Method for evaluating safety of lane changing actions of vehicle in automatic

Abstract

The invention relates to a method for safety assessment of lane changing actions in an automated driving operation of a vehicle (EGO). According to the invention, before the start of a lane change action of the vehicle (EGO) from the left lane (F1) to the intermediate lane (F2) or from the right lane (F3) to the intermediate lane (F2) of the multilane section (F), the risk of collision is determined by an assumed lane change action of other vehicles (PE1 to PE3) on the right lane (F3) or the left lane (F1), and the risk of collision is determined on the basis of a maximum lane change duration and a cut-in time (tEM). Determining a longitudinal acceleration (ax, PE) of the other vehicles (PE1 to PE3) which results in a collision, evaluating the execution of the lane changing action as a function of a relative longitudinal position # imgabs0 # of the vehicle (EGO) with respect to the other vehicles (PE1 to PE3) at the start of the lane changing action and an initial relative longitudinal speed # imgabs1 # of the vehicle (EGO) with respect to the other vehicles (PE1 to PE3) on the basis of a collision probability (PKollision) as a safety indicator (S1) and a minimum distance (dx, min) in the absence of a collision as another safety indicator (S2); taking into account the jerk (j) magnitude of the longitudinal accelerations (ax, PE) of the other vehicles (PE1 to PE3) when determining the collision probability (PKollision).

Description

Technical Field

[0001] The present invention relates to a method for evaluating the safety of a lane change maneuver of a vehicle during an autonomous driving operation by means of an environmental sensor system, wherein the environment of the vehicle and the objects located therein are detected based on signals detected by the environmental sensor system. Background Art

[0002] US 8,244,408 B2 discloses a method for evaluating risks associated with driving operations of an autonomous vehicle control system. The vehicle is configured to perform an autonomous lane change maneuver and is equipped with a monitoring system. In this case, each of a plurality of objects located near the vehicle is monitored. The position of each object is predicted relative to the projected trajectory of the vehicle, and the level of collision risk between the vehicle and each object is evaluated.

[0003] Furthermore, EP 3 281 831 A1 describes a control system and a control method for determining the lane change probability of a motor vehicle ahead. The control system is designed to detect, by means of at least one environmental sensor, another motor vehicle participating in traffic in front of its own motor vehicle, determine the lateral movement of the other motor vehicle relative to the lane in which the other motor vehicle or its own motor vehicle is located, and calculate the movement-based lane change probability of the other motor vehicle based on the determined lateral movement of the other motor vehicle. Further, the control system is set and configured to: determine the current traffic situation based on environmental data obtained by the environmental sensor, calculate the traffic situation-based lane change probability of the other motor vehicle based on the determined current traffic situation, and calculate the total lane change probability of the other motor vehicle based on the movement-based probability and the traffic situation-based probability.

[0004] Furthermore, US2010 / 0 228 419A1 also discloses a method for evaluating the collision risk associated with vehicle operation, wherein the vehicle is designed to perform an autonomous lane change maneuver. The method includes the following steps:

[0005] - Monitoring each of a plurality of target vehicles located near the vehicle;

[0006] - Predicting the position of each target vehicle relative to the projected trajectory of the vehicle in a future time step; and

[0007] - Evaluating the level of collision risk between the vehicle and each target vehicle in a future time step.

[0008] DE 10 2019 129 879 A1 describes a method for automatically controlling a motor vehicle traveling on a current lane on a road, wherein the road has another lane. The method includes the following steps:

[0009] - Generate and receive two temporary driving actions / driving maneuvers, where the temporary driving actions include a lane change from the current lane to another lane and the start time point of the lane change, and the start time points of the two temporary driving actions are at different times;

[0010] - Compare the two driving actions considering their respective start time points; and

[0011] - Select one of the start time points based on the comparison.

[0012] In addition, DE 196 47 430 A1 describes a method for automatically braking a motor vehicle driven manually, in which the relative speed with respect to an obstacle located approximately in front of the vehicle in the driving direction is determined. In addition, the distance between the vehicle and the obstacle is also determined, and the determined distance is compared with the braking distance of the vehicle at a speed approximately equivalent to the relative speed. According to the comparison result, when the determined distance is less than the braking distance, an automatic braking process is executed. Summary of the Invention

[0013] The object of the present invention is to provide a new method for evaluating the safety of lane change actions in autonomous driving operations of a vehicle.

[0014] According to the present invention, this object is achieved by a method having the features described in claim 1.

[0015] Advantageous design solutions of the present invention are the subject matter of the dependent claims.

[0016] A method for evaluating the safety of lane change actions in autonomous driving operations of a vehicle by means of an environmental sensor system, in which the environment of the vehicle and the objects located therein are detected based on the signals detected by the environmental sensor system, and the method provides that:

[0017] - Before the start of a lane change action of the vehicle from the left lane to the middle lane or from the right lane to the middle lane of a multi-lane section, determine the collision risk by means of an assumed lane change action of other vehicles in the right lane or the left lane, where

[0018] - Calculate the longitudinal acceleration based on the maximum lane change duration and the cut-in moment, which causes a collision due to the overlap of the vehicle surfaces / vehicle areas of the vehicle and other vehicles,

[0019] - Evaluate the execution of the lane change action based on the relative longitudinal position of the vehicle with respect to other vehicles and the relative longitudinal speed of the vehicle with respect to other vehicles at the start of the lane change action, based on the collision probability as a safety indicator and the minimum distance in the case of no collision as another safety indicator.

[0020] - Consider the magnitude of the jerk of the longitudinal acceleration of other vehicles when determining the collision probability.

[0021] where

[0022] - Optimize the longitudinal acceleration of the vehicle such that the longitudinal acceleration changes required for the minimum and maximum longitudinal accelerations of other vehicles have statistically low collision probabilities, respectively.

[0023] According to the present invention, the collision probability as a safety indicator is determined based on the following parameters:

[0024] - The previously determined probability of the expected longitudinal acceleration of other vehicles,

[0025] - The initial situation,

[0026] - The geometric vehicle information of the vehicle and the geometric vehicle information of other vehicles,

[0027] - The start time point of the assumed lane change action of other vehicles,

[0028] - The duration of the lane change action, and

[0029] - The planned longitudinal acceleration of the vehicle.

[0030] In particular, the method provides that it is possible to check before the vehicle starts to change lanes whether it is still possible to safely execute the lane change when there may be a deviation in the vehicle's prediction of the lane change action of other vehicles or when the lane change cannot be foreseen based on the situation. For this reason, the collision probability is determined only based on longitudinal dynamics because it is impossible to predict whether other vehicles will execute a lane change. Here, longitudinal dynamics refers to both longitudinal acceleration and jerk.

[0031] By applying this method, based on the actual (especially measured) initial longitudinal acceleration of other vehicles potentially cutting into the automated vehicle's lane, the optimization of the longitudinal acceleration of the automated vehicle is designed such that the longitudinal acceleration changes required for the longitudinal acceleration of the potentially cutting-in vehicle have a statistically low occurrence probability, thereby reducing the collision probability.

[0032] Here, jerk refers to the instantaneous time rate of change of an object's acceleration. Especially in an electrically driven vehicle, a change in acceleration results in longitudinal jerk.

[0033] In particular, by applying this method, it is possible to evaluate / quantify the collision risk of the vehicle for a lane change action towards the middle lane at the tactical level.

[0034] A vehicle system for automated, in particular autonomous, driving operations can reduce the collision risk before a lane change maneuver by adjusting its set behavior, or delay the start of the lane change maneuver if both the positive acceleration consumption and the negative acceleration consumption of the vehicle are too high and / or until the initial situation for a safe lane change has improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Embodiments of the present invention will be explained in more detail below with reference to the drawings.

[0036] Wherein:

[0037] Figure 1 A section of road with three lanes and two vehicles is schematically shown;

[0038] Figure 2 Two illustrations of a section of road with the same initial state and a changed longitudinal acceleration are schematically shown;

[0039] Figure 3 The derivation of the cut-in moment in a specific situation is schematically shown;

[0040] Figure 4 The derivation of the cut-in moment in another specific situation is schematically shown;

[0041] Figure 5 The derivation of the cut-in moment in yet another specific situation is schematically shown;

[0042] Figure 6 The derivation of the cut-in moment in still another specific situation is schematically shown;

[0043] Figure 7 An illustration of the starting position limit case / critical case and its relative longitudinal speed change curve is schematically shown;

[0044] Figure 8 The derivation of the collision probability as a safety indicator is schematically shown;

[0045] Figure 9 An illustration of the calculation of the minimum distance between the vehicle and the closest other vehicle traveling in the next lane as another safety indicator is schematically shown,

[0046] Figure 10 An illustration of various limit cases for calculating the longitudinal acceleration limit is schematically shown,

[0047] Figure 11A A diagram of the collision probability density without considering jerk is schematically shown,

[0048] Figure 11BAnother graph schematically shows the collision probability density shifted by adjusting the vehicle longitudinal acceleration.

[0049] Figure 12A A graph schematically shows the collision probability density and the longitudinal acceleration limit value.

[0050] Figure 12B Schematically, a figure shows the calculation of the longitudinal acceleration difference for calculating the jerk limit value, and

[0051] Figure 12C A graph schematically shows the situation considering the occurrence probability of the longitudinal acceleration jerk. Detailed implementation mode

[0052] In all the drawings, the parts corresponding to each other are marked with the same reference numerals.

[0053] Figure 1 A section F is shown, which has three lanes F1 to F3 extending in the same direction. The vehicle EGO travels on the left lane F1 in an autonomous driving operation and intends to perform a lane change maneuver to the middle lane F2. Here, the lane change trajectory T1 of the vehicle EGO from the left lane F1 to the middle lane F2 is shown.

[0054] Another vehicle PE1 travels on the right lane F3, and it is possible (i.e., without an identifiable intention) that this other vehicle also intends to perform a lane change maneuver to the middle lane F2. Similarly, the assumed lane change trajectory T2 of this other vehicle PE1 from the right lane F3 to the middle lane F2 is also shown. Figure 1 The lane following trajectory ST of this other vehicle PE1 only related to the right lane F3 is also shown.

[0055] For the automated, especially autonomous, driving operation of the vehicle EGO, the lane change represents a relatively complex driving action. For this reason, it is necessary to consider the environmental situation to plan and implement the longitudinal and lateral movements of the vehicle EGO.

[0056] According to Donges and Michon, the evaluation of the lane change maneuver is carried out at three levels, namely the strategic level, the tactical level, and the operational level. The following problem description particularly relates to the tactical level, which describes the attractiveness and feasibility of the lane change maneuver. Generally, the autonomous lane change analysis at this level only includes object information, and the object information is assigned to its own lane (according to Figure 1In this embodiment, the left lane F1 and the target lane ZS (i.e., the middle lane F2). However, if the predicted behavior is not related to the target lane ZS, the object information of other vehicles PE1 to PE3 located in the next lane (i.e., the right lane F3) shown in the following figure is not considered, or is only considered indirectly, for example, through a potential field. The number of other vehicles PE1 to PE3 is not fixed at 3, but can vary. Therefore, lane changes or incorrect predictions that are not visible from the context are not considered, or are only considered through general recursive trajectories. However, in a three-lane or multi-lane section F, especially on a highway, during a lane change from the left lane F1 or the right lane F3 to the middle lane F2, it is possible that other vehicles PE1 to PE3 also decide to change to the same target lane ZS without a recognizable intention during the same time period. This situation is a relatively critical one.

[0057] Therefore, during the lane change maneuver, there is a risk of collision with other vehicles PE1 to PE3 that may change to the middle lane F2 during the same time period.

[0058] The human driver of vehicle EGO can assess the situation of the surrounding traffic during the lane change maneuver based on their previous experience, including considering the objects, i.e., traffic participants, in the next lane (refer to Figure 1 i.e., the right lane F3), in order to evaluate the attractiveness and feasibility of their tactical driving decisions; while the autonomous vehicle system relies on the measurement data of the environmental sensor system to evaluate the rule set of the planned tactical behavior. In particular, there is neither risk quantification, especially in the form of safety metrics S1, S2, nor a calculation rule for the desired set behavior of the autonomous driving system of vehicle EGO.

[0059] Therefore, in order to evaluate the lane change maneuver to the middle lane F2 at the tactical level by considering the object information in the next lane (i.e., the right lane F3), it is necessary to determine the measurement variables and parameters that allow the quantification of the collision risk of these objects during the lane change maneuver. Based on this, the desired set behavior of the autonomous driving system can be derived immediately.

[0060] In the following, a method for evaluating the safety of the lane change maneuver of vehicle EGO in autonomous driving operations using an environmental sensor system is described, where the environment of vehicle EGO and the objects located therein are detected based on the signals detected by the environmental sensor system.

[0061] To execute this method, it is assumed that the lane change maneuvers of other vehicles PE1 to PE3 are unpredictable.

[0062] In Figure 2 it shows a section F and the same initial situation Two images A1 and A2, where vehicle EGO is traveling in the left lane F1 and intends to perform a lane change maneuver into the middle lane F2. Another three vehicles PE1 to PE3 are traveling in the right lane F3.

[0063] The lane change maneuver of vehicle EGO into the middle lane F2 is checked for collisions by means of the assumed lane change maneuvers of the other vehicles PE1 to PE3, where the other vehicles PE1 to PE3 can be passenger cars or trucks. This means that each of the other vehicles PE1 to PE3 is a potential cut-in vehicle for vehicle EGO. The other vehicles PE1 to PE3 can also be other means of transportation, such as motorcycles, where an acceleration range is also determined here and the same principle is applied to the risk analysis regarding the lane change of vehicle EGO.

[0064] In order to check the lane change maneuver based on the assumed lane change maneuvers of the other vehicles PE1 to PE3, the longitudinal acceleration a is calculated by means of a linearized lateral profile / lateral section, in particular based on the maximum lane change duration and the cut-in moment x,PE which longitudinal acceleration would lead to a collision due to the overlap of the vehicle surfaces between vehicle EGO and one of the other vehicles PE1 to PE3. For this purpose, the corresponding longitudinal acceleration a x,PE at which a collision occurs is defined, which simultaneously describes the collision probability P Kollision or is equivalent to the collision probability P Kollision since the longitudinal acceleration a x,PE is directly related to the overlap of the vehicle surfaces and thus directly related to the collision.

[0065] The evaluation of the lane change maneuver at the tactical level is based on two safety indicators S1 and S2 at the start of the lane change maneuver and on the relative longitudinal position (also referred to as the initial distance between the vehicle centers) as vehicle measurement variable 1 and the initial relative longitudinal speed as vehicle measurement variable 2.

[0066] where the relative longitudinal position is as follows:

[0067]

[0068] The initial relative longitudinal speed is calculated as follows

[0069]

[0070] In particular, when vehicle EGO is traveling behind one of the other vehicles PE1 to PE3, the initial longitudinal distance is negative. Similarly, when vehicle EGO has a higher longitudinal speed v x,EGO,init than one of the other vehicles PE1 to PE3, the relative speed is positive.

[0071] The safety indicator S1 forms a collision probability P Kollision , and another safety indicator S2 represents the minimum distance d between the vehicle EGO and other vehicles PE1 to PE3 in the absence of a collision threat. x,min .

[0072] Regarding the collision probability P as the safety indicator S1 Kollision the evaluation is carried out by means of the following collision probability P Kollision : This collision probability is determined by the previously determined probability of the expected longitudinal acceleration a of other vehicles PE1 to PE3, with the help of the initial situation x,PE with the help of the geometric vehicle information l of the vehicle EGO and the geometric vehicle information l of other vehicles PE1 to PE3 EGO , with the help of the lane change start time points of other vehicles PE1 to PE3, the duration of the lane change action, and the planned longitudinal acceleration a of the vehicle EGO PE and is obtained. x,EGO,n is obtained.

[0073] Regarding the minimum distance d as another safety indicator S2 x,min the evaluation is carried out based on the minimum longitudinal distance between the closest bumpers to each other. Among them, the minimum distance d during the entire lane change action x,min is selected according to the intersection of the lateral coordinates between the vehicle EGO and at least one of the other vehicles PE1 to PE3.

[0074] Both safety indicators S1 and S2 are calculated based on models.

[0075] By changing the longitudinal acceleration a represented by the numerical label n x,EGO,n and / or the lane change duration, the vehicle EGO can affect the collision probability P Kollision during the lane change action and the minimum distance d x,min .

[0076] In the first figure A1, a scenario of three other vehicles PE1 to PE3 as potential cut-in vehicles into the middle lane F2 in their respective initial situations is shown.

[0077] For the vehicle EGO and the first other vehicle PE1, there is a total collision probability of the vehicle EGO wherein, in the first figure A1, with the longitudinal acceleration a ,x,EGO,0 the following minimum distance d from the corresponding other vehicles PE1 to PE3 is as follows x,min : In the case of the total collision probability this minimum distance is set to zero.

[0078] Figure 2 The second figure A2 in it shows the same initial situation as the first figure A1 where the vehicle EGO has a changed longitudinal acceleration a x,EGO,n such that a corresponding collision probability P is thereby generated Kollision and a corresponding minimum distance d x,min with new values

[0079] Therefore, the vehicle EGO can reduce the collision risk before starting a lane change maneuver or consciously postpone the start of the lane change maneuver if the positive or negative acceleration of the vehicle EGO is too high and / or until the initial situation for a safe lane change has improved

[0080] To carry out this method, it is necessary to define the start time point and end time point of the lane change maneuver in order to distinguish the current scenario from other scenarios. These time points are determined according to the method disclosed in the following resources

[0081] Vasile, Laurin, Kiran Divakar, and Dieter Schramm, "Deep-Learning Basierte Verkehrsteilnehmer Für Hochautomatisierte Spurwechsel (Behavior Prediction of Rear Traffic Participants for Highly Automated Lane Changes Based on Deep Learning)". Transforming Mobility-What Next?-Tagungsband zum 13. Wissenschaftsforum (Transforming Mobility-What's Next?-Proceedings of the 13th Traffic Science Forum): Springer-Verlag, Wiesbaden, 2021

[0082] Based on the start and end time points of the defined lane change maneuver, and based on the collected measurement data, the average longitudinal acceleration is determined from the actual driving dataset for performing the lane change maneuver according to the lane change direction (especially the lane change direction regarding the faster / slower lanes F1 to F3) and the vehicle category. In addition, the probability of the average longitudinal acceleration for performing the lane change maneuver is also determined. Among them, a probability density function pdf is generated respectively according to the lane change direction and the vehicle category in the frequency distribution, and its integral describes the probability of the corresponding acceleration range. The probability density function pdf is then applied to other vehicles PE1 to PE3 according to the lane change direction and its vehicle category

[0083] In addition, based on the said measurement data, the lane change duration t PE (Δy ZMDepending on the start and end time points, it is determined according to the distances Δy of vehicles PE1 to PE3 from the center ZM of the target lane T,ZM , the lane change direction (especially the lane change direction with respect to the faster / slower lanes F1 to F3), and the vehicle class. The lane change duration t PE (Δy ZM ) is determined by averaging multiple lane change maneuvers with similar distances Δy from the center ZM of the target lane PE,ZM .

[0084] With the aid of the determined longitudinal acceleration, a model is developed with which, based on the corresponding initial situation , the collision probability P as the safety indicator S1 can be determined, in particular calculated Kollision and the minimum distance d between the vehicle EGO and the other vehicles PE1 to PE3 as another safety indicator S2 x,min , and this model takes into account the possibility that the vehicle EGO has an impact by changing its longitudinal acceleration a x,EGO,n .

[0085] The longitudinal acceleration range and its probability are used to calculate the collision probability P as the safety indicator S1 Kollision . Here it is checked which longitudinal accelerations a x,PE of the corresponding other vehicles PE1 to PE3 during a lane change maneuver to the middle lane F2 would result in a collision with the vehicle EGO.

[0086] The lane change maneuver of the vehicle EGO is evaluated based on the initial situation of one or more other vehicles PE1 to PE3 on the right lane F3 . This evaluation is described by the initial distance between the two vehicle center points and the initial relative longitudinal speed . These two parameters are acquired with the aid of signals detected by an environmental sensor system of an automated, in particular autonomous, vehicle EGO.

[0087] Based on the initial situation , the minimum longitudinal acceleration a x,PE,min and the maximum longitudinal acceleration a x,PE,max of the other vehicles PE1 to PE3 are subsequently determined, for which, in the case of the linearized lateral profiles of the vehicle EGO and the other vehicles PE1 to PE3, a collision just still occurs during the lane change maneuver. Values within the longitudinal acceleration limit range (including the boundary values) also result in a collision.

[0088] The necessary longitudinal acceleration range a x,PE,min to a x,PE,max of the other vehicles PE1 to PE3 that leads to a potential collision can be determined by the longitudinal acceleration ax,EGO is affected, where different longitudinal accelerations a x,EGO,n are represented by the numerical label n.

[0089] The time period to be examined is defined by the maximum value t Ego of the lane change durations t max = max(t PE (Δy ZM ), t Ego ) of the vehicle EGO and the other vehicles PE1 to PE3.

[0090] In particular, this is because a longer lane change duration provides more time to reduce a higher initial relative longitudinal speed with a lower acceleration difference between the vehicle EGO and at least one of the other vehicles PE1 to PE3 and distance, which means a more critical situation. This situation will be further described below. In addition, the starting point of the time period during which a possible collision is to be examined is defined by the time point t EM of the cut-in process.

[0091] To determine the time point t EM of the cut-in process of two vehicles EGO, PE1 to PE3, i.e., the time point of the first lateral overlap of the two vehicle planes, the lateral movements of the vehicle EGO and the corresponding other vehicles PE1 to PE3 are linearized. Figures 3 to 6 A calculation rule is shown respectively, and four possible situations are shown.

[0092] Assuming a constant lateral speed, four possible time points t PE,ZM of a cut-in process can be calculated based on the initial distance Δy EM from the center ZM of the target lane.

[0093] Situation 1:

[0094] Figure 3 The illustrated embodiment shows possible intersections of the resulting straight lines representing the linearized lateral movements of the vehicle side (ZF) facing the vehicle.

[0095] If the two vehicle planes intersect before one of the two lane change maneuvers is completed, then:

[0096]

[0097] The condition is:

[0098]

[0099] t SPDescribes the postponement of the start of lane changes of the corresponding other vehicles PE1 to PE3. By assuming that there are lane change maneuvers that cannot be recognized in the scenario, the corresponding other vehicles PE1 to PE3 can also decide to change lanes to lanes F1 to F3 at any possible time point of the lane change maneuver of vehicle EGO.

[0100] By postponing the start of the lane change of the corresponding other vehicles PE1 to PE3 by t SP to a later time point, the time period to be examined is shortened.

[0101] The initial position and initial relative speed are calculated as follows:

[0102]

[0103] Assume that the initial longitudinal accelerations of the other vehicles PE1 to PE3 cannot be accurately measured or can only be measured inaccurately, and are assumed to be zero in the method described here.

[0104] y ZF,EGO (t SP ) = y ZF,EGO,init + v y,EGO t SP (7)

[0105]

[0106] The first possible contact point between vehicle EGO and the corresponding other vehicles PE1 to PE3 is also marked and shown in Figure 3 this.

[0107] Case 2:

[0108] If the corresponding other vehicles PE1 to PE3 reach the lateral end position of vehicle EGO after vehicle EGO has completed its lateral movement but before the end of the time period to be examined (t max - t SP ), then there is:

[0109]

[0110] The condition is:

[0111]

[0112] As Figure 4 shown in the embodiment of.

[0113] Case 3:

[0114] If the corresponding other vehicles PE1 to PE3 are within the time period to be examined (t max - t SP) reaches the lateral end position only after the end, but still touches the lane boundary line (y max -t SP ) of the target lane ZS within the investigated time period (t SB,PE ), then there is:

[0115] t EM = t max -t SP (11)

[0116] The condition is:

[0117]

[0118] Although there is no actual overlap between the vehicle surfaces, the fact that the two vehicles EGO, PE1 to PE3 remain side by side in the same lane F2 will be regarded as critical and will therefore be evaluated as an overlap.

[0119] Case 4:

[0120] If the corresponding other vehicles PE1 to PE3 reach their lateral end position y ZF,PE,end earlier than the vehicle EGO reaches the lateral end position y ZF,EGO,end of the corresponding other vehicles PE1 to PE3, then there is:

[0121]

[0122] The condition is:

[0123]

[0124] Case 4 is completed by formula (13).

[0125] By the following calculation rules, the minimum longitudinal acceleration and the maximum longitudinal acceleration of the corresponding other vehicles PE1 to PE3 are determined In the given initial situation

[0126] Values (including the boundary values) within these limit ranges will result in a collision:

[0127]

[0128] Acceleration difference limit situation:

[0129]

[0130] For Δv x (t SP)<0 (18)

[0131] Wherein: l EGO = vehicle length of EGO, to vehicle length of PE3

[0132]

[0133] Figure 7 Shows an explanation of the calculation rules.

[0134] The acceleration limit case is formed by the acceleration difference through which the relative velocity Δv x (t SP ) at the time point t when the lane change ends SP or the time point t of the cut-in process EM is completely eliminated.

[0135] According to the sign of the relative velocity Δv x (t SP ) at the time point t SP the initial limit case distance of the vehicle center point can be calculated This distance is necessary so that at the given relative velocity Δv x (t SP ) before the vehicles EGO, PE1 to PE3 move away from each other again, the last approaching point is the contact of the bumpers.

[0136] If the distance Δx of the vehicle center point MM (t SP ) at the time point t SP is between the boundaries determined in formulas (15) and (16), then the differential acceleration that can cause the bumpers of the vehicles EGO, PE1 to PE3 to contact each other (Δx SS,t = 0) before the vehicles EGO, PE1 to PE3 move away from each other again is sought. This situation occurs when the following conditions are met:

[0137]

[0138] After solving for t, only one solution is obtained. This is the case when the square root of the solution of the quadratic equation of the form ax 2 + bx + c = 0 is zero. Among them, the calculation of the bumper Δx SS,t contact time point is between the time point t of the cut-in process EM and the time point when the lane change action ends, and is calculated for each delay t SP of the start time point.

[0139] According to the relative velocity Δvx (t SP ) at time point t SP The number of possible cases in equations (21) and (22) changes depending on the sign of x (t SP ) case, the maximum longitudinal acceleration Determined by formula (21a), (21b) or (21c), the minimum longitudinal acceleration Only determined by formula (22d) or (22f). At negative relative speed Δv x (t SP ), the corresponding reverse is true, where the maximum longitudinal acceleration Determined by formula (21a) or (21c). Minimum longitudinal acceleration At positive relative speed Δv x (t SP ) is obtained by combining equations (22d) and (22f); and / or maximum longitudinal acceleration At negative relative speed Δv x (t SP ) is obtained by combining equations (21a) and (21c).

[0140] Figure 10 The limiting case positions of equations (15) to (18) and the various regions of equations (21a-c) and (22d-f) are shown.

[0141] Then, the longitudinal acceleration value a determined by equations (21a) to (21c) and (22d) to (22f) is x,PE,min and a x,PE,max As Figure 8 The integration limits shown are used to calculate the collision probability P as the safety indicator S1 Kollision To do this, the probability density function pdf determined at an earlier point in time is integrated. The collision probability P Kollision According to the postponement SP The weighted processing is performed, where the weight is Figure 8 The straight line shown Definition. In particular, Figure 8 The collision probability P is shown as the safety index S1. Kollision The derivation process.

[0142] The theoretical basis of this weighted straight line is as follows: The later the lane-changing actions of the corresponding other vehicles PE1 to PE3 start, the less time is available to complete the lane-changing actions, thus reducing the risk of collision. Further, it can be inferred from this that as the lane-changing action of vehicle EGO progresses, the probability of the other vehicles PE1 to PE3 initiating a lane change also decreases because the probability of the movement of vehicle EGO being perceived by the other vehicles PE1 to PE3 increases. Then, the individual weighted collision probabilities are summed to obtain the total collision probability. Total collision probability It can also be calculated without the weighted straight line and used as the safety index S1.

[0143]

[0144] In Figure 8 the upper region, two regions B1 and B2 are shown with different shadings. The first region B1 represents a possible lateral collision due to the overlap of the vehicle surfaces between vehicle EGO and a corresponding other vehicle PE1 to PE3.

[0145] The lower region B2 represents a possible longitudinal collision between vehicle EGO and the corresponding other vehicle PE1 to PE3.

[0146] A straight line forms an area below it In addition, the intersection point with the abscissa is determined by the last relevant start time point t for the lane change of the corresponding other vehicles PE1 to PE3. SP,End This last relevant start time point t SP,End represents the time point when the corresponding other vehicles PE1 to PE3 start their lane-changing actions, and at this time point, there is sufficient time for the lane boundary line SB of the target lane ZS to contact the vehicle surface facing vehicle EGO.

[0147] The intersection point G0 with the vertical axis is obtained by deforming the area below the straight line The slope of the straight line is determined as follows: The determination method is as follows:

[0148]

[0149] Subsequently, the total collision probabilities of all PEs are accumulated (AGKol = accumulated GKol, n PE = the number of potential cut-in vehicles).

[0150]

[0151] Total collision probability P AGKol,PE,n can be incorporated into any cost function for trajectory planning in order to calculate the optimal longitudinal acceleration a under various requirements and / or constraints regarding motorization, adhesion friction coefficient, comfort requirements, legal regulations, etc. x,EGO If the acceleration cost of the vehicle EGO calculated (which is necessary to rule out potential collisions) has too great an adverse impact on other requirements, the lane change maneuver may also be carried out at a later time point when the initial situation for performing a safe lane change maneuver has changed.

[0152] Figure 9 A diagram showing the relative longitudinal distance curve of the vehicle bumper is used to calculate the minimum distance d between the vehicle EGO and the corresponding other vehicles PE1 to PE3 without a collision occurring. x,min .

[0153] If no collision occurs, the minimum distance d x,min (also referred to as the minimum longitudinal distance) is used as another safety metric S2. Among them, the minimum distance d x,min is at a minimum at the cut-in time t EM or at the time t of the maximum lane change duration. max is at a minimum.

[0154] In particular, Figure 9 a diagram shows the relationship between the minimum distance d x,min and the relative longitudinal velocity Δv x (t SP = 0).

[0155] Here, the delay t of the start time point for initiating the lane change maneuvers of the corresponding other vehicles PE1 to PE3 SP is set to zero because when the two lane change maneuvers start simultaneously, there is the most time available to reduce the relative longitudinal distance.

[0156] The distance between the two opposing vehicle bumpers at the cut-in time and at the time point of the maximum lane change duration depends on the most critical acceleration of the corresponding other vehicles PE1 to PE3 depending on the initial situation :

[0157] Δa x,data,max,n = a x,Ego,n - a x,PE,data,max ; Δa x,data,min,n = a x,Ego,n - a x,PE,data,min (27)

[0158]

[0159] Depending on the difference in the situation, the minimum distance d x,minThe minimum value is obtained from the following formula:

[0160]

[0161] This method enables the safety assessment of an automated, in particular autonomous, vehicle EGO.

[0162] By changing the longitudinal acceleration a of the vehicle EGO x,EGO,n , the longitudinal acceleration a of other vehicles PE1 to PE3, which are potential cut-in vehicles necessary for a collision, x,PE can be offset so that it lies outside the critical range determined based on real driving data. Therefore, the vehicle EGO can reduce the collision risk before performing a lane change maneuver towards the middle lane F2, or consciously postpone the start of the lane change maneuver, thereby enhancing the safety of the vehicle EGO and other vehicles PE1 to PE3.

[0163] Figure 11A shows a graph of the collision probability density pd(a x,PE ) without considering the jerk j, and another graph of the probability density region where the offset is achieved by adjusting the longitudinal acceleration a of the vehicle EGO x,EGO .

[0164] Herein, as described above, it is assumed that the initial longitudinal accelerations of other vehicles PE1 to PE3 cannot be accurately measured or can only be inaccurately measured, and are assumed to be zero in the method described herein.

[0165] Furthermore, the purpose of this method is to offset the longitudinal acceleration limits a x,PE,min , a x,PE,max of other vehicles PE1 to PE3 to a low-probability region by adjusting the longitudinal speed of the vehicle EGO, thereby minimizing the collision probability P Kollision .

[0166] To this end, the occurrence probability of the corresponding longitudinal acceleration a x,PE causing a collision is defined, which simultaneously describes the collision probability P Kollision or is equivalent to the collision probability P Kollision , because the longitudinal acceleration a x,PE is directly related to the overlap with the vehicle plane and thus directly related to the collision.

[0167] Herein, problems as shown in Figure 11A and 11B may arise:

[0168] If the initial longitudinal acceleration of one of the other vehicles PE1 to PE3 is already within the range of low occurrence probability, then the longitudinal acceleration a of the vehicle EGOx,EGO Optimization of may shift the longitudinal acceleration limits a of other vehicles PE1 to PE3 x,PE,min , a x,PE,max to the following range: From a statistical perspective regarding the longitudinal acceleration a x,EGO , this range is optimal due to its low probability, but due to the actual, especially the measured initial situation this range encompasses or is close to the initial longitudinal acceleration of other vehicles PE1 to PE3 Therefore, if other vehicles PE1 to PE3 maintain their initial longitudinal acceleration on average during lane changes there is a risk of collision.

[0169] According to Figure 11A , the collision probability P is calculated as follows without considering the jerk j Kollision :[[]]

[0170]

[0171] And the collision probability P considering the jerk j as shown in Figures 12A - 12C is ≈ 0. Kollision

[0172] Figure 11B shows another graph where the probability density pd(a x,EGO ) is shifted by adjusting the longitudinal acceleration a x,PE .

[0173] According to the above method, the collision probability P is calculated as follows Kollision :[[]]

[0174]

[0175] And according to the new solution ≈ 50%.

[0176] Figures 12a - 12c show an alternative or supplementary solution to determine the collision probability P considering the jerk j of other vehicles PE1 - PE3 at the longitudinal acceleration a x,PE . The jerk j refers to the instantaneous time rate of change of the acceleration a Kollision .

[0177] Assuming that the initial longitudinal accelerations of other vehicles PE1 - PE3 can be measured accurately enough the initial longitudinal acceleration can then be determined, especially calculated and the difference between the two longitudinal acceleration limits a x,PE,min , a x,PE,max can be found.

[0178] ​Based on this difference, the minimum jerk \(j\) that must be applied on average for other vehicles PE1 - PE3 to enter the potential collision area during a lane change can be calculated. x；PE4Δmin and the maximum jerk \(j\) x,PE,4Δmax . This calculation process is as Figures 12A - 12B shown.

[0179] According to the calculation cases (21a) to (21c) and (22d) to (22f) for calculating the longitudinal acceleration limits \(a\) x,PE,min , \(a\) x,PE,max , the maximum jerk \(j\) x,PE,max is calculated as follows:

[0180]

[0181] \(t\) Fall \(\in t\) max , \(t\) EM .

[0182] According to the calculation cases (21a) to (21c) and (22d) to (22f) for calculating the longitudinal acceleration limits \(a\) x,PE,min , \(a\) x,PE,max , the minimum jerk \(j\) x,PE,min is calculated as follows:

[0183]

[0184] \(t\) Fall \(\in t\) max , \(t\) EM .

[0185] As Figure 12C shown, based on the data set, using the same method as the above process steps, the longitudinal jerk range is determined from the average value of the longitudinal acceleration change during a lane change, and its occurrence probability is described by a probability density function.

[0186] The values of the determined maximum jerk \(j\) x,PE,4Δmax and minimum jerk \(j\) x,PE,4Δmin are used as the integration limits of the jerk \(j\) probability density function for calculating the collision probability Kollision .

[0187] Compared with the above process steps, this solution additionally considers the jerk \(j\) when calculating the collision probability \(P\) Kollision . Thus, the longitudinal acceleration \(a\) x,EGO of vehicle EGO can be optimized such that the acceleration changes required for the minimum longitudinal acceleration \(a\) x,PE,min and maximum longitudinal acceleration of other vehicles PE1 - PE3 have a statistically low occurrence probability and thus a low collision probability.

[0188] Differently from that described above, the occurrence probability P of the jerk j required to achieve the average acceleration that would cause a collision Kollision , rather than the longitudinal acceleration a x,PE of the occurrence probability is used as the collision probability P Kollision .

Claims

1. A method for using an environmental sensor system to perform a safety assessment of a lane change action in an autonomous driving operation of a vehicle (EGO), wherein, Detecting the environment of the vehicle (EGO) and objects located in the environment based on signals detected by an environmental sensor system, wherein, - Before the start of a lane change operation of the vehicle (EGO) from the left lane (F1) of a multi-lane section (F) to the middle lane (F2) or from the right lane (F3) to the middle lane (F2), determining a collision risk by means of a hypothetical lane change operation of other vehicles (PE1 to PE3) on the right lane (F3) or the left lane (F1), - Based on the maximum lane change duration and the cut-in time (t EM ), determine the longitudinal acceleration (a x,PE ) of the other vehicles (PE1 to PE3) that causes a collision, where the collision occurs due to the overlap of the vehicle surfaces of the vehicle (EGO) and the other vehicles (PE1 to PE3), - Based on the relative longitudinal position of the vehicle (EGO) with respect to the other vehicles (PE1 to PE3) at the start of the lane change maneuver and the initial relative longitudinal speed of the vehicle (EGO) with respect to the other vehicles (PE1 to PE3) By means of the collision probability (P Kollision ) as a safety indicator (S1) and the minimum distance (d x,min ) in the case of no collision as another safety indicator (S2), to evaluate the execution of the lane change maneuver - When determining the collision probability (P Kollision ), consider the magnitude of the jerk (j) of the longitudinal acceleration (a x,PE ) of the other vehicles (PE1 to PE3), - Optimize the longitudinal acceleration (a x,EGO ) of the vehicle such that the longitudinal acceleration change costs required for the minimum longitudinal acceleration (a x,PE,min ) and the maximum longitudinal acceleration (a x,PE,max ) of the other vehicles (PE1 to PE3) respectively have a statistically low collision probability (P Kollision ). characterized in that, Based on the probability of the expected longitudinal acceleration (a x,PE ) of the previously determined other vehicles (PE1 to PE3), based on the initial situation Based on the geometric vehicle information of the vehicle (EGO) and the geometric vehicle information of the other vehicles (PE1 to PE3), based on the start time point of the assumed lane change action of the other vehicles (PE1 to PE3), the duration of the lane change action, and the planned longitudinal acceleration (a x,EGO,n ) of the vehicle (EGO), to determine the collision probability (P Kollision ) as a safety index (S1).

2. The method according to claim 1, Characterized in that, Determine two safety indicators (S1, S2) based on a model, and by changing the longitudinal acceleration (a x,EGO,n ) and / or the initial situation of the vehicle (EGO) relative to the other vehicles (PE1 to PE3) closest to it during driving to affect the collision probability (P Kollision ) and the minimum distance (d x,min ) during the execution of a lane change maneuver.

3. The method according to any one of the preceding claims, Characterized in that, Evaluate the minimum distance (d x,min ) as another safety indicator (S2) based on the minimum longitudinal distance from the bumper of the vehicle (EGO) to the bumpers of the other vehicles (PE1 to PE3) traveling closest thereto, where, in the case where the lateral coordinates of these vehicles (EGO, PE1 to PE3) intersect each other, select the minimum distance (d x,min ) during the lane change operation.

Images