Methods for avoiding obstacles

Description

TECHNICAL FIELD OF THE INVENTION

[0001] The present invention relates generally to the automation of the tracking of motor vehicle trajectories.

[0002] It finds a particularly advantageous application in the context of driving aids for motor vehicles.

[0003] It relates more particularly to a method for automated control of a motor vehicle enabling this vehicle to follow an obstacle avoidance trajectory, in particular on a bend.

[0004] It also concerns a motor vehicle equipped with a computer suitable for implementing this process. STATE OF THE ART

[0005] In an effort to make motor vehicles safer, they are currently being equipped with driver assistance systems or autonomous driving systems.

[0006] Among these systems, we know in particular the automatic emergency braking systems (better known by the abbreviation AEB, from the English "Automatic Emergency Braking"), designed to avoid any collision with obstacles located in the lane taken by the vehicle. These systems are designed to detect an obstacle on the lane and to act, in this situation, on the conventional braking system of the motor vehicle.

[0007] However, there are situations in which these emergency braking systems do not prevent a collision or cannot be used (for example, if another vehicle is following too closely behind the motor vehicle).

[0008] For these situations, automatic avoidance systems (better known by the abbreviation AES, from the English "Automatic Evasive Steering" or "Automatic Emergency Steering") have been developed which allow the obstacle to be avoided by diverting the vehicle from its trajectory, by acting on the vehicle's direction.

[0009] However, it happens that this AES system, after having calculated a trajectory to follow to avoid the obstacle, does not allow the vehicle to be guided with the desired precision, which can prove very dangerous.

[0010] Document US9731755 then proposes taking into account in the calculations certain effects of the road, such as its curvature, its slope and its longitudinal slope, so that the trajectory tracking is better.

[0011] More specifically, this paper proposes to calculate a steering instruction that minimizes the deviation between the vehicle trajectory and the desired trajectory. The control used for this purpose comprises a closed-loop feedback part and an open-loop part, which takes into account a predictively determined road disturbance factor.

[0012] But even with this technical solution, the lane tracking is not as precise as desired. Documents DE 102009043070 A1 and FR 3088275 A1 present comparable driver assistance systems. PRESENTATION OF THE INVENTION

[0013] In order to overcome the aforementioned drawback of the state of the art, the present invention proposes to solve the problem of following the avoidance trajectory by taking into account the roll and / or lateral acceleration experienced by the vehicle, which have a significant impact on the trajectory taken by the vehicle.

[0014] More particularly, the invention proposes a method for automated control of a motor vehicle, comprising steps of: acquisition of parameters relating to a trajectory for avoiding an obstacle by the motor vehicle, acquisition of a lateral acceleration or a roll angle experienced by the motor vehicle, calculation by a computer of a provisional control instruction for a steering actuator for the steered wheels, as a function of said parameters and by means of a closed-loop controller, and open-loop calculation of a correction term for said provisional control instruction, as a function of the lateral acceleration or the roll angle acquired.

[0015] Thus, the closed-loop controller includes static state feedback calculated from the vehicle and road state variables at time t, without having to make any predictions that would be a source of error.

[0016] The open-loop correction term is then calculated based on vehicle roll or lateral acceleration, to account for the effect of roll on the trajectory. This roll or lateral acceleration can be measured or calculated.

[0017] The correction term can also be calculated based on the road curvature and the closed-loop controller gains.

[0018] Thus, other advantageous and non-limiting characteristics of the method according to the invention, taken individually or in all technically possible combinations, are the following: the correction term is calculated as a function of a steering coefficient induced by the roll of the motor vehicle on the front wheels of the motor vehicle and / or a steering coefficient induced by the roll of the motor vehicle on the rear wheels of the motor vehicle; the correction term is calculated as a function of the curvature of the road taken by the motor vehicle; the correction term is equal to the sum of the product of a first variable multiplied by the lateral acceleration or by the roll angle, and the product of a second variable multiplied by the curvature of the road; the correction term is calculated as a function of at least one of the gains of the closed-loop controller: the correction term is calculated as a function of an understeer gradient; the understeer gradient is calculated as a function of at least one of the gains of the closed-loop controller; the understeer gradient is calculated using the following equation: K V = l r 1 − k δ − k δ ref C f − l f 1 − k δ − k δ ref + k ψ L − k β C r ⋅ m l f + l r where m, C f , C r , lf , lr are parameters related only to the architecture of the motor vehicle, and where k δ , k δref , K ΨL and k β are gains of the closed-loop controller; the correction term is calculated using the following equation: δ Ffwd = l f + l r 1 − k δ − k δ ref + k ψ L − k β l r + k ψ L l s − k r V + K V V 2 ⋅ ρ ref − 1 − k δ − k δ ref ε 1 − 1 − k δ − k δ ref + k ψ L − K β ε 2 ⋅ ϕ where: lf , lr are parameters related only to the architecture of the motor vehicle, V is the longitudinal speed of the motor vehicle, ls is a predetermined sighting distance, ρ ref is the radius of curvature of the road at the level of the motor vehicle, ε 1 is the steering coefficient induced by the roll on the front wheelset, ε 2 is the steering coefficient induced by the roll on the rear wheelset, Φ is the roll angle, Kv is the understeer gradient, k δ , k δref , k ΨL , kr , and k β are gains of the closed-loop controller.

[0019] The invention also proposes a motor vehicle comprising a steering actuator for the steering wheels and a computer for controlling said actuator, programmed to implement a method as mentioned above.

[0020] Of course, the various features, variants and embodiments of the invention may be combined with each other in various combinations to the extent that they are not incompatible or mutually exclusive. DETAILED DESCRIPTION OF THE INVENTION

[0021] The description which follows with reference to the appended drawings, given as non-limiting examples, will make it clear what the invention consists of and how it can be implemented.

[0022] On the attached drawings: [ Fig. 1 ] is a schematic top view of a motor vehicle traveling on a road, on which the trajectory that this vehicle must take is represented; [ Fig. 2 ] is a schematic top view of the motor vehicle of the figure 1 , represented in four successive positions located along an obstacle avoidance trajectory; [ Fig. 3 ] is a schematic top view of the motor vehicle of the figure 1 , allowing observation of the influence of rolling on the trajectory of the vehicle; [ Fig. 4 ] is a graph illustrating the parameters for stabilizing a closed-loop transfer function used in a method according to the invention; [ Fig. 5 ] is a block diagram which illustrates the different algorithms making it possible to implement the method according to the invention; [ Fig. 6 ] is a diagram illustrating a transfer function used to implement a method according to the invention.

[0023] On the figure 1 , a motor vehicle 10 is shown, conventionally comprising a chassis which delimits a passenger compartment, two front steered wheels 11, and two rear non-steered wheels 12. As a variant, these two rear wheels could also be steered, which would however require an adaptation of the control law described below.

[0024] This motor vehicle 10 comprises at least one conventional steering system making it possible to act on the orientation of the front wheels 11 so as to be able to turn the vehicle. This conventional steering system notably comprises a steering wheel connected to connecting rods in order to pivot the front wheels 11. In the example considered, it also comprises an actuator 31 (shown in the figure 5 ) allowing action on the orientation of the front wheels depending on the orientation of the steering wheel and / or depending on a request received from a computer 13.

[0025] In addition, this motor vehicle could also include a differential braking system making it possible to act differently on the two front wheels 11 (or even on the two rear wheels 12) so as to slow down the motor vehicle by making it turn. Such a differential braking system comprises, for example, a controlled differential or electric motors placed at the level of the wheels of the vehicle or even braking calipers controlled independently of each other. It thus includes at least one actuator which is designed to act differently on the rotation speeds of the wheels depending on a request received from the computer 13.

[0026] For the sake of clarity, it will be assumed here that the vehicle does not have such a differential braking system. If it did, this would require an adaptation of the control law described below.

[0027] The computer 13 is designed to control the power steering actuator 31 according to the traffic conditions encountered. For this purpose, it comprises at least one processor, at least one memory and an input and output interface connected to a vehicle data network (typically a CAN network).

[0028] Thanks to its interface, the calculator 13 is adapted to receive input signals from different sensors.

[0029] Among these sensors, for example, it is planned: a device such as a front camera, making it possible to locate the position of the vehicle in relation to its traffic lane, a device such as a RADAR or LIDAR remote sensor, making it possible to detect an obstacle 20 located on the path of the motor vehicle 10 ( figure 2 ), at least one lateral detection device, such as a RADAR or LIDAR remote sensor, making it possible to observe the environment on the sides of the vehicle, a device such as a gyrometer, making it possible to determine the yaw rotation speed (around a vertical axis) of the motor vehicle 10, a steering wheel position and angular speed sensor, and a torque sensor applied by the driver to the steering wheel.

[0030] Thanks to its interface, the computer 13 is adapted to transmit an instruction to the power steering actuator 31.

[0031] It thus makes it possible to force the vehicle to follow an avoidance trajectory T0 of obstacle 20 (see figure 2 ).

[0032] Thanks to its memory, the computer 13 stores data used in the process described below.

[0033] In particular, it stores a computer application, consisting of computer programs comprising instructions whose execution by the processor allows the computer to implement the method described below.

[0034] Before describing this process, we can introduce the different variables and data that will be used, some of which are illustrated in the figure 1 .

[0035] We first considered a reference frame (CG, X, Y, Z) attached to the vehicle, whose origin is located at the center of gravity of the vehicle, whose X axis coincides with the longitudinal axis A1 of the vehicle (going from the rear to the front), whose Y axis is a lateral axis (going here from right to left), and whose Z axis is a vertical axis when the vehicle is on a horizontal road.

[0036] The total mass of the motor vehicle will be noted “m” and will be expressed in kg.

[0037] The inertia of the motor vehicle around the Z axis will be noted “J” or “I z” and will be expressed in Nm

[0038] The inertia of the motor vehicle around the X axis will be noted I x and will be expressed in Nm

[0039] The distance between the center of gravity CG and the front axle of the vehicle will be noted “lf” and will be expressed in meters.

[0040] The distance between the center of gravity CG and the rear axle will be noted “lr” and will be expressed in meters.

[0041] The front wheel drift stiffness coefficient will be noted “C r” and will be expressed in N / rad.

[0042] The rear wheel drift stiffness coefficient will be noted “C r” and will be expressed in N / rad.

[0043] These wheel drift stiffness coefficients are concepts well known to those skilled in the art. For example, the front wheel drift stiffness coefficient is thus the one that allows the equation F r = 2.C f .α f to be written, with F f the lateral sliding force of the front wheels and α f the front wheel drift angle.

[0044] The steering angle that the front steered wheels make with the longitudinal axis A1 of the motor vehicle 10 will be noted “δ” and will be expressed in rad.

[0045] The variable δ ref , expressed in rad, will designate the saturated steering angle setpoint, as it will be transmitted to the power steering actuator.

[0046] The variable δ K , expressed in rad, will designate the unsaturated steering angle setpoint. At this stage, we can only specify that the concept of saturation will be linked to value limits or value variation.

[0047] The vehicle's yaw rate (around the vertical axis passing through its center of gravity CG) will be noted "r" and will be expressed in rad / s.

[0048] The relative heading angle between the longitudinal axis A1 of the vehicle and the tangent to the avoidance trajectory T0 (desired trajectory of the vehicle) will be noted “Ψ L” and will be expressed in rad.

[0049] The lateral deviation between the longitudinal axis A1 of the motor vehicle 10 (passing through the center of gravity CG) and the avoidance trajectory T0, at a sighting distance “Is” located in front of the vehicle, will be noted “y L” and will be expressed in meters.

[0050] The lateral deviation instruction between the longitudinal axis A1 of the motor vehicle 10 (passing through the center of gravity CG) and the avoidance trajectory T0, at a sighting distance “Is” located in front of the vehicle, will be noted “y L-ref” and will be expressed in meters.

[0051] The trajectory tracking error will be noted “e yL” and will be expressed in meters. It will be equal to the difference between the lateral deviation setpoint y L-ref and the lateral deviation y L .

[0052] The aforementioned aiming distance “Is” will be measured from the center of gravity CG and will be expressed in meters.

[0053] The drift angle of the motor vehicle 10 (angle that the speed vector of the motor vehicle makes with its longitudinal axis A1) will be noted “β” and will be expressed in rad.

[0054] The speed of the motor vehicle along the longitudinal axis A1 will be noted “V” and will be expressed in m / s.

[0055] The constant "g" will be the acceleration of gravity, expressed in ms -2< .

[0056] The roll angle taken by the vehicle relative to the vertical, expressed in radians, will be noted Φ.

[0057] When calculating this roll angle, the following parameters can be considered: br is the roll damping coefficient, expressed in N.rad -1< .s -1< , kr is the roll stiffness, expressed in N.rad -1< , h is the distance between the vehicle's roll center and the ground.

[0058] The lateral acceleration experienced by the vehicle along the Y axis will be noted ay and will be expressed in ms -2< .

[0059] The average curvature of the road at the level of the motor vehicle will be noted ρ ref and will be expressed in m -1< . This curvature is equal to the inverse of the radius of curvature of an arc of a circle which is formed by the road at the precise level of the center of gravity of the motor vehicle.

[0060] The constants “ξ” and “ω” will represent dynamic characteristics of the steering angle of the vehicle’s front wheels, more precisely ξ represents the damping and ω is the natural frequency of the electric power steering (EPS).

[0061] The constant “ω f” will represent a dynamic characteristic of an arbitrary limited disturbance “w” applied to the vehicle.

[0062] The steering speed will refer to the angular steering speed of the front steered wheels.

[0063] As a hypothesis, for the sake of clarity, the road on which the motor vehicle is traveling is assumed to be flat and horizontal.

[0064] Before describing the method which will be executed by the computer 13 to implement the invention, we can describe the calculations which made it possible to establish the algorithms making it possible to implement the invention, so as to clearly understand where these algorithms come from and on what springs they are based.

[0065] As a preliminary point, it should be noted here that, to simplify the presentation, only power steering will be used to divert the vehicle from its initial trajectory. Differential braking will not be installed or activated.

[0066] On the figure 5 , we modeled the process of calculating the saturated steering angle setpoint δ ref to be transmitted to the power steering actuator 31.

[0067] In summary, it is expected that the controller determines a trajectory to follow, determines a first steering angle setpoint (hereinafter called unsaturated steering angle setpoint δ K ) taking into account this trajectory and parameters linked to the dynamics of the vehicle, then saturates this setpoint in order to obtain the saturated steering angle setpoint δ ref to be transmitted to the actuator 31.

[0068] On the figure 5 , block Z1 corresponds to the block which makes it possible to determine the trajectory to follow to avoid obstacle 20.

[0069] The obstacle 20 may be in the same lane as that of the motor vehicle 10 or in the adjacent lane, provided that it is located on the initial trajectory of the vehicle at a short distance taking into account the speed of the vehicle.

[0070] Since the method of determining this avoidance trajectory T0 is not the subject of the present invention, it will not be described here. It will only be noted that this block Z1 makes it possible to determine, when the AES function is activated, a lateral deviation instruction y L-ref and its time derivative. It will be noted that this avoidance trajectory T0 will be considered here as static but that as a variant, a dynamic trajectory (recalculated according to new obstacles detected or according to the actual trajectory of the vehicle) could be used.

[0071] Block Z6 illustrates the controller K Fbck which would be usable and reliable if the motor vehicle 10 were traveling in a straight line. This controller corresponds to a mathematical operator making it possible to calculate a provisional steering angle setpoint (hereinafter called the first component δ Fbck of the unsaturated steering angle setpoint δ K ) making it possible to control the wheels of the vehicle so that the latter follows the avoidance trajectory T0.

[0072] Blocks Z2, Z3 and Z4 illustrate algorithms that allow the calculation of a correction term to be applied to the first component δ Fbck to take into account the impact of the vehicle's roll on its trajectory. This correction term will hereinafter be called the second component δ Ffwd of the unsaturated steering angle setpoint δ K .

[0073] The sum of the first component δ Fbck and the second component δ Ffwd makes it possible to calculate the unsaturated steering angle setpoint δ K .

[0074] Block Z5 illustrates a constraint that we wish to apply to the unsaturated steering angle setpoint δ K .

[0075] This constraint here is a limitation of the steering wheel turning speed. In fact, we want the steering wheel turning angular speed not to exceed a threshold noted υ , in order to make the vehicle controllable if the driver had to take back control of the vehicle during the obstacle avoidance phase.

[0076] Other constraints could be considered. For example, a saturation constraint on the steering angle amplitude could be used. But here, for the sake of clarity, this second constraint will not be considered. Alternatively, one could also not provide any constraint.

[0077] Block Z5 therefore makes it possible to provide a saturated steering angle setpoint, hereinafter referenced δ sat . This saturated setpoint serving as a reference for controlling the power steering actuator 31, will also be referenced δ ref .

[0078] We can now detail the calculations that led to the invention.

[0079] The steering of the front wheels 11 of the vehicle can be modeled simply by the following mathematical formula. δ ¨ δ ˙ = − 2 ξω − ω 2 1 0 δ ˙ δ + ω 2 0 δ ref

[0080] First, we will consider the hypothesis that the curvature of the road taken is zero. Thus, if we do not consider the roll and curvature of the road, we can model the dynamic behavior of the vehicle using the following equation. β ˙ r ˙ ψ ˙ L y ˙ L = − C f + C r mv − 1 + C r l r − C f l f mv 2 − C f l f − C r l r J − C r l r 2 + C f l f 2 Jv 0 0 0 0 0 1 v l s 0 0 v 0 β r ψ L y L + 0 0 − v 0 ρ ref + C f mv C f l f J 0 0 δ

[0081] This modeling of vehicle behavior is called the "bicycle model" because it does not take into account the fact that each axle of the vehicle has not one, but two wheels. However, it allows for a good modeling of the vehicle's trajectory as a function of the steering angle of its steered wheels (here the front wheels), when the vehicle is traveling straight.

[0082] However, as shown in the figure 3 , the roll taken by the motor vehicle 10 when cornering has an influence on the actual trajectory of the vehicle, so much so that it is desired to take it into account.

[0083] Here, we can define the roll by an angle, hereinafter called "roll angle Φ", measuring the angle of inclination of the vehicle body by the road around the longitudinal axis A1 of this vehicle.

[0084] The steering coefficient induced by the roll on the front axle of the vehicle will be noted ε 1 . Once multiplied by the roll angle Φ, it will make it possible to evaluate how this roll modifies the steering angle of the front wheels of the vehicle and therefore the trajectory T1 represented on the figure 3 .

[0085] The steering coefficient induced by the roll on the rear axle will be noted ε 2 . Once multiplied by the roll angle Φ, it will allow us to evaluate how this roll modifies the steering angle of the rear wheels.

[0086] As shown in the figure 3 , the roll actually tends to move the vehicle away from its trajectory.

[0087] To take this phenomenon into account, the bicycle model presented above is improved by means of the following equation. β ˙ r ˙ ψ ˙ L y ˙ L = − C f + C r mv − 1 + C r l r − C f l f mv 2 − C f l f − C r l r J − C r l r 2 + C f l f 2 Jv 0 0 0 0 0 1 v l s 0 0 v 0 β r ψ L y L + 0 0 − v 0 ρ ref + C f mv C f l f J 0 0 δ + ε 1 ϕ + C r mv − C r l r J 0 0 ε 2 ϕ

[0088] At this point, we can note that the roll angle Φ can be measured, or estimated. Here, it will be modeled by the following equation. I x ϕ ¨ + b r ϕ ˙ + k r ϕ = ma y h + mghϕ

[0089] It could then be estimated by a Kalman filter. However, here it is more simply estimated at block Z4 of the figure 5 , using the following equation. ϕ = ma y h k r − mgh

[0090] To establish this equation, we considered that the variation and acceleration of the roll are substantially zero. Indeed, most of the time, the vehicle is more or less in an established turn, so the dynamics of the roll (its derivatives) are substantially zero.

[0091] Here, we seek to ensure good trajectory tracking by minimizing the trajectory tracking error e yL , which we recall is expressed using the following equation. e yL = y L − y L _ ref

[0092] In this embodiment of the invention, it is assumed that the dynamics of the reference trajectory is represented by the equation below: y ¨ L _ ref = − ω f y ˙ L _ ref + ω f w

[0093] The bicycle model described so far, however, does not allow the steering speed of the front wheels 11 of the vehicle to be limited. However, such limitations are particularly important to ensure that the driver of the vehicle is able to regain control of the vehicle at any time.

[0094] The saturation of the steering speed illustrated on the figure 5 by block Z5 can be put into an equation as follows: δ ˙ ref ≤ υ

[0095] In this equation, the threshold υ is for example equal to 0.0491 Rad / s, which corresponds to 0.785 Rad / s at the steering wheel (i.e. 45° / s) if the steering gear ratio is equal to 16.

[0096] As shown in the figure 6 , the steering speed limiter used here is particular in that it forms a pseudo-closed-loop controller which includes: a direct chain transfer function equal to the product of the threshold υ (to respect the condition stipulated by the aforementioned equation), of an integrator in 1 / s, and of a corrector which is a hyperbolic tangent type function of Δ.α, an indirect chain transfer function (or “feedback chain”) equal to one.

[0097] It receives as input the unsaturated steering angle setpoint δ K and transmits as output the saturated steering angle setpoint δ ref .

[0098] In this figure, the coefficient Δ corresponds to the difference between the variables δ K and δ ref . The coefficient α is a constant between 0 and infinity, which is the only parameter allowing to influence the fast or flexible nature of the steering speed limiter.

[0099] This steering speed limiter has the advantage of being simple to set up since it is sufficient to adjust the coefficient α. It ensures continuous and smooth control (infinitely derivable).

[0100] So, given the shape of this steering speed limiter, we can write the equation: δ ˙ ref = υ . tanh α δ K − δ ref

[0101] This gives a vehicle controllability model which is pseudo-linear.

[0102] We can actually introduce the following parameter θ:

[0103] θ = tanh α . δ K − δ ref α . δ K − δ ref

[0104] Then rewrite the above equation in linear form: δ ˙ ref = − υ . α . θ . δ ref + υ . α . θ . δ K

[0105] This equation is characteristic of a state representation and it shows that the setpoint variation limiter model is linear as a function of the parameter θ.

[0106] It is then possible, on this basis, to determine the controller K Fbck , which we recall ensures good monitoring of the avoidance trajectory T0, but that it does not take into account the roll taken by the vehicle.

[0107] Given the above equations, we can further enrich the bicycle model to obtain a new usable model. This new enriched model is written: β ˙ r ˙ ψ ˙ L e ˙ y L δ ¨ δ ˙ y ¨ L ref δ ˙ ref = − C f + C r mV − 1 + C r l r − C f l f mV 2 0 0 0 C f mV 0 0 − C f l f − C r l r J − C r l r 2 + C f l f 2 JV 0 0 0 C f l f J 0 0 0 1 0 0 0 0 0 0 V l s V 0 0 0 − 1 0 0 0 0 0 − 2 ξω − ω 2 0 ω 2 0 0 0 0 1 0 0 0 0 0 0 0 0 0 − ω f 0 0 0 0 0 0 0 0 − υαθ β r ψ L e y L δ ˙ δ y ˙ L ref δ ref + 0 0 0 0 0 0 0 vαθ δ K + 0 0 − V 0 0 0 0 0 ρ ref ε 1 C f + ε 2 C r mv ε 1 C f l f − ε 2 C r l r J 0 0 0 0 0 0 ϕ + 0 0 0 0 0 0 ω f 0 w y ref

[0108] We remind you that we can write. δ K = δ Fbck + δ Ffwd

[0109] This equation clearly illustrates that the unsaturated steering instruction δ K can be obtained using a first controller K Fbck and a correction term taking into account the influence of roll on the vehicle trajectory.

[0110] Firstly, the K Fbck controller, which is of the "feedback" type (or closed loop), makes it possible to determine the first component δ Fbck in order to guarantee the stability and robustness of the closed loop. Secondly, the correction term (calculated in open loop or "feed-forward") makes it possible to determine the second component δ Ffwd in order to compensate for the effects of the road curvature and the roll (induced by the road curvature) on the nominal performance of the K Fbck controller.

[0111] We can then first look at the synthesis of the K Fbck controller.

[0112] We can write:

[0113] δ Fbck = k β k r k ψ L k e y L k δ ˙ k δ k y ˙ L _ ref k δ ref β r ψ L e y L δ ˙ δ y ˙ L _ ref δ ref where k β , kr , k ψL , k e y L , k δ̇ , k δ , k ẏ L_ref And k δref are the "gains" of the K Fbck controller. This K Fbck controller is calculated with the assumption that curvature and roll are negligible. Its objective is to guarantee the stability of the closed loop and to minimize the impact of the disturbance w yref on the reference avoidance trajectory tracking. In fact, the global controller is separated into two parts: on the one hand, the "feedback" which guarantees stability and the reference trajectory tracking in the nominal case (without the effect of curvature), and, on the other hand, the "feed-forward" compensates for the effect of curvature to have the iso performance (as in a case of a road without curvature).

[0114] We can then consider a state vector x, which we write in the form: x = β r Ψ L e y L δ ˙ δ y ˙ L _ ref δ ref T

[0115] The objective is then to determine the form of the controller K Fbck . To do this, we can write our behavioral model in a generic form: x ˙ = A θ x + B u θ . δ ref + B w w y = C y . x

[0116] In this equation, C y is the identity matrix, A is a dynamic matrix, B u is a control matrix and B w is a perturbation matrix, which can be written in the form: A = − C f + C r mV − 1 + C r l r − C f l f mV 2 0 0 0 C f mV 0 0 − C f l f − C r l r J − C r l r 2 + C f l f 2 JV 0 0 0 C f l f J 0 0 0 1 0 0 0 0 0 0 V l s V 0 0 0 − 1 0 0 0 0 0 − 2 ξω − ω 2 0 ω 2 0 0 0 0 1 0 0 0 0 0 0 0 0 0 − ω f 0 0 0 0 0 0 0 0 − υαθ B u = 0 0 0 0 0 0 0 vαθ , B w = 0 0 0 0 0 0 ω f 0

[0117] The K Fbck controller, which is defined as a static state return, can be expressed in the form: δ Fbck = K Fbck . x

[0118] To find an optimal K Fbck controller, different methods can be used.

[0119] The method used here is that of linear matrix inequalities. It is thus carried out from convex optimization criteria under constraints of linear matrix inequalities.

[0120] The objective is more precisely to optimize the gains of the closed loop defined by the K Fbck controller by playing on the choice of poles.

[0121] The matrix inequalities used are three in number and are defined by the following inequalities. A i Q + B i R + A i Q + B i R T + 2 μQ ≺ 0 − γQ A i Q + B i R ∗ − γQ ≺ 0

[0122] sin φ A i Q + B i R + A i Q + B i R T cos φ A i Q + B i R − A i Q + B i R T ∗ sin φ A i Q + B i R + A i Q + B i R T ≺ 0

[0123] In these inequalities, the index i is equal to 1 or 2, and we can then define the matrices A i and B i as follows: A 1 = A θ min , A 2 = θ max , B 1 = B u θ min , B 2 = B u θ max

[0124] A matrix of the form X Y Y T W is written in the form X Y ∗ W .

[0125] The K Fbck controller is defined by the equation: K Fbck = R . Q − 1

[0126] The vehicle speed is assumed to be constant (so all matrices in the system are considered constant).

[0127] The three inequalities ensure that the dynamics of the closed loop remain limited. Indeed, thanks to these constraints, the poles of the closed loop are bounded in a zone defined by a radius γ, a minimum distance from the imaginary axis µ, and an opening angle ϕ (see figure 4 ).

[0128] This method is effective when it comes to determining the steering wheel angle at all times in a reasonable way (and controllable by a driver with average skills) and in a way that is achievable by the actuator. These constraints also ensure the stability of the closed loop.

[0129] The objective here is to minimize the γ ray. Once the controller gains K Fbck are obtained, we can calculate the controller result using the following formula: δ Fbck = k β k r k ψ L k e y L k δ ˙ k δ k y ˙ L _ ref k δ sat β r ψ L e y L δ ˙ δ y ˙ L _ ref δ ref

[0130] The values ​​θ min and θ max were introduced into the three matrix inequalities.

[0131] The value of θ, which is related to the gap between δ K and δ ref , reflects the level of violation by the controller K Fbck of the controllability limit stated by equation [Math.8].

[0132] By definition, θ is between 0 (exclusive) and 1 (inclusive). When θ is equal to 1, the calculated unsaturated steering wheel angle setpoint δ K respects the controllability limit. When it is close to 0, the calculated unsaturated steering wheel angle setpoint δ K has a value that imposes too great a steering dynamic, which generates a risk of vehicle instability. When θ takes intermediate values ​​between 0 and 1, the controllability limit is not respected but it is possible that there is no risk of vehicle instability.

[0133] In other words, the choice of the values ​​θ min and θ max has a direct impact on the performance and robustness of the K Fbck controller. The larger the range [θ min ,θ max ], the less efficient the K Fbck controller is but the more robust it is. On the contrary, the smaller this range, the more efficient the K Fbck controller is but the less robust it is.

[0134] Logically the value θ max is chosen equal to 1 (case in which the K Fbck controller operates in linear mode, as is generally the case, without violation of controllability constraints).

[0135] Determining the value θ min, on the other hand, requires a compromise between performance and robustness. Determining this value amounts to imposing a maximum threshold for the difference, in absolute value, between δ K and δ ref .

[0136] When cornering, the dual effect of curvature and roll becomes more impactful on the performance of the K Fbck controller.

[0137] The invention then proposes to calculate the second component δ Ffwd to compensate for the effect of the curvature and that of the roll in order to maintain a good level of tracking of the avoidance trajectory T0.

[0138] This second component δ Ffwd can be calculated during each avoidance phase or only for those occurring on curved roads (when the radius of curvature of the road is less than a predetermined threshold), the curvature of the road being for example derived from data perceived by a camera.

[0139] As it appears clearly on the figure 5 , to obtain this second component δ Ffwd , we use the gains of the controller K Fbck .

[0140] In practice, in equation [Math.12], we replace the unsaturated steering angle setpoint δ K by its expression from equation [Math.13], which makes it possible to obtain the equation of a following closed-loop system. β ˙ r ˙ ψ ˙ L e ˙ y L δ ¨ δ ˙ y ¨ L _ ref δ ˙ ref = − C f + C r mV − 1 + C r l r − C f l f mV 2 0 0 0 C f mV 0 0 − C f l f − C r l r J − C r l r 2 + C f l f 2 JV 0 0 0 C f l f J 0 0 0 1 0 0 0 0 0 0 V l s V 0 0 0 − 1 0 0 0 0 0 − 2 ξω − ω 2 0 ω 2 0 0 0 0 1 0 0 0 0 0 0 0 0 0 − ω f 0 k β vαθ k r vαθ k ψ L vαθ k e y L k δ ˙ vαθ k δ vαθ k y ˙ L r ef vαθ − vαθ 1 − k δ sat β r ψ L e y L δ ˙ δ y ˙ L _ ref δ ref + 0 0 0 0 0 0 0 vαθ δ Ffwd + 0 0 − V 0 0 0 0 0 ρ ref + ε 1 C f + ε 2 C r mv ε 1 C f l f − ε 2 C r l r J 0 0 0 0 0 0 ϕ + 0 0 0 0 0 0 ω f 0 w y ref

[0141] Knowing that the disturbance w yref has already been compensated when calculating the component δ Fbck , it will then only be necessary in the calculation of the second component δ Ffwd to minimize the influence of the other disturbances (curvature and roll) on the static avoidance trajectory tracking error e yL .

[0142] To do this, first of all, we can calculate the static values ​​of the states of the system defined by the equation [Math.25], where the derivatives are zero, which amounts to solving the following system of equations: − C f + C r mV − 1 + C r l r − C f l f mV 2 0 0 0 C f mV 0 0 − C f l f − C r l r J − C r l r 2 + C f l f 2 JV 0 0 0 C f l f J 0 0 0 1 0 0 0 0 0 0 V l s V 0 0 0 − 1 0 0 0 0 0 − 2 ξω − ω 2 0 ω 2 0 0 0 0 1 0 0 0 0 0 0 0 0 0 − ω f 0 k β vαθ k r vαθ k ψ L vαθ k e y L vαθ k δ ˙ vαθ k δ vαθ k y ˙ L _ ref vαθ − υαθ 1 − k δ sat β static r static ψ L static e y L static δ ˙ static δ static y ˙ L _ ref static δ ref static + 0 0 0 0 0 0 0 vαθ δ Ffwd + 0 0 − V 0 0 0 0 0 ρ ref + ε 1 C f + ε 2 C r mv ε 1 C f l f − ε 2 C r l r J 0 0 0 0 0 0 ϕ = 0

[0143] Solving this system of equations above allows us to calculate the static trajectory tracking error e yL, as a function of the curvature ρ ref of the road, the speed V of the motor vehicle 10, the roll angle Φ, the controller gains K Fbck , the second component δ Ffwd sought, and other parameters relating to the vehicle itself (lf, lr, ε1, ε2...). We can thus write: e y L static = f δ Ffwd ϕ V ρ ref

[0144] So, to find the second component δ Ffwd allowing to obtain a zero static trajectory tracking error e yL, we can write the equation below: e y L static = 0

[0145] Which allows us to obtain the following equation. δ Ffwd = l f + l r 1 − k δ − k δ ref + k ψ L − k β l r + k ψ L l s − k r V + K V V 2 . ρ ref − 1 − k δ − k δ ref ε 1 − 1 − k δ − k δ ref + k ψ L − k β ε 2 . ϕ

[0146] In this equation, the term K v is called the “understeer gradient” and is expressed as: K V = l r 1 − k δ − k δ ref C f − l f 1 − k δ − k δ ref + k ψ L − k β C r . m l f + l r

[0147] This understeer gradient is a coefficient which makes it possible to express and quantify the tendency of the motor vehicle to understeer in a bend at a given speed.

[0148] In summary, to obtain the second component δ Ffwd , it is therefore sufficient to calculate the term Kv using equation [Math.30] (block Z2 of the figure 5 ), then use equation [Math.29] (block Z3).

[0149] At this stage, we can then describe the method which will be executed by the computer 13 of the motor vehicle 10 to implement the invention.

[0150] The computer 13 is programmed here to implement this process recursively, that is to say step by step, and in a loop.

[0151] To do this, during a first step, the computer 13 checks that the autonomous obstacle avoidance (AES) function is activated.

[0152] If this is the case, it attempts to detect the presence of a possible obstacle located in the path of the motor vehicle 10. It uses its RADAR or LIDAR remote sensor for this.

[0153] In the absence of an obstacle, this step is repeated in loops.

[0154] As soon as a potentially dangerous obstacle 20 is detected (see figure 2 ), the computer 13 plans an avoidance trajectory T0 allowing this obstacle 20 to be avoided.

[0155] The computer 13 will then seek to define a control instruction for the power steering actuator 31, namely the saturated steering angle instruction δ ref , making it possible to follow this avoidance trajectory T0 as best as possible.

[0156] To do this, it begins by calculating or acquiring parameters such as: the measured steering angle δ, the drift angle β of the motor vehicle 10, the derivative with respect to time of the measured steering angle δ, the saturated steering angle setpoint δ ref obtained at the previous time step, the yaw rate r, the relative heading angle Ψ L , and the lateral deviation setpoint y L-ref .

[0157] As shown in the figure 5 , the computer 13 then calculates the gains of the controller K Fbck , then it deduces the value of the first component δ Fbck of the unsaturated steering angle setpoint δ K .

[0158] It also deduces the values ​​of the term K v then of the second component δ Ffwd of the unsaturated steering angle setpoint δ K .

[0159] It can then deduce the value of this unsaturated steering angle setpoint δ K then that of the saturated steering angle setpoint δ ref.

[0160] The latter is then transmitted to the power steering actuator 31 in order to deviate the vehicle from its initial trajectory.

[0161] The present invention is in no way limited to the embodiment described and shown, but those skilled in the art will be able to provide any variation in accordance with the invention.

Claims

1. Control method for the automated steering of a motor vehicle (10) comprising wheels (11, 12) of which at least two wheels (11) are steered wheels, comprising steps of: - acquiring parameters (β, r, ΨL, yL, δ, δref) relating to an avoidance path whereby the motor vehicle (10) avoids an obstacle (20), and of - calculating, using a computer (13) a temporary control instruction (δFbck) for a steering actuator (31) that steers the steered wheels (11), as a function of said parameters (β, r, ΨL, yL, δ, δref) and by means of a closed loop controller (KFbck), characterized in that it further comprises steps of: - acquiring a lateral acceleration (ay) or a roll angle (Φ) experienced by the motor vehicle (10), and - calculating, using open loop calculation, a correction term (δFfwd) for said temporary control instruction (δFbck), as a function of the lateral acceleration (ay) or of the roll angle (Φ) acquired.

2. Control method according to the preceding claim, wherein the correction term (δFfwd) is calculated as a function of a coefficient of steering (ε1) induced by the rolling of the motor vehicle (10) on the front wheels (11) of the motor vehicle (10) and / or of a coefficient of steering (ε2) induced by the rolling of the motor vehicle (10) on the rear wheels (12) of the motor vehicle (10).

3. Control method according to one of the preceding claims, wherein the correction term (δFfwd) is calculated as a function of the curvature (ρref) of the road taken by the motor vehicle (10).

4. Control method according to the preceding claim, wherein the correction term (δFfwd) is equal to the sum of: - the product of a first variable multiplied by the lateral acceleration (ay) or by the roll angle (Φ), and - the product of a second variable multiplied by the curvature (ρref) of the road.

5. Control method according to one of the preceding claims, wherein the correction term (δFfwd) is calculated as a function of at least one of the gains of the closed loop controller (KFbck).

6. Control method according to one of the preceding claims, wherein the correction term (δFfwd) is calculated as a function of an understeer gradient (Kv).

7. Control method according to the preceding claim, wherein the understeer gradient (Kv) is calculated as a function of at least one of the gains of the closed loop controller (KFbck).

8. Control method according to the preceding claim, wherein the understeer gradient (Kv) is calculated by means of the following equation: K V = l r 1 − k δ − k δ ref C f − l f 1 − k δ − k δ ref + k ψ L − k β C r . m l f + l r , where m, Cf, Cr, lf, lr are parameters connected only with the architecture of the motor vehicle, and where kδ, kδref, kΨL and kβ are gains of the closed loop controller.

9. Control method according to the preceding claims, wherein the correction term (δFfwd) is calculated by means of the following equation: δ Ffwd = l f + l r 1 − k δ − k δ ref + k ψ L − k β l r + k ψ L l s − k r V + K V V 2 . ρ ref − 1 − k δ − k δ ref ε 1 − 1 − k δ − k δ ref + k ψ L − K β ε 2 . ϕ where: lf, lr are parameters connected only with the architecture of the motor vehicle, V is the longitudinal velocity of the motor vehicle, ls is a predetermined sighting distance, ρref is the radius of curvature of the road at the motor vehicle, ε1 is the coefficient of steering induced by the rolling on the front wheelset, ε2 is the coefficient of steering induced by the rolling on the rear wheelset, Φ is the roll angle, Kv is the understeer gradient, kδ, kδref, kΨL, kr, and kβ are gains of the closed loop controller.

10. Motor vehicle (10) comprising a steering actuator (31) for steering the steered wheels (11) and a computer (13) for controlling said actuators (31), characterized in that the computer (13) is programmed to implement a method according to one of the preceding claims.