Method of avoiding obstacles

Abstract

The invention relates to a method for evading objects (C1, C2, C3, C4) by a motor vehicle (10), comprising the following steps: - detecting objects located in the environment of the motor vehicle; - if several objects have been detected, acquiring data characterizing the position and / or the dynamics of each object; - verifying whether at least one proximity criterion between at least two of the detected objects is satisfied, and if so; - combining these two objects into one group (G2); - calculating data characterizing the position and / or the dynamics of the group; and - activating a system for obstacle avoidance and / or determining an evasive trajectory as a function of the data characterizing the position and / or the dynamics of the group.

Description

Technical Field

[0001] This invention generally relates to driving assistance for motor vehicles.

[0002] More specifically, the present invention relates to a method for avoiding obstacles.

[0003] The present invention also relates to a motor vehicle equipped with a computer designed to implement the method. Background Technology

[0004] To improve the safety of motor vehicles, these vehicles are now equipped with driver assistance systems or autonomous driving systems.

[0005] These systems are known, in particular, to include Automatic Emergency Braking (AEB) systems, which are designed to avoid any collision with obstacles located in the vehicle's lane by simply acting on the vehicle's conventional braking system.

[0006] However, in some situations, these emergency braking systems cannot prevent a collision or are unusable (for example, if another vehicle is following closely behind the vehicle).

[0007] To address these situations, advanced avoidance steering or automatic emergency steering (AES) systems have been developed, which enable vehicles to avoid obstacles by deviating from their trajectory through steering inputs.

[0008] To ensure the effectiveness of this AES function, it is necessary to reliably detect the parts of the motor vehicle environment that are related to calculating the optimal avoidance trajectory.

[0009] One parameter typically used for this purpose is called the Time to Collision (TTC). The second parameter is formed by the deviation required to allow the vehicle to pass the side of a detected obstacle without colliding with it. Therefore, each potential obstacle is considered individually to determine the most dangerous one and derive the optimal avoidance trajectory from it.

[0010] However, it has been found that by relying on these parameters, the AES obstacle avoidance function will activate in autonomous mode (without driver intervention) in some dangerous situations (where it is preferable for the driver to perform the avoidance maneuver). In another situation, the AES obstacle avoidance function may be unable to activate due to too many obstacles to avoid in the same area, rendering the function relatively unusable. Summary of the Invention

[0011] To overcome the aforementioned shortcomings of the prior art, the present invention proposes not to process detected objects independently, but to group them together when possible.

[0012] More specifically, according to the present invention is a method for motor vehicles to avoid objects initially considered potential obstacles. The method includes the following steps: - Detect objects in the environment of a motor vehicle. - If multiple objects have been detected, acquire data representing the position and / or dynamics of each detected object. - Check whether at least one proximity criterion is met between at least two of the detected objects, and if so, - Combine these two objects into one group. - Calculate data characterizing the location and / or dynamics of the group, and - Activate the obstacle avoidance system and / or determine the avoidance trajectory based on data characterizing the position and / or dynamics of the group.

[0013] The method of grouping targets has many advantages.

[0014] The main advantage is that this means the AES function can be activated in more situations than usual.

[0015] In fact, when there are too many objects to process, it is usually stipulated that the AES function should be inactive. In this case, a large number of individual objects can be processed by grouping them.

[0016] Similarly, when a vehicle is traveling on a two-lane road and there are one or more objects in each lane, it is generally prohibited to activate the AES function. In contrast, in this invention, only the deviations between various objects or groups of objects are considered in order to check whether the AES function can be activated.

[0017] Furthermore, it is typically stipulated that the lane in which each object is located be detected before activating the AES function. When an object spans two lanes, the clearance for passing beside it is more limited compared to when the object is in the center of its lane. Therefore, a larger safety margin is usually considered before activating the AES function. In contrast, in this invention, each target group defines an area to be avoided, the location of which is independent of the lane position. Therefore, the safety margin to be considered can be limited, allowing the AES function to be activated in more situations.

[0018] More generally, this method can ignore the concept of lanes when calculating avoidance trajectories, and instead preferably manage the environment as a single space in which the object moves.

[0019] It should also be noted that the present invention can simplify calculations.

[0020] This allows for the management of a large number of targets simultaneously, which has proven particularly beneficial when detecting a group of cyclists.

[0021] It should also be noted that this method does not focus on the category of the detected objects (cyclists, motor vehicles, etc.), but rather prefers to group objects of any category only according to one or more proximity criteria.

[0022] The following are other advantageous and non-limiting features of the method according to the invention, which can be considered individually or in any technically possible combination: - Calculate data characterizing the position and / or dynamics of the group based on data that individually characterizes the objects in the group; - Next, without considering these data that individually represent the objects in this group: - Activate the obstacle avoidance system and / or specifically determine the avoidance trajectory, independent of the data that individually represent the objects in this group; - Proximity criteria involve the lateral distance between two objects (an axis orthogonal to the tangent of the road at the level of the object). - In the acquisition step, one of the data is the lateral trajectory deviation that the motor vehicle must take to avoid each object; - The proximity criterion includes checking whether the difference between the lateral trajectory deviation to be taken to avoid the first object on the side oriented toward the second object and the lateral trajectory deviation to be taken to avoid the second object on the side oriented toward the first object is greater than or equal to a predetermined threshold. - This threshold is greater than or equal to 0; - If at least three objects have been detected, it is stipulated that these objects be sorted in a continuous order from one edge of the road to the other, and then it is checked whether the proximity criteria are met between each pair of consecutive objects in the continuous order. - The relative lateral velocity is calculated based on the deviation between the lateral velocity of the motor vehicle relative to the road it travels in a first reference frame oriented along the tangent of the road at the level of the motor vehicle and the lateral velocity of the object relative to the road in a second reference frame oriented along the tangent of the road at the level of the object, and then each lateral trajectory deviation is determined based on the relative lateral velocity. - The proximity criterion refers to the longitudinal distance between two objects; - In the acquisition step, one of the data points involves the remaining time before the motor vehicle collides with each object; - To check whether the proximity criteria are met, check whether the deviation between the collision times of the two objects is less than a threshold. - This threshold is greater than or equal to 0; In the acquisition step, one of the data is the lateral trajectory deviation to be taken to avoid each object on the same left or right side, and in the calculation step, one of the data representing the group is selected as the maximum value among the lateral trajectory deviations to be taken to avoid objects on the same side in the group. In the acquisition step, one of the data representing each object is the remaining time before the motor vehicle hits each object, and in the calculation step, one of the data representing the group is selected as the minimum value of the remaining time before the motor vehicle hits one of the objects in the group.

[0023] The present invention also relates to a motor vehicle including at least one steering wheel, wherein the steering system for each steering wheel is designed to be operated by a computer-controlled actuator, the computer being designed to implement the triggering method as described above.

[0024] Of course, the various features, variations and embodiments of the present invention can be combined with each other in various combinations, as long as they are not incompatible or mutually exclusive. Attached Figure Description

[0025] The following description, given by way of non-limiting example with reference to the accompanying drawings, will provide a better understanding of the scope of the invention and how it can be practiced.

[0026] In the attached diagram: [ Figure 1 [Illustration] is a schematic diagram of a motor vehicle and two target vehicles traveling in two different lanes according to the present invention; [ Figure 2 ]yes Figure 1 A diagram of the motor vehicle and one of the two target vehicles; [ Figure 3 ] is with Figure 2 The similar view shown illustrates the second step in the process of determining the location of one of the target cars. [ Figure 4 ] is with Figure 2 The similar views shown illustrate four reference frames used as part of this invention; [ Figure 5 ]yes Figure 4 The representation of the four reference frames in the diagram; [ Figure 6 ]yes Figure 1 A schematic diagram of one of the target vehicles and its driving lane; [ Figure 7 [Illustration] is a schematic diagram of an exemplary configuration of the motor vehicle and four vehicles traveling nearby, according to the present invention. [ Figure 8 ] is with Figure 7A similar view, in which the car's reference numerals have been reordered; [ Figure 9 ] is with Figure 8 A similar view, in which the cars have been grouped for the first time; [ Figure 10 ] is with Figure 8 A similar view, in which the cars have been grouped a second time; [ Figure 11 [Illustration of another exemplary configuration of a group of three vehicles, including the motor vehicle according to the invention, and the vehicle traveling nearby;] [ Figure 12 [Illustration] is a schematic diagram of another exemplary configuration of the motor vehicle according to the invention and three vehicles traveling nearby. Detailed Implementation

[0027] Figure 1 The illustration shows a motor vehicle 10 traveling on a road with two "objects" on the road that could potentially form obstacles to the motor vehicle 10. In this case, the two objects are formed by cars C1 and C2. As a variation, the two objects could be other types of objects (pedestrians, cyclists, etc.). Preferably, the objects in question are moving.

[0028] In the remainder of the specification, motor vehicle 10 is the vehicle on which the invention will be implemented and will be referred to as "this vehicle 10".

[0029] The vehicle 10 typically includes a chassis defining the passenger compartment, wheels, at least two of which are steering wheels, a powertrain, a braking system, and a conventional steering system for orientation acting on the steering wheels.

[0030] In the example under consideration, the steering system is controlled by an auxiliary steering actuator that allows the steering wheel to be oriented based on the steering wheel orientation and / or, in certain cases, based on instructions issued by a computer C10.

[0031] The computer C10 includes at least one processor, at least one memory, and various input and output interfaces.

[0032] The C10 computer is able to receive input signals from various sensors thanks to its input interface.

[0033] These sensors include, for example, the following: - Devices used to identify the vehicle's position relative to its lane, such as a front-facing camera. - Devices used to detect obstacles located on the trajectory of this vehicle 10, such as radar or lidar remote detectors. - At least one lateral device, such as a radar or lidar remote detector, for observing the environment to the side of the vehicle.

[0034] Therefore, computer C10 receives data from multiple sensors relating to objects present in the environment of the vehicle 10. Typically, this data is combined to provide reliable fused data about each object.

[0035] The computer C10 can transmit commands to the auxiliary steering actuator through its output interface.

[0036] Therefore, if conditions permit, it can be ensured that the vehicle follows the obstacle avoidance trajectory as closely as possible.

[0037] The computer C10 uses its memory to store data used as part of the following method.

[0038] The computer stores, in particular, computer application programs consisting of computer programs including instructions that, when executed by a processor, enable the computer to perform the methods described below.

[0039] These procedures include, in particular, an "AES system" designed to calculate an obstacle avoidance trajectory and control the vehicle 10 to follow that trajectory, or to assist the driver in controlling the vehicle 10 to follow that trajectory. The AES system has an autonomous mode and a manual mode. In autonomous mode, following the trajectory requires no driver assistance, while in manual mode, the AES system assists the driver in avoiding obstacles while the driver retains control. Since this AES system is well-known to those skilled in the art, it will not be described in detail here.

[0040] The computer program also includes activation software for activating the AES system, which can determine whether the AES system should be activated (taking into account the trajectory of the vehicle and the trajectories of objects in the vehicle's environment) and wait for the optimal activation time of the system. Here, this activation software is more specifically the subject of the present invention.

[0041] The software is activated once the vehicle 10 begins to move.

[0042] The software is implemented in a cyclical manner with fixed time increments.

[0043] The software includes preliminary steps for acquiring data related to the vehicle 10 and its environment, followed by nine main steps. These successive steps can then be described one by one.

[0044] In the preliminary step, computer C10 receives at least one image acquired by the front-facing camera of vehicle 10. The computer also receives data from a remote detector. These images and data are then fused.

[0045] At this stage, computer C10 thus possesses images of the road ahead of vehicle 10, as well as fused data, which in particular characterize each detected object located in the environment of vehicle 10. Here, this environment is considered to be the area surrounding the vehicle, in which the vehicle's sensors are designed to acquire data.

[0046] exist Figure 1 In the example, this vehicle is in the central central lane V. C Driving on the central lane, there are two additional driving lanes on either side of this central lane. R V L .

[0047] Then, computer C10 attempts to determine these driving lanes V C V R V L The position and shape of the boundary lines NL, L, R, NL.

[0048] Therefore, in this case, each of these lines is modeled using a polynomial equation. Here, the chosen polynomial is of order 3, and thus can be written as: [Math. 1]

[0049] In this equation: - The term yLine represents the lateral coordinate of the lane boundary line under consideration. - The term x represents the longitudinal coordinate of the boundary line, and - Terms a, b, c, and d are the coefficients of a polynomial, which are determined based on the shape of the boundary lines as seen by the vehicle's front-facing camera (or obtained by the computer C10 from a navigation system that includes a detailed map of the location in which the vehicle is moving).

[0050] In fact, these items are provided through data fusion. When visibility conditions are good, they can model the shape of lane boundary lines up to a distance of about one hundred meters.

[0051] At this stage, it should be noted that in the remainder of this disclosure, the term “longitudinal” will correspond to the vector component along the horizontal axis in the frame of reference under consideration, and the term “lateral” will correspond to the vector component along the vertical axis in the frame of reference under consideration (the frame of reference under consideration is always orthogonal).

[0052] Equation [Math. 1] here is attached to vehicle 10 and Figure 1 The reference frame (X) shown in the figure EGO YEGO The reference frame is oriented such that its horizontal axis extends along the longitudinal axis of the vehicle 10. It is centered on the front radar of the vehicle 10.

[0053] As a variation, other simpler or more complex methods can be used to model the geometry of lane boundary lines.

[0054] Once the coefficients a, b, c, and d for each lane boundary line have been determined, the computer C10 can implement these nine steps of the method, thereby sensing the degree of danger posed by detected objects to the vehicle, so as to trigger the AES obstacle avoidance system if necessary.

[0055] The first step involves determining the distance between the vehicle and the object under consideration (one of cars C1 or C2).

[0056] The distance calculated here is not Euclidean distance. In fact, the shape of the road needs to be taken into account in order to determine the distance that the vehicle 10 and the object must travel before colliding with each other.

[0057] Therefore, here, computer C10 calculates the arc distance L. AB .

[0058] Therefore, as detailed in document FR3077547, a computer can use the following equation: [Math. 2]

[0059] in: - L AB It is the arc distance between two points A and B (corresponding to the positions of the vehicle and the object under consideration). - x A This refers to the vehicle's longitudinal position (at radar level), and - x B In the reference frame (X) EGO Y EGO The longitudinal position of the object considered in ().

[0060] The second step involves determining the position of each detected object relative to the road lane, taking into account the equations of each lane boundary line and the fused data.

[0061] Computer C10 knows that the feature points (hereinafter referred to as "anchor points") of each detected object are located in a reference frame (X) attached to the vehicle. EGO Y EGOThe coordinates in the diagram are used to determine the feature point. This feature point is typically the center of the object as seen by a front-facing camera or a radar remote detector. Here, we will consider the feature point to be the center of the radiator grille of cars C1 and C2.

[0062] Two objects have already been detected (the two cars C1 and C2). Figure 1 In the example, the coordinates of the anchor points are labeled as (X_rel1, Y_rel1) and (X_rel2, Y_rel2), respectively.

[0063] Should Figure 1 In the reference frame (X) EGO Y EGO The following values ​​are also shown on the ordinate axis: - Y_road_NL_1, which is the value of the term yLine in the equation of the lane boundary line NL [Math. 1] at the x-coordinate point X_rel1. - Y_road_NL_2, which is the value of the term yLine in the equation of the lane boundary line NL [Math. 1] at the x-coordinate point X_rel2. - Y_road_L_1, which is the value of the term yLine in the equation of the lane boundary line L [Math. 1] at the x-coordinate point X_rel1. - Y_road_L_2, which is the value of the term yLine in the equation of the lane boundary line L [Math. 1] at the abscissa point X_rel2. - Y_road_R_1, which is the value of the term yLine in the equation of the lane boundary line R [Math. 1] at the x-coordinate point X_rel1. - Y_road_R_2, which is the value of the term yLine in the equation of the lane boundary line R [Math. 1] at the x-coordinate point X_rel2. - Y_road_NR_1, which is the value of the term yLine in the equation [Math. 1] of the lane boundary line NR at the x-coordinate point X_rel1. - Y_road_NR_2, which is the value of the term yLine in the equation of the lane boundary line NR [Math. 1] at the x-coordinate point X_rel2.

[0064] Then, by comparing these values ​​with the lateral coordinates Y_rel1 and Y_rel2 of cars C1 and C2, the lane in which each of the two cars is located can be determined.

[0065] For example, the lateral coordinate Y_rel1 of car C1 is between the values ​​Y_road_R_1 and Y_road_L_1, which means that the car is located between the lane boundary lines L and R.

[0066] At this stage, computer C10 can therefore determine the lane V where each detected object is located. L V C V R .

[0067] The purpose of the third step is to determine the parameters that characterize the kinematics of each object relative to the lane boundary line.

[0068] In the rest of the description of this step, for the sake of clarity of this disclosure, only a single object (car C1) will be considered.

[0069] This step includes a first sub-step in which the computer C10 determines the position of the object relative to one of the lane boundary lines. The lane boundary line under consideration is preferably the lane boundary line that separates the central lane from the driving lane where the object is located.

[0070] As a variation, the lane boundary line under consideration can be another boundary line, such as the lane edge line (see...). Figure 2 and Figure 3 This is especially true when no boundary line is detected between the lane of the object and the lane of the vehicle 10.

[0071] The idea is to discretize the lane boundary line intervals into a finite number of N points, and then select the point closest to the object under consideration. This operation is performed several times by re-discretizing the lane boundary line within each reduced interval located on either side of the selected point, in order to eventually find a good estimate of the point on the driving lane that is closest to the object under consideration.

[0072] In fact, such as Figure 2 As shown, the computer begins by discretizing the lane boundary line R into N coordinate points (X, X, Y) in the reference frame of the vehicle 10. i Y i These points are regularly distributed along the boundary line (in fact, along the X-axis). EGO (The interval between two consecutive points is always the same). The first point of these points is level with the vehicle (its x-coordinate is zero) or located at a first predetermined distance from the vehicle, and the last point of these points is located at a second predetermined distance from the vehicle.

[0073] Then, knowing the anchor point of car C1, which is represented here as (X) relGiven the coordinates of Yrel, the computer can use the following equation to derive the Euclidean distance between each discretized point of the lane boundary line R and the anchor point of car C1. Distance : [Math. 3]

[0074] Euclidean distance Bird Distance The shortest discretized point is the point closest to car C1. Therefore, this coordinate point (X...) is chosen. S Y S ).

[0075] Then, as Figure 3 As shown, this discretization operation is repeated with a finer degree of discretization at smaller intervals. The boundaries of the intervals are preferably defined by coordinate points (X, Y, X). s-1 Y s-1 ) and (X s+1 Y s+1 The number of discretization points is preferably still equal to n. Then, the new operation can select a new coordinate point (X). s Y s ).

[0076] After a certain number of iterations (e.g., 10 times) or when the interval between two discretized points is small enough (e.g., less than 10 cm), the computer stops repeating these iterations.

[0077] The last point selected is called the "projection point F". It is considered a good approximation of the point on the boundary line that is closest to car C1.

[0078] The x-coordinate of this point is X s The value is called Distance Xproj .

[0079] The Euclidean distance between the projection point F and the car C1 (Bird) Distance The value is called Dist Target2Lane .

[0080] The second sub-step includes computer C10 determining the speed of vehicle 10 in a reference frame attached to the road and level with the vehicle, and the speed of vehicle C1 in a reference frame attached to the road and level with vehicle C1.

[0081] In this sub-step, it will be assumed that, starting from projection point F, the road extends along the tangent at that point. Therefore, starting from car C1, the road is considered a straight line.

[0082] To better understand these calculations, Figure 4Four reference frames used in the remainder of this disclosure are shown.

[0083] The first frame of reference is the existing frame of reference attached to this vehicle (X). EGO Y EGO ).

[0084] It should be noted that this reference frame moves simultaneously with the vehicle 10. Therefore, an absolute reference frame (X) is also shown. abs Y abs This absolute reference frame is consistent with the first reference frame during measurement but is considered fixed.

[0085] Another frame of reference is represented as (X) lineEGO Y lineEGO The reference frame is attached to the lane boundary line R, and is oriented so that its abscissa is tangent to the boundary line, and is centered on the vehicle's radar (whose abscissa is zero in the second reference frame).

[0086] Another frame of reference is represented as (X) obj Y obj The reference frame is attached to vehicle C1 and is oriented so that its horizontal coordinate is aligned with the direction of travel of vehicle C1, and it is centered on the anchor point of vehicle C1.

[0087] The last frame of reference is represented as (X) LineObj Y LineObj The reference frame is attached to the lane boundary line R, and is oriented so that its horizontal coordinate is tangent to the boundary line, and is centered on the anchor point of car C1.

[0088] Figure 5 The angles separating these reference frames are shown: - Angle lineEGO / EGO From the reference frame (X) EGO Y EGO Transform to reference frame (X) lineEGO Y lineEGO ), - Angle lineObj / EGO From the reference frame (X) EGO Y EGO Transform to reference frame (X) LineObj Y LineObj ), - Angle Obj / EGO From the reference frame (X) EGO Y EGO Transform to reference frame (X) Obj Y Obj ), - Angle Obj / LineObjFrom the reference frame (X) LineObj Y LineObj Transform to reference frame (X) Obj Y Obj ).

[0089] Angle LineX / EGO More generally, it will be used to refer to the reference frame (X). EGO Y EGO The x-coordinate of the lane boundary line R is at the x-coordinate point X (with reference frame X). EGO Y EGO () indicates the angle separated by the tangent at ().

[0090] Therefore, it can be written as: [Math. 4]

[0091] in: [Math. 5]

[0092] Therefore, when x = 0, it can be written as: [Math. 6]

[0093] At the x-axis point x = Distance Xproj When, it can be written as: [Math. 7]

[0094] Computers can use the following formula to calculate in reference frame (X) lineEGO Y lineEGO The longitudinal component Vx of the velocity of vehicle 10 in this context EGO / LineEGO and the transverse component Vy EGO / LineEGO : [Math. 8]

[0095] [Math. 9]

[0096] In these formulas: - V EGO / abs In the absolute reference frame (X) abs Y abs The speed of the vehicle 10 is measured, for example, by a sensor located at the axle of the vehicle; - Angle VEgo / Ego This is the vehicle 10 relative to the reference frame (X).EGO Y EGO The angle of the velocity vector on the x-axis of the vector is given. Here, we assume that this angle is zero.

[0097] The computer can also calculate the speed V of car C1 in an absolute frame of reference. Obj / abs The longitudinal component Vx Obj / abs and the transverse component Vy Obj / abs Therefore, the computer uses the following formula: [Math. 10]

[0098] [Math. 11]

[0099] In these formulas: - Vx EGO / abs and Vy EGO / abs This vehicle is along the absolute reference frame (X). abs Y abs The velocity components of the x and y coordinates of the graph, and - Vx Obj / EGO and Vy Obj / EGO It is the reference frame (X) of car C1 relative to the vehicle itself. EGO Y EGO The velocity components of the x and y coordinates of ().

[0100] Therefore, it can be written as: [Math. 12]

[0101] As shown in the following two equations, it can be based on the angle Angle LineObj / Ego and Angle LineEgo / Ego To determine the component Vx of the relative velocity of car C1 with respect to the boundary line at projection point F "along the lane boundary line R". Obj / lineObj Vy Obj / lineObj This allows for better representation of information.

[0102] [Math. 13]

[0103] [Math. 14]

[0104] In these two equations, Angle VObj / Obj It is the angle of the vehicle's velocity vector in a reference frame attached to vehicle C1, and Angle Obj / Ego It is a frame of reference (X)EGO Y EGO The heading angle of the car in the figure.

[0105] In fact, this assumes that the object's velocity vector is collinear with its heading angle, therefore the angle Angle VObj / Obj It is zero.

[0106] A similar process was applied to determine the relative acceleration "along the lane boundary line R" between the vehicle and the lane boundary line at the zero abscissa point, and between the vehicle C1 and the lane boundary line at the projection point F.

[0107] Therefore, a computer can use the following formula to calculate the value in the reference frame (X). LineEGO Y LineEGO The longitudinal component Ax of the acceleration of vehicle 10 in the middle) EGO / LineEGO and the horizontal component Ay EGO / LineEGO : [Math. 15]

[0108] [Math. 16]

[0109] Computers can also use the following formula to calculate in reference frame (X) LineObj Y LineObj The longitudinal component Ax of the acceleration of car C1 in ( ) Obj / LineObj and the horizontal component Ay Obj / LineObj : [Math. 17]

[0110] [Math. 18]

[0111] In these formulas: - A EGO / abs It is the absolute acceleration of this vehicle 10 in the absolute frame of reference; - A Obj / abs It is the absolute acceleration of car C1 in an absolute frame of reference.

[0112] The calculated velocity and acceleration can then be combined to obtain the longitudinal components of the relative velocity and relative acceleration of the vehicle and car C1 relative to the road, VRelRoute, using the four equations defined below. Longi ,ARelRoute Longi and lateral component VRelRoute Lat ,ARelRouteLat .

[0113] In fact, the longitudinal component of the relative velocity between this vehicle and car C1 is considered to be VRelRoute Longi This is equivalent to using a reference frame (X) attached to the driving lane at the same level as the vehicle. LineEGO Y LineEGO The longitudinal component of the vehicle's speed is represented by ( ) and on the other hand, by a reference frame (X) attached to the lane at the same level as the vehicle C1. LineObj Y LineObj The deviation between the longitudinal components of the speed of car C1 is represented by ).

[0114] Similarly, the lateral component of the relative velocity between this vehicle and car C1, VRelRoute, will be considered... Lat This is equivalent to using a reference frame (X) attached to the driving lane at the same level as the vehicle. LineEGO Y LineEGO The lateral component of the vehicle's speed is represented by ( ) and on the other hand, by a reference frame (X) attached to the lane at the same level as the vehicle C1. LineObj Y LineObj The deviation between the lateral components of the speed of car C1 is represented by ).

[0115] Therefore, it can be written as: [Math. 19]

[0116] [Math. 20]

[0117] The components of acceleration can be calculated in a similar way: [Math. 21]

[0118] [Math. 22]

[0119] As will become apparent in detail in the remainder of this disclosure, the use of relative speed can provide an indication of collision risk that is difficult to obtain by other means.

[0120] At this stage, it can be recalled that computer C10 knows the distance Dist between the lane boundary line (at projection point F) and the anchor point of car C1. Target2Lane The value of .

[0121] In the third sub-step, computer C10 will determine the projection point F and the point P of the car C1 closest to the lane boundary line R. proxDistance between Lane DY (See Figure 6 ).

[0122] Here, the computer calculates the distance using the following equation.

[0123] [Math. 23]

[0124] In this equation, the term Width corresponds to the width of the car C1.

[0125] Then, the calculations can determine the collision time TTC with the object under consideration (car C1) in the fourth sub-step, that is, the time required for the vehicle to collide with the car while both the vehicle and the car C1 maintain their speeds.

[0126] In fact, at this stage, according to equation [Math. 2], the computer knows the length L of the arc separating vehicle 10 from car C1. AB According to equation [Math. 19], the computer also knows the longitudinal component VRelRoute of the relative velocity between vehicle 10 and car C1, relative to the shape of the road. Longi Finally, according to equation [Math. 21], the computer knows the longitudinal component of the acceleration, ARelRoute. Longi .

[0127] Using these longitudinal components, a good approximation of the time to collision (TTC) can be obtained when the road is curved and the vehicle trajectories are not parallel.

[0128] Here, the computer C10 then determines the collision time TTC sought using the following equation: [Math. 24]

[0129] It should be noted that two validity conditions for this equation must be met beforehand. These conditions are as follows.

[0130] [Math. 25] and

[0131] Conversely, if the longitudinal component of relative acceleration ARelRoute Longi If it is zero, then it can be written as: [Math. 26]

[0132] As a variant, the collision time TTC can be calculated in another way, for example, by assuming that the relative velocity and / or relative acceleration are constant.

[0133] In summary, at this stage, the computer, by utilizing fused data, possesses various parameters that characterize the diverse objects in its environment and also represent potential obstacles along its trajectory. For each object, the computer specifically possesses: - Collision time TTC (Equation [Math. 24]). - The object's position on the road (determined in step 2), and - Confirm the existence of the object (provided in the data fusion step).

[0134] Then, in the fourth step, the computer C10 will perform a first filtering on the various detected objects based on the parameters it has, so as to retain only the objects that are relevant to the implementation of the AES function (i.e., objects that form potential obstacles).

[0135] Therefore, the filtering operation includes considering relevant objects (hereinafter referred to as "targets") as objects that were identified during data fusion, whose locations are potentially hazardous (in this example, this is equivalent to checking that these objects are located in one of the driving lanes), and whose time to collision (TTC) is less than a predetermined threshold.

[0136] If multiple targets are detected in the same lane, only a limited number of targets (e.g., 4) may be considered, specifically the target that is closest to vehicle 10.

[0137] The fifth step involves the computer C10 identifying the lateral trajectory deviation (or overlap) required to avoid each target or group of targets to the right and left while avoiding other objects present on the road.

[0138] This step is implemented in five sub-steps.

[0139] Prior to the first sub-step, computer C10 identifies the target using a reference specific to each target.

[0140] Figure 7 An example is shown where four targets are located in the environment of this vehicle 10 and are ahead of this vehicle.

[0141] For example, each target is identified by the computer using a reference symbol written here in the form Cn, where n is a natural number equal to 1, 2, 3, or 4.

[0142] The number n of the target Cn is given here randomly.

[0143] The first sub-step involves independently considering each target Cn and calculating the deviation ovLn required to avoid the target Cn to the right and the deviation ovRn required to avoid the target Cn to the left.

[0144] Figure 7 The deviations ovL1 and ovR1 to avoid the target C1 have been shown.

[0145] These right and left deviations ovLn and ovRn are determined by taking into account the trajectory of the vehicle 10 and the fused data. In fact, data fusion provides kinematic information about the target relative to the vehicle 10, which, along with the vehicle's trajectory, is used to calculate these deviations. Therefore, these deviations are dynamically calculated based on the dynamics of the vehicle 10, the target's dynamics, and the shape of the lane.

[0146] Figure 7 The example corresponds to the case where the vehicle 10 and the target are moving in a straight line.

[0147] It was observed that if a deviation must be made to avoid the target to the right, the value of the deviation ovRn is greater than 0. Otherwise, the value of the deviation is less than or equal to 0. More specifically, if the vehicle 10 does not need to change its trajectory to avoid the target and passes as close to the target as possible, the deviation ovRn is equal to 0. In contrast, if the vehicle 10 does not need to change its trajectory to avoid the target, but should change its trajectory if it wishes to pass as close to the target as possible, the deviation ovRn is strictly less than 0.

[0148] Similarly, if a deviation must be taken to avoid the target to the left, the value of the deviation ovLn is greater than 0. Otherwise, the value of the deviation is less than or equal to 0.

[0149] If the road is not straight, it is suggested that the following information be considered to refine the calculation of these deviations: - The lateral component of the relative velocity between this vehicle and the target under consideration along the road traveled (VRelRoute) Lat (Equation [Math. 20]) - Angle Obj / LineObj ,as well as - Collision Time TTC.

[0150] Taking into account the shape of the driving lane, the first of these pieces of information can be the actual lateral speed between the vehicle 10 and the target.

[0151] To better understand the benefits of this parameter, consider the following example: In this example, the vehicle and the target are traveling in opposite directions on two different lanes, correctly following the curvature of these lanes. Therefore, it should be understood that, theoretically, the risk of an accident is zero. In this example, the lateral component of the relative velocity of the vehicle relative to the target is VRelRoute. Lat The value will be zero, which means that the calculated deviation will be less than or equal to zero, which correctly expresses the idea that the theoretical collision risk is zero.

[0152] In other words, the lateral component VRelRoute Lat The collision time TTC allows for a trade-off between the effects of relative lateral velocity and relative longitudinal velocity on the calculation of left deviation ovLn and right deviation ovRn.

[0153] Similarly, by weighting the length and width of the vehicle, information Angle Obj / LineObj This allows for the determination of more precise values ​​for the impact surface of the target under consideration.

[0154] Here, these right biases E are calculated. right and left deviation E left The method is similar to that described in document FR1907351, except that these calculations can take into account the three types of information mentioned above.

[0155] Therefore, based on information Angle Obj / LineObj The target's half-width is calculated. Then, this half-width is used to calculate a preliminary value for each deviation, which does not consider the safety radius for avoiding the target. Lateral component VRelRoute Lat Multiply it by the collision time TTC, and then add it to the initial deviation to obtain the desired deviation.

[0156] In other words, if we consider reference FR1907351, in order to calculate these deviations, it would be necessary to use the horizontal component VRelRoute in the calculation of the horizontal coordinate dVy. Lat The product of the collision time TTC. The coordinates Y used in this literature. a (Here it corresponds to the coordinate Y) rel In itself, it will be considered equal to the coordinates generated by data fusion and equal to the length and angle of the target. Obj / LineObj The sum of terms of the product of the cosines.

[0157] Therefore, it can be written as: [Math. 27]

[0158] In this equation, er is a term used to compensate for lateral measurement errors, and Long is the length of the target.

[0159] [Math. 28]

[0160] If data fusion is used, the item also takes into account the error resulting from the data fusion. This item is predefined and stored in the computer's memory.

[0161] The second sub-step will involve sorting the targets in order of their position on the road (more precisely, depending on the deviation of the target from one of the edges of the road).

[0162] Here, the operation is based on the calculated left deviation ovL. n Execute in descending order. As a variation, another sorting method can, of course, be applied.

[0163] The advantage of the method used here is that, based on the lateral relative velocity of the vehicle 10 and the target, and also based on the calculated time to collision (TTC), the deviation calculation takes into account the relative dynamics of the scene, which integrates the concept of predicting the relative position of the vehicle and the target.

[0164] Here, as Figure 8 As shown, the target previously referred to as Cn is now called C. n .

[0165] from Figure 7 and Figure 8 It can be seen from this: - Target C1 becomes target C2. - Target C2 becomes target C4. - Target C3 becomes target C3. - Target C4 becomes target C1.

[0166] Similarly, the deviation previously referred to as ovLn and ovRn is now called ovL. n ,ovR n .

[0167] therefore: - Deviations ovL1 and ovR1 become ovL2 and ovR2. - Deviations ovL2 and ovR2 become ovL4 and ovR4. - Deviations ovL3 and ovR3 become ovL3 and ovR3. - Deviations ovL4 and ovR4 become ovL1 and ovR1.

[0168] This classification of targets allows them to be ordered from left to right relative to the vehicle 10.

[0169] In the following text, TTC n This will be used to represent the target C. n Calculated collision time.

[0170] By considering each target in the order determined therein, the computer C10 can then determine whether it is possible to pass through from the right or left side of that target.

[0171] Therefore, in the third sub-step, the computer solves the following equation (in the example, all n values ​​are in the range of 1 to 4).

[0172] [Math. 29]

[0173] [Math. 30]

[0174] In these equations, the parameter dSafe has a strictly positive value and corresponds to the desired safe distance to be formed around the target to avoid getting too close to it when passing. Its value may vary, for example, based on the speed of the target and the vehicle itself, or traffic conditions (weather, etc.). It is at least equal to the width of the vehicle.

[0175] Parameter Gap Left_n The following is referred to as the left clearance, which corresponds to the width that can be passed from the left side of the target, taking into account other targets.

[0176] Parameter Gap Right_n The right clearance, referred to below, corresponds to the width that can be passed through from the right side of a target, taking into account other targets.

[0177] It should be noted here that when n equals 1, it is unnecessary to calculate the left gap. Left_n Since it is known that it can pass through from the left side of target C1.

[0178] Similarly, when n equals 4, it is not necessary to calculate the right gap. Right_n Since it is known that it can pass through from the right side of target C4.

[0179] At this stage, it can be assumed that the left gap is greater than or equal to the left gap. Left_n When the value is greater than or equal to zero, this vehicle 10 can move from the considered target C. n Passing through on the left.

[0180] Similarly, it can be assumed that the right gap (Gap) is true if and only if... Right_n When the value is greater than or equal to zero, this vehicle 10 can move from the considered target C. nIt passes through on the right side.

[0181] exist Figure 8 In the example, we thus obtain the following: - Gap Right_1 > or = 0, - Gap Left_2 > or = 0, - Gap Right_2 < 0, - Gap Left_3 < 0, - Gap Right_3 < 0, - Gap Left_4 < 0.

[0182] Figure 9 This situation is illustrated in the image.

[0183] It can be observed that the value Gap Right_2 Gap Left_3 Gap Right_3 Gap Left_4 All are strictly less than zero, which means that theoretically, it is impossible to pass through the target C2, C3 and C4.

[0184] It can also be observed that the value Gap Right_4 and Gap Left_2 A value greater than or equal to 0 means that it is possible to pass through any side of this group containing targets C2, C3, and C4. This is because the term Gap... Right_1 It is greater than or equal to 0, so it can also pass through the right side of target C1.

[0185] At this stage, when it is theoretically impossible for targets to pass each other, the computer can therefore group the targets.

[0186] Therefore, at least one proximity criterion between targets is used.

[0187] The first proximity criterion involves the lateral distance between targets.

[0188] Therefore, in order to form a group, the computer C10 identifies the target C under consideration. n With neighboring target C n+1 (in consecutive order) gaps Right_n or Gap Left_n+1 Is it strictly less than 0? If so, the two targets are grouped together.

[0189] Therefore, this is in Figure 9 The example gives a group with three objectives, referred to below as the initial group G1.

[0190] This concludes the process of grouping objectives into sets with multiple objectives.

[0191] However, this document specifies that a second proximity criterion should be considered to form these groups. This second proximity criterion involves the longitudinal distance between targets.

[0192] Specifically, the idea is that a group is formed only when the vehicle 10 cannot be accommodated among some targets in the initial group G1.

[0193] For this purpose, computer C10 is programmed here to divide one or more preliminary groups G1 when they include targets that are longitudinally separated from each other.

[0194] Here, by comparing each target C n Collision Time TTC n This sorting is performed on the targets within each group. As a variant, this sorting can be performed using other parameters (specifically, the arc distance L). AB This is executed using the TTC parameter. n The advantage is that it takes into account objective C. n The relative longitudinal velocity and acceleration.

[0195] The C10 computer works in the same way for each group with multiple objectives.

[0196] The computer is based on the time-of-collision (TTC) of the group of targets. n Sort the group of targets, for example, in ascending order.

[0197] exist Figure 10 In the example, the targets are ordered as follows: C4, C2, C3.

[0198] Then, the computer calculates the deviation between each pair of consecutive targets (in a determined order).

[0199] Therefore, in Figure 10 In the example, the computer calculates: - The deviation Δ between the collision time TTC4 associated with target C4 and the collision time TTC2 associated with target C2 4-2 , - The deviation Δ between the collision time TTC2 associated with target C2 and the collision time TTC3 associated with target C3. 2-3 .

[0200] Then, the computer will take these deviations Δ 4-2 Δ 2-3 Each of them is compared with the threshold Sx.

[0201] This threshold may remain constant. However, preferably, the threshold will be selected based at least on the longitudinal speed of the vehicle 10.

[0202] If the deviation between the collision times associated with two consecutive targets is greater than the threshold, the computer will initially divide the group into two groups.

[0203] exist Figure 10 In the example, at the end of the operation, targets C4 and C2 form the first group G2, while target C3 is removed.

[0204] Therefore, at this stage, the computer will treat each group as a separate target in the same way. Thus, the computer will correlate the left deviation ovL. i , right deviation ovR i and collision time TTC i (i is the index of the group under consideration).

[0205] Then calculate these parameters as follows.

[0206] For easier understanding, please refer to Figure 11 For example, the group under consideration, G3, includes three objectives, C1, C2, and C3.

[0207] The collision time TTC for this group i The collision time TTC will be selected to be equal to that of the targets C1, C2, and C3 in that group. n Minimum Collision Time (TTC) n .

[0208] The left deviation ovL of this group i The left deviation ovL equal to the target set will be selected. n The maximum left deviation ovL in n .

[0209] The right deviation ovR of this group i The right deviation ovR equal to the target set will be selected. n The maximum right deviation ovR in n .

[0210] In the remainder of this disclosure, for simplicity, the general term “target” is used to refer to both a group of multiple targets and a single target that does not form part of any group.

[0211] The sixth step involves the computer performing a second filtering process on the targets to distinguish critical targets from other targets.

[0212] If activating the AES function is required to avoid a target, that target is called a critical target. The most critical target (target MCT) is the target that requires the earliest activation of the AES function.

[0213] If the location of a target should be taken into account when determining the avoidance trajectory to be followed, then the target is designated as a "medium-risk" target. Therefore, medium-risk targets may inhibit the activation of AES functionality.

[0214] The idea is to consider each detected target independently, in turn (and therefore, regardless of its environment).

[0215] First, computer C10 considers all targets in the driving lane of vehicle 10 to be critical targets.

[0216] For targets located in lanes adjacent to the 10 lanes in which this vehicle is traveling, the computer checks whether they meet additional criteria.

[0217] Here, these standards relate to the following parameters: - Arc distance L AB , - The distance between the projection point F and the point closest to the lane boundary line. DY (Equation [Math. 23]), and - The lateral component Vy of the target's velocity relative to the lane boundary line at the projection point F Obj / lineObj (Equation [Math.14]).

[0218] Figure 12 Three targets, C4, C5, and C6, are shown in two lanes adjacent to the lane in which this vehicle is traveling.

[0219] To determine whether each objective is a critical objective, the computer checks whether the following two conditions are met (i) and (ii).

[0220] To check the first condition i), the computer begins by calculating the distance between the target and the lane boundary line under consideration. DY (Equation [Math. 23]). This can determine whether the target is relatively close to or relatively far from the lane boundary line under consideration.

[0221] The C10 computer derives the minimum lateral velocity threshold from this, denoted as Vy. thresholdMin .

[0222] Then, if component Vy Obj / lineObj Exceeding the minimum threshold Vy thresholdMin If the target is critical, then the first condition (the target is considered a critical target) is met. Otherwise, the target is simply considered a medium-risk target.

[0223] It should be noted that the threshold used is therefore a variable that depends on the distance between the target and the lane boundary line under consideration, thus allowing us to consider the fact that the shorter the distance, the greater the risk of collision.

[0224] The threshold used can also depend on how the target is represented in previous time increments (critical or non-critical state). Specifically, the idea is that the representation of the target does not change with each time increment because there is noise in the measurements of the data to be fused. For this reason, the threshold used to transition the target from a non-critical state to a critical state is higher than the threshold used to transition the target from a critical state to a non-critical state (similar to a hysteresis function).

[0225] The second condition ii) can disregard targets arising from potential perceptual errors, as well as targets with excessively large or abnormal lateral velocities. For this purpose, the computer will use the lateral component Vy... Obj / lineObj The absolute value of Vy and the predetermined maximum threshold thresholdMax A comparison is made. If the component is greater than the maximum threshold, the second condition is not met and the target is considered a medium-risk target. The maximum threshold should have certain limitations to avoid considering false detections that are sometimes unreasonable in value.

[0226] The maximum threshold is preferably greater than 2 m / s, and preferably equal to 3 m / s.

[0227] Objectives that do not meet one of conditions i) and ii) and / or the other condition are considered medium-risk objectives. Other objectives are considered critical objectives.

[0228] Classifying perceived targets into critical and medium-risk targets reduces computation time. Furthermore, this target classification simplifies the decision of whether to activate the AES system, and the decision made can be proven reasonable for a human driver.

[0229] exist Figure 12 In this context, it can be assumed that only target C4 satisfies both conditions i) and ii), because the other two targets do not satisfy condition i).

[0230] The seventh step involves the computer C10 determining a critical time Tcrit for each critical target. This critical time combines two different but related pieces of information to ensure obstacle avoidance while minimizing the intrusiveness of the AES function to the driver.

[0231] The benefit of calculating this parameter is especially in identifying which objective among critical objectives is the most critical objective MCT.

[0232] Before describing how this parameter is obtained, it can be recalled that the AES avoidance system can be operated fully automatically (in which case the auxiliary steering actuator autonomously follows the avoidance trajectory) or semi-automatically (in which case the driver manually initiates the avoidance maneuver, and once the driver has triggered the avoidance maneuver, the auxiliary steering actuator is controlled to assist the driver in following the avoidance trajectory). In the remainder of the manual, the terms autonomous mode and manual mode are used to refer to these two methods respectively.

[0233] Avoidance maneuvers in manual mode may be less effective than in autonomous mode. Therefore, the avoidance trajectory calculated by the AES system will differ between manual and autonomous modes. It should be noted that the calculation of this avoidance trajectory (in the form of a spiral) is not the subject of this invention. To briefly recap, the shape of this trajectory is calculated based on the vehicle's dynamic performance and the driver's ability (in manual mode).

[0234] Regardless of the circumstances, taking into account the calculated deviation ovL n ,ovR n It can construct four avoidance trajectories (four spiral lines) to avoid critical targets to the right and left in both manual and autonomous modes.

[0235] The curvature of these spirals depends on the vehicle's maximum dynamic capability. Therefore, the shape of these spirals is read from a database based on the speed of the vehicle 10.

[0236] Taking into account these deviations and the determined avoidance trajectory, four maneuvering times (TTS) can be derived (see...). Figure 7 These four manipulation times correspond to the time required to perform an avoidance maneuver to the left or right in each mode.

[0237] The method of obtaining these handling time TTS will not be described here, as it will depend on the dynamic characteristics of the vehicle 10 and the driver's performance. In practice, these handling time TTS can be read from a database established using test batteries.

[0238] Knowing the collision time (TTC) of each critical target n After manipulating the time series time (TTS), the computer can derive the critical time (Tcrit) from it using the following formula: [Math. 31]

[0239] Therefore, the critical time becomes zero at the last moment. At that moment, the AES function can still be activated, and collision with the target can be avoided by avoiding the critical target on the considered side in the considered manual or automatic mode.

[0240] Therefore, the critical time Tcrit carries three basic pieces of information about the target under consideration: on which side to avoid the target, the performance of the system (and the driver), and the time of collision with the target TTC. n .

[0241] Then, the eighth step will involve classifying critical targets using the calculated critical time Tcrit and finding the most critical target MCT.

[0242] If the driver is in control in manual mode (for example, because autonomous mode cannot be activated), then the computer selects the critical target with the shortest critical time when avoiding a right-hand maneuver and the critical target with the shortest critical time when avoiding a left-hand maneuver.

[0243] Then, once the computer C10 detects that the driver has initiated evasive maneuvers, the computer can decide to activate the AES system to assist the driver in maneuvering based on two selected critical times.

[0244] To this end, the computer determines which side (right or left) the driver turns the steering wheel when initiating their evasive maneuver, and then selects the critical time from two chosen critical times that corresponds to the side on which the driver has begun their evasive maneuver.

[0245] The most critical target (MCT) is the target corresponding to the selected critical time.

[0246] In autonomous mode, the computer behaves differently. In fact, it needs to determine which side of one or more obstacles the vehicle 10 should avoid.

[0247] The next step was to determine the optimal avoidance side for each critical target, and then determine the most critical target MCT.

[0248] In practice, the computer selects the side with the highest critical time Tcrit for each critical target.

[0249] The computer then selects the smallest critical time from the thus chosen critical times.

[0250] Then, the computer considers the most critical target MCT to be the target whose critical time Tcrit has already been selected.

[0251] The ninth step involves the computer C10 triggering the AES function when necessary and at the optimal moment.

[0252] In this scenario, under autonomous mode, the AES function is triggered once the selected critical time Tcrit drops below a predetermined threshold (e.g., equal to 0 seconds).

[0253] In contrast, the AES function is triggered differently in manual mode.

[0254] The idea is to trigger the AES function in a way that suits the situation and coordinate with the driver to avoid obstacles.

[0255] In fact, if the driver initiates steering too early, the action will be considered non-emergency because the situation does not require this type of action from the driver at that moment. In this case, the AES system will not be activated.

[0256] Similarly, if a collision is imminent and autonomous mode cannot be activated beforehand, it can be considered too late, and the AES system cannot provide any assistance to the driver. At this point, responsibility is transferred to another safety system to minimize the impact.

[0257] Therefore, in manual mode, a time interval needs to be determined during which the AES system will be activated if the driver is detected to have performed an evasive maneuver.

[0258] This time interval will be defined by two boundaries.

[0259] The first boundary will correspond to a critical time Tcrit that is strictly greater than 0, from which the driver will receive assistance if they initiate evasive maneuvers.

[0260] The second boundary corresponds to a critical time Tcrit that is less than or equal to 0 and preferably strictly less than 0, from which it is considered too late to trigger the AES system.

[0261] Therefore, in order to activate the AES function, the computer determines whether the selected critical time Tcrit is between these two boundaries, and the AES function will only be activated if it is between these two boundaries.

[0262] The present invention is by no means limited to the embodiments described and illustrated, but those skilled in the art will know how to provide any variations of the embodiments according to the invention.

[0263] A single proximity criterion that is neither horizontal nor vertical can be considered, but only the deviation between targets should be taken into account.

[0264] As another variation, the first step could be to group the vehicles according to their longitudinal deviation, and then divide each group as needed according to the lateral deviation between the vehicles in that group.

Claims

1. A method for a motor vehicle (10) to avoid objects (C1, C2, C3, C4), the method comprising the following steps: - Detect objects (C1, C2, C3, C4) in the environment of the motor vehicle (10). - Acquisition steps are used to acquire data characterizing the position and / or dynamics of each detected object (C1, C2, C3, C4). The feature is that if multiple objects (C1, C2, C3, C4) have been detected, the following steps are specified: - Check whether at least one proximity criterion is met between at least two of the detected objects (C1, C2, C3, C4), and if so, - Combine the two objects (C1, C2, C3, C4) into a group (G1, G2, G3). - Calculation steps for calculating data characterizing the position and / or dynamics of the groups (G1, G2, G3), and - Based on the data characterizing the position and / or dynamics of the group (G1, G2, G3), activate the obstacle avoidance system (AES) and / or determine the avoidance trajectory (T1). The proximity criterion refers to the lateral distance between the two objects (C1, C2, C3, C4). In this acquisition step, one of the data is the lateral trajectory deviation (E) that the motor vehicle (10) must take to avoid each object (C1, C2, C3, C4). left E right ), and wherein the proximity criterion includes examining the lateral trajectory deviation (E) to be taken, on the one hand, to avoid the first object on the side oriented toward the second object. left ) and on the other hand, the lateral trajectory deviation (E) required to avoid a second object on the side oriented towards the first object. right The difference between (Gap) Left_n Gap Right_n Is it greater than or equal to a predetermined threshold (d)? safe ).

2. The avoidance method as described in claim 1, wherein, If at least three objects (C1, C2, C3, C4) have been detected, it is stipulated that these objects (C1, C2, C3, C4) be ordered in a continuous order from one edge of the road to the other, and wherein the proximity criterion is checked between each pair of consecutive objects in the continuous order.

3. The avoidance method as described in claim 1 or 2, wherein, The first reference frame (X) is defined based on the orientation of the road tangentially along the point where the motor vehicle (10) is aligned. LineEgo Y LineEgo The lateral velocity (Vy) of the motor vehicle (10) relative to the road it is traveling on. Ego / LineEgo ) and a second reference frame (X) oriented along the tangent of the road at the level of the objects (C1, C2). LineObj Y LineObj In the diagram, the lateral velocity (Vy) of the object (C1, C2) relative to the road is... Obj / LineObj The deviation between the two is used to calculate the relative lateral velocity (VRelRoute). Lat ), and among them, based on the relative lateral velocity (VRelRoute) Lat ) to determine each lateral trajectory deviation (E) left E right ).

4. The avoidance method as described in claim 1 or 2, wherein, The proximity criterion refers to the longitudinal distance between the two objects.

5. The avoidance method as described in claim 4, wherein, In this acquisition step, one of the data involves the remaining time (TTC) before the motor vehicle (10) collides with each object (C1, C2, C3, C4), and wherein, in order to check whether the proximity criterion is met, the deviation (Δ) between the collision times (TTC) with the two objects is checked. 2-3 Δ 4-2 Is it less than the threshold? 6. The avoidance method as described in claim 1 or 2, wherein, In this acquisition step, one of the data points is the lateral trajectory deviation (E) to be taken to avoid each object (C1, C2, C3, C4) on the same left or right side. left E right ), and wherein, in this calculation step, one of the data representing the group is selected as equal to the lateral trajectory deviations (E) to be taken to avoid each object (C1, C2, C3, C4) on the same side of the group (G1, G2, G3). left E right The maximum value in ).

7. The avoidance method as described in claim 1 or 2, wherein, In the acquisition step, one of the data representing each object is the remaining time (TTC) before the motor vehicle (101) collides with each object (C1, C2, C3, C4), and wherein, in the calculation step, one of the data representing the group is selected as the minimum of the remaining time (TTC) before the motor vehicle (101) collides with each object (C1, C2, C3, C4) in the group (G1, G2, G3).

8. A motor vehicle (10) comprising at least one steering wheel, wherein a steering system for each steering wheel is designed to be operated by an actuator controlled by a computer (C10), characterized in that, The computer (C10) is designed to implement the avoidance method as described in any one of claims 1 to 7.

Images