A method for controlling an automated vehicle to avoid an object by improving the clothoid curve path.

Abstract

A method for controlling a motor vehicle to avoid an object, the motor vehicle having at least two sensing sensors, the control method comprising the following steps: - determining data from the at least two sensors; - fusing the data from the at least two sensors to determine at least a steering wheel angle, a vehicle speed, and a vehicle heading; - planning an avoidance path for avoiding the object, the avoidance path taking the form of a clothoid curve; - refining the avoidance path used to avoid the object according to the steering wheel angle, the vehicle speed, and the vehicle heading and based on solving an optimization problem; and - controlling the vehicle to follow the improved path, wherein the path refinement step comprises a first series of sub-steps for enhancing the avoidance path according to the path length, the steering wheel rotation direction, and the final heading, and a second series of sub-steps for refining the enhanced trajectory according to the initial heading and the steering wheel rotation direction.

Description

[Technical Field] 【0001】 The technical field of the present invention is a route control system, and more specifically, a system based on such clothoid-shaped routes. [Background technology] 【0002】 Some autonomous vehicle path control systems generate reference clothoid paths for automatic avoidance systems or "automatic avoidance steering" systems (AES) to help drivers maintain a comfortable steering position while avoiding collisions with obstacles detected on the road. 【0003】 These clothoid-shaped reference paths take into account the system's controllability (limitations on steering angle magnitude / gradient) and certain comfort criteria (e.g., jerks, sequence of curves). However, despite its overall performance level, this rigid approach to path planning lacks flexibility because it imposes strict constraints on the desired path (i.e., the final steering angle is equal to zero, the integral of the steering angle profile is equal to zero) and ignores the system's non-zero initial conditions. This risks destabilizing the path tracking control system and the vehicle system. For example, if the absolute value of the vehicle's heading is sufficiently large when the AES is activated, the control system will force the heading to zero to prevent deviation from the reference clothoid path. As a result, the steering wheel will turn in the opposite direction to the natural avoidance tendency, a situation that is unpleasant and difficult for the driver to understand. Furthermore, if the driver acts contrary to this operation by turning the steering wheel in the logical avoidance direction, and the system applies saturation constraints, the latter becomes unstable, resulting in vibrations due to driver-induced oscillations ("driver-induced vibrations"). 【0004】 Therefore, the aforementioned problems need to be corrected. 【0005】 The following literature is known from prior art. 【0006】 French Patent Application Publication No. 2003457 describes a very restricted method that imposes strict constraints on the planned route (for example, the final steering angle is equal to 0, the integral of the steering angle profile is equal to 0, etc.). When the initial conditions are non-zero, behaviors that potentially make the driver unstable occur. For example, when there is a fairly significant positive initial orientation, the direction of the steering wheel at the start of the avoidance phase is directed in the wrong direction. This instability may be amplified when the driver interferes with the steering wheel during the avoidance phase. 【0007】 The document "Clothoid-based model predictive control for autonomous driving, P. Lima, J. Martensson, and M. Trincavelli, ECC, Linz (2015)" discloses a control method for making the vehicle center on the lane. Furthermore, this method does not consider various constraints essential for the stability of the system, such as controllability. Also, with this method, the calculated route may become discontinuous. 【0008】 The document of US Patent Application Publication No. 2008 / 0255728 describes the construction of a clothoid curve from known initial and final states by means of a geometric shape of an isosceles triangle containing a combination of clothoid shapes. The purpose of this document is not to minimize the time to collision TTC under dynamic constraints but to find an easy way to guide from one point to another point so as to be applied to the orientation problem. Furthermore, this application does not consider dynamic limitations and there is no procedure for optimizing the operation time for a certain situation. SUMMARY OF THE INVENTION 【0009】 The subject of the present invention is a method for controlling an automotive vehicle to avoid an object, the automotive vehicle comprising at least two sensing sensors, and the control method comprising the following steps: - a step of determining data from at least two sensors; - Merging data from at least two sensors to determine at least the steering angle, the speed of the vehicle, and the orientation of the vehicle; - Planning an object avoidance path in the shape of a clothoid; - Improving the object avoidance path as a function of the steering angle, the speed of the vehicle, and the orientation of the vehicle, and based on the solution of an optimization problem; - Controlling the vehicle to follow the improved path, including: The step of improving the path includes a first series of sub-steps for further improving the avoidance path as a function of the length of the path, the direction of rotation of the steering wheel, and the final orientation, and a second series of sub-steps for improving the improved path as a function of the initial orientation and the direction of rotation of the steering wheel. 【0010】 The optimization problem is a function of a decision vector including a reference longitudinal displacement, a reference lateral displacement, a reference vehicle orientation, a reference vehicle path curvature, the speed of change of the curvature of the path, and the distance traveled relative to the origin, and empirically determined setting parameters, and may be constraints on the initial longitudinal displacement, the initial lateral displacement, the final lateral displacement, the steering angle, the rotational speed of the steering wheel, and the final orientation value. Thus, this improvement makes it possible to correctly take into account the controllability imposed on the system by considering the constraints on the steering angle and the rotational speed of the steering wheel in the optimization of the path. By taking these into account, it becomes possible to find a more reasonable and acceptable path for the driver, and indirectly, a steering angle profile. 【0011】 The first set of substeps may include the following substeps: the initialization parameters of the optimization problem are determined as a function of the reference path and merged data, and the initialization parameters include reference longitudinal displacement, reference lateral displacement, reference vehicle orientation, curvature of the reference vehicle path, initial vehicle orientation, and initial steering angle; the optimization problem is solved, and a determination is made as to whether an optimal solution exists by determining whether the solution of the optimization problem corresponds to the minimum of a predetermined cost function and whether the solution of the optimization problem satisfies constraints on the final orientation of the vehicle; in such cases, the steering angle profile is determined from the moment the control method is initiated, and then a determination is made as to whether the sign of the steering angle associated with the first extremum of the steering angle profile has the same sign as the steering angle of the reference path, and whether, over the duration of the optimization path, the extremum with the opposite sign to the first extremum is also the maximum; in such cases, the solution of the optimization problem is considered an improved path. 【0012】 If it is determined that no optimal solution exists, the following steps can be taken: the path is extended for a predetermined duration, and then the method is restarted when the optimization problem is solved by taking the path extension into account, and the predetermined duration is increased each time it is determined that no optimal solution exists. 【0013】 If the handle angle profile is not met, the following steps are taken: a longitudinal offset is applied to the reference path, the optimization problem is solved again, and then the method is restarted when determining the handle angle profile, and it is possible to increase the longitudinal offset each time it is determined that the handle angle profile is not met. 【0014】 One constraint on the optimization problem may be that, in order to ensure smoothing of the improved path, the vertical displacement of a point on the desired path must be equal to the vertical displacement of a point on the reference path. 【0015】 One constraint in the optimization problem may be that, in order to stabilize the vehicle's path after avoidance, the final value of the vehicle's heading on the improved path must be within a limited range of values. 【0016】 One constraint on the optimization problem may be that the longitudinal displacement of points along the improved path must be smaller than the longitudinal displacement of the extreme line that is not exceeded during the correction. 【0017】 One constraint on the optimization problem may be that the final lateral displacement of the improved path must be greater than or equal to the final lateral displacement of the reference path. 【0018】 The second set of substeps may include the following: determining new initialization parameters as a function of the improved path and merged data; applying a predetermined lateral offset to the improved path to solve the optimization problem; determining whether an optimal solution exists by determining whether it corresponds to the minimum of a predetermined cost function; if so, determining the steering angle profile from the moment of the control method to determine whether the sign of the steering angle associated with the first extremum of the steering angle profile has the same sign as the sign of the steering angle of the reference path; and determining whether, over the duration of the optimized path, the extremum with the opposite sign to the first extremum is also the maximum; if so, the solution to the optimization problem may be considered the final improved path. Thus, constraints on the steering angle and the lateral position combined with the offset in the initial lateral position ensure that an optimal path can be found at the start of an avoidance operation without requiring the steering wheel to be turned in the wrong direction, thereby improving driver confidence and vehicle stability. 【0019】 If it is determined that no optimal solution exists, the following steps are taken: the initial orientation is reduced by a predetermined angular deviation, and then the method is restarted when solving the optimization problem, and as soon as no optimal solution exists, it is possible to continuously increase the angular deviation each time it occurs. 【0020】 If it is determined that the optimal solution does not satisfy the steering angle constraint, the following steps are taken: the longitudinal offset is applied to the improved path, the optimization problem is solved, and then the method is restarted when determining the steering angle profile, and it is possible to increase the longitudinal offset each time it is determined that the steering angle profile is not satisfied. 【0021】 The predetermined duration, angular deviation, and longitudinal offset can be continuously increased by predetermined limit values. In such cases, the improvement step is interrupted, and the vehicle control step follows the reference path. 【0022】 Furthermore, the subject of the present invention is a control system for an automated vehicle to avoid an object, the automated vehicle comprising at least one sensing sensor and at least one computing means configured to perform the control method set forth above. 【0023】 Other objects, features, and advantages of the present invention will become apparent from reading the following description, which is given merely as a non-limiting example, and from referring to the accompanying drawings. [Brief explanation of the drawing] 【0024】 [Figure 1] This figure shows the main steps of the control method of the present invention. [Figure 2] This figure shows the main substeps of improving the route according to the present invention. [Modes for carrying out the invention] 【0025】 The objective of the control system described below is to correct the problems of current technology by improving the clothoid generated by path planning such that the system's behavior during avoidance is understandable and acceptable to the driver. These reference paths are reconstructed by taking into account the initial conditions of the vehicle. While maintaining the clothoid shape, the reference paths are still subject to controllability constraints. 【0026】 Figure 1 shows the connection between the control method of the present invention and the current technology. 【0027】 The automated vehicle comprises at least one sensor and a merging means connected as input to at least one computing means configured to merge data determined by the sensor and to perform the steps of the control method described below. 【0028】 The control method according to the present invention includes steps 1 (acquiring data), 2 (merging the acquired data), 3 (determining an object avoidance path), 4 (calculating an improved path), and 5 (controlling the movement of the vehicle ("motion control"). Steps 3 (determining an object avoidance path) and 4 (calculating an improved path) are included in the AES-type automatic avoidance method 6. 【0029】 Calculation step 4 makes it possible to improve the path obtained from step 3, which determines the object avoidance path that constitutes the conventional path plan. 【0030】 Route improvement (or "route refinement") makes it possible to improve the route generated by conventional route planning as a function of data received from merging data, particularly the direction, steering angle, and vehicle speed at the moment of method initiation. 【0031】 The principle of path improvement is, above all, to find an approximate / linear model that allows for a simplified representation of the clothoid. This approximation / model will then be used to re-formulate the "improved path" problem as a quadratic optimization problem. 【0032】 The linear model of the clothoid will be explained here. 【0033】 The following formula defines a clothoid-shaped path. [Formula 1] TIFF0007872790000001.tif15170[Formula 2] TIFF0007872790000002.tif15170 【0034】 Two consecutive points i and i + 1 located on the same cycloid, and the distances s they have traveled from the starting point , s i+1 For these, their equations can be approximated as follows. [Formula 3] TIFF0007872790000003.tif23170[Formula 4] TIFF0007872790000004.tif24170[Formula 5][[ID=@17]]<00@0188> TIFF0007872790000005.tif13170[Formula 6] TIFF0007872790000006.tif10170 Here s i : The distance traveled from the starting point to point i along the path x(s i ): The vertical displacement of the vehicle at point i on the path with respect to the starting point y(s i ): The lateral displacement of the vehicle at point i on the path with respect to the starting point θ(s i ): The orientation of the vehicle at point i on the path κ(s i ): The curvature of the vehicle's path at point i<00@0195>c i : The speed of the variation of the curvature of the path at point i 【0035】 The above equations ([Formula 3] to [Formula 6]) are used as the model of the present invention representing the desired path of the vehicle. X i =(x i , y i , θ i , κ i ) In the case of, U i =(c i , s i ) is determined, and these equations can be simply expressed as follows: [Formula 7] Note: There seems to be an "@" symbol in the original text's tags which might be an error. I've translated the text as accurately as possible while keeping those tags intact. If this "@" is significant, it might need further clarification for a more precise translation.TIFF0007872790000007.tif24170 【0036】 The linearization of this model around the reference point is expressed as follows: [Equation 8] TIFF0007872790000008.tif9170 【0037】 X ref_i =(x ref_i ,y ref_i ,θ ref_i ,κ ref_i ) In TIFF0007872790000009.tif9170, TIFF0007872790000010.tif18170 【0038】 Coefficient A i and B i The matrix is ​​determined by the following formula: [Formula 9] TIFF0007872790000011.tif72170[Formula 10] TIFF0007872790000012.tif55170 【0039】 The reference path is sampled to obtain a set of N reference points. For a set of N reference points across a given path, the following formula applies: [Equation 11] TIFF0007872790000013.tif9170 Here, [Formula 12] TIFF0007872790000014.tif31170[Formula 13] TIFF0007872790000015.tif31170[Formula 14] TIFF0007872790000016.tif26170 【0040】 To find a path close to the reference path that is subject to controllability constraints without losing the clothoid shape, the research problem of the present invention is reformulated in the same way as for optimization calculations, where the cost function is as follows. [Formula 15] TIFF0007872790000017.tif9170 【0041】 The cost function J is subject to the following conditions. [Formula 16] TIFF0007872790000018.tif7170 【0042】 U is a decision variable This is a set of definitions for TIFF0007872790000019.tif7170. This set is determined based on constraints regarding the controllability of the system, which will be detailed in equations ([Equation 25]~[Equation 26]). 【0043】 In the cost function, TIFF0007872790000020.tif8170 is a weighting factor, and in this case, The filename is TIFF0007872790000021.tif14170. 【0044】 Matrices Q and R are 4x4 diagonal matrices. 【0045】 The cost function ([Equation 15]) can be expanded into a quadratic optimization problem ([Equation 16]) of constraints required to minimize J: [Formula 17] TIFF0007872790000022.tif12170 Here, [Formula 18] TIFF0007872790000023.tif9170[Formula 19] TIFF0007872790000024.tif9170[Formula 20] TIFF0007872790000025.tif9170 【0046】 Without loss of generality, the term "d" can be deleted from equation ([Equation 17]). 【0047】 The problem of finding an improved path (close to the reference path defined by French Patent Application Publication No. 2003457 dated April 7, 2020) is formulated here in a manner similar to that of conventional quadratic optimization problems. 【0048】 In contrast, controllability constraints (with respect to the maximum overshoot of the final bearing, etc.) and limitations on improved paths will be discussed later in this specification. 【0049】 Solving equations [Equation 17] to [Equation 20] yields a path that closely resembles a clothoid-shaped reference path, satisfying the controllability constraint. However, this does not guarantee the certainty of finding a path that adequately satisfies the system's service and the driver's perception of driving. In this part, constraints on the system's state (such as the maximum lateral deviation and the final orientation) are taken into account in an improved reference path optimization to avoid the problem of turning in the wrong direction when the AES function is activated. As a result, certain constraints are added to the optimization problem below. 【0050】 The first constraint is the constraint on the equation relating to the horizontal coordinate x (longitudinal displacement of the vehicle) of a point on the path, which must be such that its value is the same as the value of a point on the reference path. [Formula 21] x i =x ref_i ∀i=1, ..., N 【0051】 The constraints of the equation make it possible to smooth the discovered path, that is, to ensure its continuity. 【0052】 The following constraints are inequality constraints. Therefore, the second constraint concerns the maximum overshoot from the improved path, and the lateral displacement at a point on the improved path is the lateral displacement Y of the extreme line that must not be exceeded during the correction. maxIt must be smaller. [Formula 22] y i ≦Y max ∀i=1, ..., N 【0053】 The final value of the lateral displacement of the improved path must always be greater than or equal to the final value of the lateral displacement of the reference path. [Formula 23] y N ≧y ref_N 【0054】 The third constraint concerns the handle angle corresponding to the desired path, which is bound in terms of gradient and magnitude so that the controllability constraint is satisfied throughout the entire path. Although it is known that a strict relationship between the handle angle and the decision variable is unavailable, an approximation based on a bicycle model is used. 【0055】 Furthermore, the relationship between the steering angle and curvature can be obtained. [Formula 24] TIFF0007872790000026.tif13170 【0056】 The relationship between the steering wheel's rotation speed and curvature can also be obtained. [Formula 25] TIFF0007872790000027.tif13170 Here, δ i : (In the Cartesian plane) Handle angle at point i TIFF0007872790000028.tif8170l f and l r : The distance from the vehicle's center of gravity to the front axle and rear axle, respectively. m: Vehicle weight c f , c r : Drift stiffness of the front and rear wheels v: Vehicle speed c i : Velocity of the curvature at point i Ratio_DAE: Ratio between handle angle and wheel angle 【0057】 The constraints on steering angle and speed are expressed as follows: [Formula 26] TIFF0007872790000029.tif44170[Formula 27] TIFF0007872790000030.tif45170 Here, TIFF0007872790000031.tif7170 【0058】 The final steering angle must be within an acceptable range to ensure the vehicle's path stabilizes after avoiding an obstacle. [Formula 28] Δ final_min ≤δ N ≤Δ final_max 【0059】 Offset of vertical displacement at the start of correction When the value of TIFF0007872790000032.tif6170 is given, the steering angle must be greater than or equal to the angle value at the moment of AES activation. It should be noted that the moment of AES activation corresponds to the moment of control method activation. The same constraints apply to the vehicle's lateral deviation. When AES is activated and the vehicle's orientation is non-zero at that moment, these constraints are added to avoid orienting the vehicle towards the reference path. It should be noted that the longitudinal displacement value x is not determined during the optimization calculation, but is determined in the optimization strategy derived from equation [Equation 37] described below: [Formula 29] δ i ≥δ1∀i=1, ..., n [Formula 30] y i ≥y1∀i=1, ..., n Here, n is the vertical displacement. This corresponds to TIFF0007872790000033.tif6170. 【0060】 The fourth constraint concerns the stabilization of the vehicle's path after avoidance, based on the final value of the vehicle's heading, which should be located within a sufficiently small range. The purpose of adding this type of constraint is to avoid cases where an optimal solution cannot be found. This type of constraint is often called a "relaxation constraint" or "soft constraint," and its addition involves a cost function ([Equation 17]) and a decision vector. This requires modification of TIFF0007872790000034.tif7170. Therefore, the paper "Soft constraints and exact penalty functions in model predictive control," by Eric C. Kerrigan and Jan M. Maciejowski, UKACC, Cambridge (2000), includes this type of constraint in the control. The addition of this relaxation constraint also makes it possible to converge the vehicle's orientation at the end of the path to a desired range of values. 【0061】 The equation ("Equation 17") can also be expressed as follows: [Formula 31] TIFF0007872790000035.tif13170 Here, [Formula 32] TIFF0007872790000036.tif11170[Formula 33] TIFF0007872790000037.tif13170[Formula 34] TIFF0007872790000038.tif26170TIFF0007872790000039.tif7170P: Weighting parameter for ∈ 【0062】 moreover, [Formula 35] TIFF0007872790000040.tif10170 Here, [Formula 36] TIFF0007872790000041.tif18170θ min and θ max These are the acceptable extreme values ​​of the direction at the end of the path. Assume the following: θ ref_N : The bearing at the last point on the reference route. V: Weighting parameter 【0063】 Based on the theoretical development described above, the optimized decision vector TIFF0007872790000042.tif7170 is evaluated and expressed in equation [Equation 32] by solving the optimization problem in equation [Equation 37] below, which is subject to the constraints defined by equations [Equation 21] to [Equation 23], [Equation 26] to [Equation 30], and [Equation 35]. The decision vector is considered to be optimized when the associated cost J, determined by applying equation [Equation 31], reaches its minimum value. [Formula 37] TIFF0007872790000043.tif14170 【0064】 It should be noted that the solution to the optimization problem obtained in Equation [Equation 37] is effectively advantageous and essential for the AES control system when the vehicle's initial orientation is positive when avoiding to the left. If this initial orientation is negative, the original reference path (not recalculated) is already sufficient to completely avoid to the left by locking the steering wheel to the right. For avoidance to the right, the case of a negative initial orientation is conceivable. Therefore, for avoidance to the right, the clothoid is conceivable, which is derived only from improvements when the vehicle's orientation is strictly negative. In the case of avoidance to the right with a zero or positive initial orientation, the system is stable without requiring any improvement to the path. Furthermore, the initial curvature can be used in place of the initial steering angle by applying the relationship between steering angle and curvature expressed in Equation ([Equation 24]). 【0065】 The determination of whether the steering behavior matches what the driver expects (steps 5 and 12 below) can be verified for left-side avoidance by counting the number of negative extremes in the angle profile when the initial heading is positive. Similarly, for right-side avoidance, the number of positive extremes is counted. If there are multiple negative extremes, it is possible that the steering angle profile is not suitable for the driver's reference. In fact, under the assumption that there are no more negative extremes for the steering angle profile, the first one acts to move the vehicle closer to the reference path, even if the system is currently performing avoidance. This behavior of the system is unacceptable to the driver. In this case, to eliminate this unexpected behavior, the longitudinal displacement The offset value of TIFF0007872790000044.tif6170 is adjusted until only a single negative extremum remains on the angular profile. The file size is increased by increasing it by TIFF0007872790000045.tif6170 each time. 【0066】 The initial value of the lateral deviation of the desired path is the lateral offset value y. L_ref_init This will result in an offset. This offset value is determined before solving the optimization problem presented in equation [equation 37]. The lateral offset value is the angle requirement δ from the AES's robust control device at startup. request It is calculated so that it is equal to the measured handle angle δ(t0). 【0067】 therefore, [Formula 38] TIFF0007872790000046.tif26170, and here, δ request (t0)=δ(t0) θ(t0): The steering wheel bearing measured at system startup. δ(t0): Handle angle measured at system startup. TIFF0007872790000047.tif7170K δ : Gain of the AES control unit corresponding to the vehicle's steering angle K θGain of AES control device corresponding to vehicle orientation D LookAhead : The projected distance of the vehicle in the Cartesian plane 【0068】 Distance D LookAhead This corresponds to the product of the time to collision (TTC) and the vehicle's velocity. To reduce the complexity of implementation, in this study case, it is assumed that the initial steering angle can be considered to be very small and equal to zero. As a result, equation ([Equation 38]) becomes: [Formula 39] TIFF0007872790000048.tif14170 【0069】 Furthermore, to ensure that the final path does not contradict the improved path, an inequality constraint is added to the optimization problem given in equation [Equation 37], which prevents the lateral deviation of this solution from being smaller than that of the improved path. 【0070】 The main substeps of calculation step 4 for improving the avoidance route, as shown in Figure 2, are described below. 【0071】 In the first substep 11, a reference path is obtained from the data of step 3, which determines the object avoidance path, and step 2, which is merged. Then, the initialization parameters are set to the reference path (reference vertical displacement x ref , reference lateral displacement y ref , the vehicle's reference direction θ ref , the reference curvature κ of the vehicle's path ref ) as a function of merged data (initial heading θ of the vehicle at startup t) init , and the initial angle δ of the handle at the time of activation t. init Based on this, the variables ITER, ITER2, and ITER3 are initialized to zero. 【0072】 In the second substep 12, the cost optimization problem represented by equation [Equation 37] is solved, along with the determined initialization parameters, subject to the constraints defined by equations [Equations 21] to [Equation 23], [Equations 26] to [Equation 30], and [Equation 35]. 【0073】 In the third substep 13, the optimized decision vector Equation [Equation 37] is solved by TIFF0007872790000049.tif7170, while the constraints on the final orientation determined by Equation [Equation 35] are related to the minimum value of the cost function J [Equation 31]. A determination is made as to whether TIFF0007872790000050.tif7170 is satisfied, and whether the product of the ITER variable and the value Δt is smaller than a predetermined value, in particular, the collision time TTC obtained from step 2 of merging. 【0074】 Otherwise, the method proceeds to the fourth step 14, where the ITER variable is incremented by one unit and the reference path is extended for a duration equal to ITER times Δt. The method then restarts at the second substep 12. 【0075】 In the third substep 13, if it is determined that an optimal solution exists and the constraints on the final orientation are satisfied, the method proceeds to the fifth substep 15, where a determination is made as to whether the steering behavior is consistent with what the driver expects. To make this determination, the sign of the steering angle associated with the first extremum of the steering angle profile is determined by whether the sign of the steering angle is the same as the sign of the steering angle of the reference path, and whether the extremum with the opposite sign to the first extremum is the largest over the duration of the optimized path, and furthermore, the steering angle profile δ from the moment of AES activation i This will be determined. 【0076】 If the steering wheel behavior does not match what the driver expects, the method proceeds to the sixth substep 16, where the ITER2 variable is incremented by one unit, and the longitudinal displacement is also increased. The offset value of TIFF0007872790000051.tif6170 is The value is increased in TIFF0007872790000052.tif7170, and then the process proceeds to the seventh substep 17, where the cost optimization problem represented by equation [equation 37] is solved again, along with the initialization parameters determined in substep 13 and the increased vertical displacement, subject to the constraints represented by equations ([equation 21]~[equation 23]), ([equation 26]~[equation 30]), and ([equation 35]). The process is restarted in the fifth substep 15. 【0077】 In the fifth substep 15, if the steering wheel behavior matches what the driver expects, the method proceeds to the eighth substep 18, where the last determined path is considered the improved path. 【0078】 If the method were to stop here, the improved path found would still not be sufficient to avoid the problem of the handle locking in the wrong direction when the AES is activated. To overcome this problem, the following steps of the method ensure that the handle rotates in the expected direction, and that this is done regardless of speed. 【0079】 The method proceeds to substep 19, where the cost optimization problem represented by equation [equation 37] is further resolved by considering the improved path as the base path and by reinitializing the ITER2 variable, thereby reducing the offset y defined by equation [equation 39]. L_ref_int This is further solved under the constraints expressed by equations ([Equation 21] to [Equation 23], [Equation 26] to [Equation 30], and [Equation 35]). In other words, the cost optimization problem is solved by initialization parameters defined as a function of the improved path. 【0080】 During the 10th substep 20, the optimized decision vector Equation [Equation 37] is solved using TIFF0007872790000053.tif7170, and a determination is made as to whether an optimal solution exists by checking whether it is associated with the minimum value of the cost function J [Equation 31] and whether the lateral displacement of the solution path is greater than the lateral displacement of the improved path. It should be noted that the optimal orientation is no longer a constraint when determining the optimal solution here, because only the initial part of the improved path is changed, and the improved path already satisfies the constraint on the final orientation. 【0081】 Otherwise, the method proceeds to the 11th substep 21, where the ITER3 variables are incremented by one unit, and the initial azimuth θ init The value Δθ is reduced by Δθ·ITER3. The value Δθ is predetermined, in particular, as a function of the sensitivity of the orientation sensor and by performing an inspection. The method then resumes in the 9th substep 19. 【0082】 If it is determined in substep 20 that an optimal solution exists, the method proceeds to the 12th substep 22, where a determination is made in the same manner as the determination made in substep 23 regarding whether the driver's behavior matches what the driver expects. 【0083】 If the steering wheel behavior does not match what the driver expects, the method proceeds to the 13th substep 23, where the ITER2 variable is incremented by one unit, and the longitudinal displacement is also increased. The offset value of TIFF0007872790000054.tif6170 The value is increased by TIFF0007872790000055.tif7170. The method proceeds to the 14th substep 24, where the cost optimization problem represented by equation [Equation 37] is further solved, subject to the constraints represented by equations ([Equations 21] to [Equation 23], [Equation 26] to [Equation 30], and [Equation 35]) along with the increased longitudinal displacement as an initialization parameter. The method then resumes at the 12th substep 22. 【0084】 If the steering wheel behavior matches what the driver expected in the 12th substep 22, the method proceeds to the 15th substep 25, where the last determined path is considered the final path. 【0085】 It should be noted that the predetermined duration, angular deviation, and vertical offset are continuously increased by predetermined limit values. In such cases, the improvement step is interrupted, and the vehicle control step follows the standard path. 【0086】 This control method improves path-following stability and client service, along with steering wheel rotation in the expected direction. Operation duration is also reduced by the proposed method.

Claims

[Claim 1] A control method for controlling an automated vehicle to avoid an object, wherein the automated vehicle is equipped with at least two sensing sensors, and the control method comprises the following steps: The steps include determining the data from the at least two sensors, A step of merging the data from the at least two sensors to determine at least the steering angle, the vehicle's speed, and the vehicle's direction, The steps include: planning a reference path which is a path to avoid objects with a clothoid shape, The steps include improving the reference path based on solving an optimization problem that minimizes a predetermined cost function as a function of the steering angle, the vehicle's speed, and the vehicle's direction, The steps include controlling the vehicle to perform the improved reference path, A method for improving the reference path, comprising: a first set of substeps for further improving the reference path as a function of the length of the reference path, the direction of rotation of the handle, and the final bearing; and a second set of substeps for further improving the improved reference path as a function of the initial bearing and the direction of rotation of the handle. [Claim 2] The control method according to claim 1, wherein the optimization problem is a function of a reference longitudinal displacement, a reference lateral displacement, the orientation of a reference vehicle, the curvature of the reference path of the reference vehicle, a decision vector including the rate of change of the curvature of the reference path and the distance traveled relative to the origin, and empirically determined setting parameters, and constraints on the initial longitudinal displacement, initial lateral displacement, final lateral displacement, the steering angle, the steering rotation speed, and the final orientation value. [Claim 3] The first series of substeps described above consists of the following substeps: The initialization parameters of the optimization problem are determined to be a function of the reference path and the merged data, and the initialization parameters include a reference longitudinal displacement, a reference lateral displacement, the orientation of the reference vehicle, the curvature of the reference vehicle path, the initial orientation of the vehicle, and the initial steering angle. A determination is made as to whether an optimal solution exists by determining whether the optimization problem is solved, whether the solution to the optimization problem corresponds to the minimum value of the predetermined cost function, and whether the solution to the optimization problem satisfies the constraints on the final orientation of the vehicle. In such a case, the handle angle profile is determined from the moment the control method is activated, and then a determination is made as to whether the sign of the handle angle associated with a first extremum of the handle angle profile has the same sign as the sign of the handle angle of the reference path, and whether, over the duration of the optimized path which is the solution to the optimization problem, there is also a maximum extremum with the opposite sign to the first extremum, in which case the solution to the optimization problem is considered an improved path. [Claim 4] The control method according to claim 1, wherein if it is determined that no optimal solution exists, the reference path is extended for a predetermined duration, and the method is then resumed when the optimization problem is solved by taking into account the extension of the reference path, and the predetermined duration is increased each time it is determined that no optimal solution exists. [Claim 5] The control method according to claim 3, wherein if the steering angle profile is not satisfied, a longitudinal offset is applied to the reference path, the optimization problem is solved again, and then the method is restarted when determining the steering angle profile, increasing the longitudinal offset each time it is determined that the steering angle profile is not satisfied. [Claim 6] The control method according to any one of claims 3 to 5, wherein the constraint on the optimization problem is that the vertical displacement of a point in the improved path must be equal to the vertical displacement of the point in the reference path. [Claim 7] The control method according to any one of claims 3 to 6, wherein the constraint on the optimization problem is that the final value of the vehicle's direction on the improved route must be within a limited range of values. [Claim 8] The control method according to any one of claims 3 to 7, wherein the constraint on the optimization problem is that the lateral displacement of the points in the improved path must be less than the lateral displacement of the extreme line that is not exceeded during the correction. [Claim 9] The second series of substeps described above consists of the following substeps: A control method according to claim 3 or 5, comprising determining new initialization parameters as a function of the improved path and the merged data, determining whether an optimal solution exists by determining whether a predetermined lateral offset is applied to the improved path to solve the optimization problem and whether it corresponds to a minimum of a predetermined cost function, in which case the handle angle profile is determined from the moment of the control method, determining whether the sign of the handle angle associated with the first extremum of the handle angle profile has the same sign as the sign of the handle angle of the reference path, and determining whether, over the duration of the optimized path, the extremum with the opposite sign to the first extremum is also the largest, in which case the solution to the optimization problem is considered the final improved path. [Claim 10] The control method according to claim 9, wherein if it is determined that no optimal solution exists, the initial orientation is reduced by a predetermined angular deviation, and then the method is restarted when the optimization problem is solved, and the angular deviation is increased each time that the absence of an optimal solution occurs consecutively. [Claim 11] The control method according to claim 9 or 10, wherein if it is determined that the optimal solution does not satisfy the steering angle constraint, a longitudinal offset is applied to the improved path, the optimization problem is solved, and the method is then restarted when determining the steering angle profile, increasing the longitudinal offset each time it is determined that the steering angle profile is not satisfied. [Claim 12] If it is determined that no optimal solution exists, the reference path is extended for a predetermined duration, and the initial orientation is reduced by a predetermined angular deviation. If the handle angle profile is not met, a longitudinal offset is applied to the reference path. The control method according to any one of claims 1 to 11, wherein the predetermined duration, the angular deviation, and the vertical offset are continuously increased up to a predetermined limit value, in which case the improvement step is interrupted and the reference path is executed in the step of controlling the vehicle. [Claim 13] A system for controlling an automated vehicle to avoid an object, wherein the automated vehicle comprises at least one sensing sensor and at least one computing means configured to perform the control method described in any one of claims 1 to 12.

Images