lane change decision method according to claim 1, wherein: The vehicle's behavioral intention includes at least one of roadside starting, pulling over, abandoning lane change, forced lane change, meeting, obstacle avoidance, lane change and overtaking, and lane deviation; The determining of the vehicle's behavioral intention based on the obstacle information, the vehicle's position information, and the road information of the current navigation task includes: If the vehicle is in the starting state and the distance between the vehicle's origin and the reference line of the current lane is greater than the preset roadside starting distance threshold, the vehicle's behavioral intention is determined to be a roadside start; If the number of lane changes in the current lane is 0 and the distance between the vehicle and the end point of the navigation task is less than the preset pullover distance threshold, the vehicle's behavior intention is determined to be pullover parking; If the vehicle is currently in the process of changing lanes and there is a risk of collision with an obstacle in the target lane, the vehicle's behavioral intention is determined to be to abandon the lane change; If the distance from the vehicle to the end of the current lane minus the length of the solid line ahead and the length of the current road section intersection is less than the total lane change distance, the vehicle's intention is determined to be a forced lane change; If the vehicle is in the leftmost lane and the distance between the oncoming vehicle in the left oncoming lane and the centerline of the vehicle's current lane is within a preset first threshold range, the vehicle's behavior intention is determined to be a passing intention; If there is a risk of collision between the vehicle's planned navigation route and a stationary obstacle ahead, the vehicle's behavior intention is determined to be obstacle avoidance. If there is a moving obstacle in front of the current lane that limits the vehicle's speed, and the vehicle's speed meets the preset conditions, the vehicle's behavior intention is determined to be a lane change and overtaking; If there is an obstacle in front of the vehicle and the distance from the obstacle boundary to the center of the current lane is within a preset second threshold range, it is determined that the vehicle's behavior intention is to deviate within the lane.

3. The automatic driving lane change decision method according to claim 2, wherein: The determining of the vehicle's behavioral intention based on the obstacle information, the vehicle's position information, and the road information of the current navigation task includes: The movement route of the obstacle is determined based on the obstacle information, and the obstacles existing on each lane are determined based on the movement route; wherein, when the movement route of any obstacle is located on any lane, it is determined that the obstacle exists on the lane.

4. The automatic driving lane change decision method according to claim 2, wherein: The determining of the vehicle's behavioral intention based on the obstacle information, the vehicle's position information, and the road information of the current navigation task includes: A target obstacle with a collision risk with the vehicle is determined based on the obstacle information, and a behavior intention of the vehicle is determined based on the obstacle information of the target obstacle, the vehicle position information, and the road information of the current navigation task.

5. The automatic driving lane change decision method according to claim 2, wherein: The determining whether to calculate the lane change cost according to the vehicle's behavioral intention includes: When any of the vehicle's behavioral intentions, including roadside starting, pulling over, abandoning lane change, forced lane change, meeting, obstacle avoidance, lane change and overtaking, or lane deviation, is established, the lane change cost calculation is determined; When the vehicle's behavioral intentions of starting from the roadside, pulling over, giving up lane change, forced lane change, meeting, going around obstacles, changing lanes to overtake, and shifting within the lane are not established, the lane change cost calculation is determined not to be performed.

6. The automatic driving lane change decision method according to claim 2, wherein: The lane change cost value of the reference line of each lane includes at least one of the global navigation cost Global_cost, the distance cost Dis_cost, the speed cost Speed_cost, the lane change cost Lc_cost, the stability cost Stable_cost, the safety cost Safety_cost, and the side lane cost Side_cost.

7. The automatic driving lane change decision method according to claim 6, wherein: in: The global navigation cost Global_cost is equivalent to the number of lane changes required for the vehicle to reach the end of the navigation task; The distance cost Dis_cost is related to the distance between the vehicle and the end of the current lane. The closer the distance between the vehicle and the end of the current lane, the greater the corresponding distance cost Dis_cost. Speed ​​cost = C*Speed_cost+B*A*V_max / v_ego; where V_max is the speed limit of the current lane, v_ego is the vehicle speed, and A, B, and C are preset coefficients. The lane change cost Lc_cost represents the number of lane changes from the current lane to the target lane; Stability cost Stable_cost = (T_stable – t_do_lc / T_stable) + dis_line + delta_angle; where T_stable is the preset time threshold, t_do_lc is the time after the vehicle completes a lane change, and when t_do_lc>T_stable, T_stable – t_do_lc / T_stable = 0; dis_line is the lateral distance error between the vehicle origin and the target lane reference line, delta_angle is the tangent error between the vehicle heading and the target lane reference line; When there is a collision risk in the target lane, the safety cost Safety_cost = MAX_COST1; when there is no collision risk in the target lane, the safety cost Safety_cost = 0; where MAX_COST1 is the preset first-generation value. When the current lane, the left adjacent lane, and the right adjacent lane are all impassable, the side costs Side_cost of the current lane, the left adjacent lane, and the right adjacent lane are 0, and the side costs Side_cost of the left lane of the left adjacent lane and the right lane of the right adjacent lane are MAX_COST2; when any one of the current lane, the left adjacent lane, and the right adjacent lane is passable, the side cost Side_cost of this one lane is MAX_COST2, and the side costs Side_cost of the left lane of the left adjacent lane and the right lane of the right adjacent lane are 0; where MAX_COST2 is the preset second-generation value.

8. The automatic driving lane change decision method according to any one of claims 1 to 7, wherein: Calculating the space between any two obstacles in the lane corresponding to the target reference line according to the obstacle information of the lane corresponding to the target reference line includes: Calculating the speed V_start of the obstacle that the vehicle needs to overtake, the speed V_end of the obstacle that the vehicle needs to follow, the distance GAP - Min_s from the vehicle to the obstacle that needs to be overtaken, and the distance GAP - Max_s from the vehicle to the obstacle that needs to be followed according to the obstacle information of the lane corresponding to the target reference line. When GAP - Max_s - GAP - Min_s < D_gap_0, it is determined that the space between the corresponding two obstacles is unavailable; where D_gap_0 is the preset first space threshold. When D_gap_0 < GAP - Max_s - GAP - Min_s < D_gap_1 and V_start > E * V_end, it is determined that the space between the corresponding two obstacles is unavailable; where D_gap_1 is the preset second space threshold and E is the preset coefficient. When D_gap_0 < GAP - Max_s - GAP - Min_s < D_gap_1 and V_start > E * V_end is not satisfied, it is determined that the space between the corresponding two obstacles is available. When D_gap_1 < GAP - Max_s - GAP - Min_s, it is determined that the space between the corresponding two obstacles is available.

9. An automatic driving lane change decision device, characterized in that: The device for implementing the autonomous driving lane-changing decision-making method according to any one of claims 1 to 8 includes: An information acquisition module for acquiring the vehicle position information and the road information of the current navigation task. A lane generation module for taking the lane where the vehicle is located as the current lane, generating five lanes including the left adjacent lane, the left lane of the left adjacent lane, the right adjacent lane, and the right lane of the right adjacent lane, and the reference line of each lane is its center line. An intention determination module for acquiring the obstacle information detected by the vehicle-mounted sensing module and determining the vehicle behavior intention according to the obstacle information, the vehicle position information, and the road information of the current navigation task. a cost calculation module, configured to determine whether to perform lane change cost calculation based on the vehicle's behavioral intention; if so, calculate the lane change cost for each lane's reference line according to a preset cost function, and select the reference line of the lane with the smallest lane change cost as the target reference line; if not, determine that the vehicle maintains its current lateral action; and a lane keeping module, configured to determine that the vehicle maintains its current lateral motion when the target reference line is the reference line of the current lane; The lane change calculation module is configured to determine that a lane change is required when the target reference line is the reference line of any one of the left adjacent lane, the left vehicle lane of the left adjacent lane, the right adjacent lane, and the right lane of the right adjacent lane; obtain obstacle information for the lane corresponding to the target reference line; calculate the space between any two obstacles in the lane corresponding to the target reference line based on the obstacle information for the lane corresponding to the target reference line; select the optimal space between the two obstacles as the lane change space; and perform the lane change based on the lane change space.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the automatic driving lane change decision method according to any one of claims 1 to 8 is implemented.

Images