Method for controlling trajectory of vehicle for obstacle avoidance

Abstract

The invention relates to a method (1) for controlling the trajectory of a vehicle when approaching an obstacle, comprising a step (12) of controlling the steering angles of the wheels, carrying out a calculation (I) of a yaw rate setpoint of the vehicle and a loop (II) of controlling the steering angles of the wheels as a function of the calculated yaw rate setpoint, the step (12) of controlling the steering angle of the wheels uses the longitudinal speed of the vehicle, the control method (1) being characterized in that the control method further comprises a step (10) of controlling the speed of the vehicle, and in that the longitudinal speed of the vehicle is a set longitudinal speed (Vox) determined during the step (12) of controlling the speed of the vehicle.

Description

[0001] The present invention relates to the field of the motor vehicle industry and more precisely to a method for controlling the path of an autonomous vehicle or a vehicle equipped with an advanced driver assistance system.

[0002] These vehicles are equipped with sensors that allow them to capture their surroundings, as well as actuators that allow them to control the vehicle's transmission and steering without driver intervention, these actuators themselves being controlled by one or more computers that are usually embedded in the vehicle. Thus, such vehicles manage their path autonomously, at least in certain driving situations.

[0003] Such vehicles must handle often complex scenarios. In order for the vehicle’s computer to make and execute the best decision about the path to follow (for example, in terms of steering wheel angle or vehicle speed), it is necessary for multiple software modules to interact and fuse the information delivered by these modules in real time.

[0004] One of the recurring and complex problems such vehicles must solve is obstacle avoidance at high speeds. Specifically, this problem requires interaction between the vehicle's lateral control system, which manages the steering angle of the vehicle's wheels, and the vehicle's longitudinal control system, which manages the vehicle's speed. To solve this problem, integrating the data from these systems results in nonlinear equations that are difficult to solve in real time to simultaneously adjust the vehicle's speed and the steering angle of its wheels.

[0005] Existing control solutions for managing obstacle avoidance maneuvers are mainly based solely on the regulation of the vehicle's steering, which dynamically modifies the vehicle's yaw rate towards a new reference path generated by the vehicle's computer, depending on the vehicle's measured longitudinal speed.

[0006] Because the generation of the yaw rate setpoint is closely related to the vehicle's measured longitudinal velocity, during an obstacle avoidance maneuver at high speeds, the vehicle will reach its physical limits and will not be able to turn as quickly as requested and achieve the yaw rate setpoint. This can result in vehicle oscillations or even instability, even if the vehicle eventually manages to avoid the obstacle and reach a new reference path.

[0007] These solutions do not take into account the longitudinal evolution of the vehicle when calculating its yaw rate, nor the physical limitations of the vehicle's braking and steering systems, which may result in the vehicle being unable to satisfactorily perform avoidance maneuvers.

[0008] The present invention aims to at least partially remedy the shortcomings of the prior art by providing a method for controlling the path of a vehicle approaching an obstacle, wherein the vehicle's longitudinal speed setpoint and yaw rate setpoint are optimized such that the physical limitations of the vehicle's steering system can be respected at high speeds, thereby allowing the vehicle to avoid the obstacle via rapid maneuvers without instability.

[0009] To this end, the invention provides a method for controlling the path of a vehicle approaching an obstacle, the method comprising a step of controlling the steering angles of the wheels, the step implementing calculation of a yaw rate setpoint for the vehicle and a circuit for controlling the steering angles of the wheels as a function of the calculated yaw rate setpoint, the step of controlling the steering angles of the wheels using the longitudinal speed of the vehicle,

[0010] The control method is characterized in that the control method further comprises the step of controlling the speed of the vehicle, and the control method is characterized in that the longitudinal speed of the vehicle is a set-point longitudinal speed determined in the step of controlling the speed of the vehicle.

[0011] The present invention solves the complex obstacle avoidance problem simply by decoupling the step of controlling the vehicle's speed, on the one hand, and the step of controlling the steering angle of the vehicle's wheels, on the other. The latter step uses only one exogenous parameter generated during the step of controlling the vehicle's speed. Specifically, these two steps are executed in parallel, and the step of controlling the steering angle of the vehicle's wheels has no effect on the step of controlling the vehicle's speed. Instead, only the vehicle's setpoint longitudinal speed influences the step of controlling the steering angle of the vehicle's wheels. In this patent application, the longitudinal direction of the vehicle is oriented parallel to the vehicle's length. By utilizing this decoupling, each step can take into account vehicle-specific limitations of the braking system or steering system.

[0012] According to an advantageous feature of the path control method according to the present invention, the step of controlling the vehicle's speed includes a sub-step of optimizing an objective function taking into account the distance, calculated as a function of the measured distance between the vehicle and the obstacle, on the one hand, and the vehicle's yaw angle, on the other. This optimization sub-step is capable of providing a setpoint longitudinal acceleration for the vehicle, the integration of which provides the setpoint longitudinal velocity. This feature incorporates information provided by the vehicle's sensors into the control loop. Specifically, the setpoint longitudinal velocity takes this information into account and requires the vehicle to decelerate when approaching an obstacle, allowing the yaw rate to be adjusted below the setpoint longitudinal velocity of the prior art and to avoid the obstacle without oscillation. The calculated distance is based on the distance measured for at least a certain period of time before the obstacle is circumvented. The setpoint longitudinal acceleration is a value that can be positive or negative, i.e., it can be a longitudinal acceleration or deceleration setpoint.

[0013] In one embodiment of the present invention, the calculated distance depends on the minimum safe distance between the vehicle and the obstacle to be circumvented. Thus, obstacles are avoided with a configurable safety margin.

[0014] Advantageously, the calculated distance increases with increasing yaw angle, at least until the vehicle's yaw angle allows the obstacle to be circumvented. Thus, as the vehicle gets closer to its target of avoiding the obstacle, braking is reduced, allowing the vehicle to change direction to follow the new path without excessive deceleration.

[0015] In one embodiment of the invention, the optimization sub-step satisfies a constraint according to which the distance traveled by the vehicle during a predetermined number of calculation increments must be less than said calculated distance. This embodiment of the optimization sub-step allows for a configurable adaptation of the braking distance in real time.

[0016] According to an advantageous feature of the path control method according to the present invention, once the yaw angle allows for circumventing the obstacle, the calculated distance is updated, according to a predefined driving mode selection, so that it no longer depends on the measured distance between the vehicle and the obstacle. Thus, once the vehicle has steered to avoid an obstacle, the vehicle's deceleration is configurable, depending on a selected driving style, such as that selected by the vehicle's user. In particular, the user can select the speed at which the complete obstacle avoidance maneuver is executed by selecting a fast or sporty mode, or a mode that delivers a safer and smoother experience.

[0017] For example, once the yaw angle allows the obstacle to be circumvented, the calculated distance is updated so that the objective function is no longer constrained, or the objective function is constrained by another obstacle in the vehicle's path than the obstacle to be circumvented. As a variant, the constraint on the calculated distance is removed in the sub-step of optimizing the objective function. This removal of the dependency of the setpoint longitudinal speed on the calculated distance allows the vehicle to recover speed once it is no longer turning toward the obstacle and reach its new path very quickly.

[0018] Alternatively, the vehicle's environment is divided orthogonally to the vehicle's initial path into a first region that excludes the obstacle and is located between the vehicle and a first end of the obstacle, and a second region that includes the obstacle and is located between the first end and the second end of the obstacle. As long as the vehicle is in the first region, once the yaw angle allows the obstacle to be circumvented, the calculated distance is updated so that the calculated distance depends only on the yaw angle and the distance between the vehicle and the target path. In this alternative, the vehicle only actually regains speed once it begins to pass the obstacle, resulting in smoother maneuvering than in the previous example.

[0019] In yet another alternative, the vehicle's environment is divided orthogonally to the vehicle's initial path into a first region that excludes the obstacle and is located between the vehicle and a first end of the obstacle, and a second region that includes the obstacle and is located between the first end and the second end of the obstacle. As long as the vehicle is in either the first or second region, once the yaw angle allows for circumventing the obstacle, the calculated distance is updated so that the calculated distance depends solely on the yaw angle and the distance between the vehicle and the target path. In this further alternative, the vehicle only actually regains speed once it has passed the obstacle, making its handling even more comfortable and smoother than in the previous example.

[0020] The invention also relates to a computer program comprising program code instructions for carrying out the steps of the control method according to the invention when the program is executed on one or more computers of a vehicle. The computer program according to the invention has advantages similar to those of the control method according to the invention.

[0021] Other characteristics and advantages of the invention will become more apparent from the following description on the one hand and from a number of non-limiting examples of embodiments given by way of indication with reference to the accompanying drawings on the other hand, in which:

[0022] [ Figure 1 ] schematically illustrates a coordinate system for positioning a vehicle implementing a control method according to the invention in one embodiment of the invention, these coordinate systems being used by the control method according to the invention,

[0023] [ Figure 2 ] shows the devices and steps used or implemented by the control method according to the present invention in this embodiment of the present invention,

[0024] [ Figure 3 ] shows how Figure 2 The control method models the vehicle environment,

[0025] [ Figure 4 ] shows the path taken by a vehicle without implementing the control method according to the invention during an obstacle avoidance maneuver,

[0026] [ Figure 5 ] shows that during this obstacle avoidance maneuver Figure 4 The yaw rate of the vehicle,

[0027] [ Figure 6 ] shows the Figure 1 Implementation Figure 2 The path taken by the vehicle of the control method,

[0028] [ Figure 7 ] shows Figure 1 The vehicles in Figure 6 the longitudinal speed on the path taken by the vehicle as shown,

[0029] [ Figure 8 ] shows Figure 1 The vehicles in Figure 6 the steering wheel angle on the path taken by the vehicle as shown, and

[0030] [ Figure 9 ] shows Figure 1 The vehicles in Figure 6 The lateral acceleration along the path taken by the vehicle is shown.

[0031] In one embodiment of the present invention, vehicle 2 (in Figure 1 ) implements the path control method 1 according to the invention (see Figure 2 ), and has a center of gravity G. Its position is fixed relative to the vehicle 2 and in which the yaw rate of the vehicle is measured The axis (OX) is parallel to the vehicle's initial path, on which the obstacle Obs is found (see Figure 3 ).

[0032] In the so-called "bicycle" model, the rear wheels of vehicle 2 are combined into a single wheel Wr, and the front wheels of vehicle 2 are combined into a single wheel Wf. An orthogonal coordinate system (G, x, y) attached to vehicle 2 includes a longitudinal axis (Gx) that passes through the center of gravity G of vehicle 2 and through the centers of the combined wheels Wr and Wf, and is oriented toward the front of vehicle 2. This longitudinal axis (Gx) is referred to as the longitudinal axis of vehicle 2. The orthogonal coordinate system (G, x, y) includes a transverse axis (Gy) that passes through the center of gravity G of the vehicle, is parallel to the vehicle floor, and is orthogonal to the longitudinal axis (Gx) of vehicle 2. Here, this transverse axis (Gy) points to the left of vehicle 2 and is referred to as the transverse axis of vehicle 2.

[0033] In the orthogonal coordinate system (G, x, y) attached to the vehicle 2, the measured speed of the vehicle 2 is (relative to a fixed orthogonal coordinate system (O, X, Y)) along the longitudinal axis (Gx) is resolved into the longitudinal velocity , and is resolved into the lateral velocity along the lateral axis (Gy) of vehicle 2 The front wheels Wf form a steering angle δ with the longitudinal axis (Gx) of the vehicle 2. This steering angle δ is the steering angle of the wheels of the vehicle 2, indicating the direction taken by them at a given time.

[0034] like Figure 2 As shown, vehicle 2 includes means 20 for perceiving its immediate surroundings. These perceiving means 20 include one or more sensors, such as radar, LiDAR (LiDAR is an acronym for Light Detection and Ranging), or a camera. Perceiving means 20 derives a measured distance Dobs between vehicle 2 and obstacles Obs located in the initial path of vehicle 2 from these sensors.

[0035] The vehicle 2 further comprises a positioning device 21 , such as a GPS module (GPS is the acronym for Global Positioning System), which is able to provide the position (X, Y) of the vehicle 2 in a fixed orthogonal coordinate system (O, X, Y).

[0036] The vehicle 2 also includes a measured yaw rate of the vehicle 2 The gyroscope 25 and the measured speed of the vehicle 2 are provided Speed sensor 26.

[0037] In this embodiment of the invention, the control method 1 is implemented in a computer of the vehicle 2, which receives, for example via a central CAN bus (CAN is the acronym for Controller Area Network) of the vehicle 2, the measured distance Dobs to the obstacle Obs, the position (X, Y) of the vehicle 2 in a fixed orthogonal coordinate system (O, X, Y), the measured yaw rate of the vehicle 2 and the measured speed , these data are updated in real time by various devices and sensors of vehicle 2.

[0038] The computer of the vehicle 2 comprises a navigation module 22 which generates a new path Tr to be followed by the vehicle 2 based on the position (X, Y) of the vehicle 2 and based on the measured distances Dobs to the obstacles Obs (see Figure 3 ) and the reference speed of the curve following this new path Tr and the reference yaw angle ψr.

[0039] The computer of the vehicle 2 also comprises a module 27 for estimating the lateral state of the vehicle 2. This estimation module 27 estimates the lateral velocity of the vehicle 2 in the orthogonal coordinate system (G, x, y) of the vehicle 2 , and uses the model to provide, in particular, the yaw rate based on the measured Corrected yaw rate c.

[0040] The computer of the vehicle 2 implements a step 10 of controlling the speed of the vehicle 2 using a longitudinal planner 23 which optimizes a reference speed according in particular to the physical characteristics of the braking, propulsion and transmission systems of the vehicle 2 and the measured distance Dobs to the obstacle Obs. The longitudinal planner 23 delivers the setpoint longitudinal speed Vox as output. Figure 3 Describe in detail how to obtain this set point longitudinal velocity.

[0041] The step 10 of controlling the speed of the vehicle also uses a longitudinal controller 24 which applies a setpoint acceleration or deceleration ax_cons to the actuators of the vehicle 2 in order to cause them to achieve a setpoint longitudinal speed Vox.

[0042] In parallel with this step 10 of controlling the speed of the vehicle 2, the computer of the vehicle 2 implements a step 12 of controlling the yaw rate of the vehicle 2 using a lateral planner 28, the operation of which will be described below with reference to Figure 3 The lateral planner 28 receives the lateral velocity of the vehicle 2 estimated by the estimation module 27. , the corrected yaw rate provided by the estimation module 27 c, a reference yaw angle ψr provided by the navigation module 22, and a setpoint longitudinal velocity Vox delivered by the longitudinal planner 23. The lateral planner 28 uses these inputs to provide a yaw rate setpoint o. This set point takes into account the physical limitations of the vehicle 2's steering system.

[0043] The step 12 of controlling the yaw rate of the vehicle 2 also uses a lateral controller 29 which adjusts the yaw rate according to a yaw rate setpoint. o Implementing a circuit for controlling the setpoint δcons of the steering angle of the wheels, which is sent to the actuators of the vehicle 2 in order to make them achieve the yaw rate setpoint o.

[0044] In step 10 of controlling the speed of the vehicle 2 , the longitudinal planner 23 determines the calculated distance Dmax (see Figure 3 ):

[0045] Dmax = Dobs – Ds, where Ds is the safety distance in meters around the obstacle Obs.

[0046] Next, the longitudinal planner 23 determines the variable Ytr (see Figure 3 ), such that:

[0047] Ytr = Wobs – Dmax / tan(ψ), where Wobs is the dimension of the obstacle orthogonal to the axis OX and therefore to the initial path of the vehicle 2, and tan(ψ) is the tangent of the yaw angle ψ of the vehicle measured by the module 27 for estimating the lateral state of the vehicle 2, this yaw angle ψ representing the change in direction of the vehicle 2 relative to the axis (OX) of the fixed orthogonal coordinate system (O, X, Y) and therefore relative to its initial path.

[0048] If the variable Ytr has a negative value, this means that the vehicle 2 is in such a direct direction towards its new path Tr that the vehicle 2 should not encounter the obstacle Obs. In contrast, if the variable Ytr has a strictly positive value, this is because the vehicle 2 is in such a direct direction towards its new path Tr that the vehicle 2 will encounter the obstacle Obs if it maintains this direction. Figure 3 In , the critical point Pc corresponds to the zero value of the variable Ytr and therefore to the target point that the longitudinal axis (Gx) of the vehicle 2 must strive to reach when turning towards its new path Tr.

[0049] The longitudinal planner 23 then modifies the value of the calculated distance Dmax as a function of the variable Ytr and of the driving mode selected, for example, by the user of the vehicle 2 .

[0050] In the Sport First driving mode, when the variable Ytr has a strictly positive value, the longitudinal planner 23 increases the calculated distance Dmax as a function of the vehicle's yaw angle ψ, and more precisely inversely proportional to the cosine of this measured yaw angle ψ:

[0051] Dmax = Dmax / cos(ψ)

[0052] In other words, the calculated distance Dmax is set not to be equal to the actual distance Dobs to the obstacle Obs, but to be equal to the distance the vehicle travels to the obstacle Obs in a constant vehicle direction with the yaw angle ψ measured when processed by the longitudinal planner 23, with the safety distance (Ds / cos(ψ)) subtracted from the travel distance.

[0053] In contrast, when the variable Ytr has a negative value, in this sporty first driving mode, the longitudinal planner 23 assigns a very large value, for example one hundred meters, to the calculated distance Dmax, so that braking of the vehicle 2 is no longer necessary:

[0054] Dmax = 100.

[0055] Because the distance to the obstacle, Dobs, is updated in real time, the calculated distance, Dmax, is also updated in real time before being modified by the longitudinal planner 23 based on the variable, Ytr. Since the calculated distance, Dmax, is used to determine the setpoint longitudinal speed, Vox, the smaller the calculated distance, Dmax, the more braking the vehicle 2 will experience. In this sport-first driving mode, the modifications made by the planner 23 cause the vehicle 2 to brake until it steers to avoid obstacle, Obs. Then, when the longitudinal axis (Gx) of the vehicle 2 reaches the target point, Pc, the vehicle is no longer forced to brake due to obstacle, Obs, but may instead be forced to brake due to other path constraints.

[0056] According to the smooth second driving mode, three different regions are distinguished through which vehicle 2 moves during its obstacle avoidance maneuver. The first region z1 extends parallel to vehicle 2's initial path in the fixed orthogonal coordinate system (O, X, Y) between vehicle 2's initial position and obstacle Obs. The second region z2 extends parallel to vehicle 2's initial path in the fixed orthogonal coordinate system (O, X, Y) over the entire dimension Lobs of obstacle Obs parallel to the axis (OX) (i.e., between the first end of obstacle Obs, located on a straight line perpendicular to vehicle 2's initial path and closest to vehicle 2, and the second end of obstacle Obs, located on a straight line perpendicular to vehicle 2's initial path and farthest from vehicle 2). The third region z3 extends parallel to vehicle 2's initial path in the fixed orthogonal coordinate system (O, X, Y) from the second end of obstacle Obs.

[0057] In the stationary second driving mode, when the variable Ytr has a strictly positive value, the longitudinal planner 23 also increases the calculated distance Dmax as a function of the vehicle's yaw angle ψ in a manner inversely proportional to the cosine of this measured yaw angle ψ:

[0058] Dmax = Dmax / cos(ψ)

[0059] In contrast, when the variable Ytr has a negative value, in this stationary second driving mode, as long as the vehicle is in the zone z1, the longitudinal planner 23 gives the calculated distance Dmax a value that no longer depends on the measured distance Dobs to the obstacle, but on the measured yaw angle ψ and the distance Ycg between the vehicle and the new path Tr of the vehicle 2, which is the target path:

[0060] Dmax =Ycg / sin(ψ)

[0061] It should be noted that the target path Tr is parallel to the vehicle's initial path, but is laterally offset relative to the initial path.

[0062] Since Dmax is set equal to the distance Ycg between vehicle 2 and the new path Tr of vehicle 2 divided by the sine of the measured yaw angle ψ, this means that once vehicle 2 has been steered to avoid obstacle Obs, braking of vehicle 2 depends on the distance vehicle 2 has traveled to the new path Tr in a constant vehicle direction having the yaw angle ψ measured when processed by the longitudinal planner 23. In other words, vehicle 2 continues to brake in zone z1 after having steered to avoid obstacle Obs, allowing the vehicle to steer toward its new path Tr more smoothly than in the first driving mode.

[0063] According to a third driving mode that delivers a safer-feeling experience to the user, when the variable Ytr has a strictly positive value, the longitudinal planner 23 always sets the calculated distance Dmax equal to:

[0064] Dmax = Dmax / cos(ψ)

[0065] In contrast, when the variable Ytr has a negative value, in this third driving mode, as long as the vehicle is located in the zones z1 and z2, the longitudinal planner 23 gives the calculated distance Dmax a value that no longer depends on the measured distance Dobs to the obstacle, but on the measured yaw angle ψ and the distance Ycg between the vehicle and the new path Tr of vehicle 2:

[0066] Dmax =Ycg / sin(ψ)

[0067] In other words, as in the second driving mode, in this third mode, once the vehicle 2 has been steered to avoid the obstacle Obs, braking of the vehicle 2 depends on the distance the vehicle 2 has traveled to the new path Tr in a constant vehicle direction having the yaw angle ψ measured when processed by the longitudinal planner 23. Therefore, the vehicle 2 continues braking in its maneuver to redirect it to the new path Tr until it bypasses the obstacle Obs, making it possible to deliver an experience that feels safer than in the case of the second driving mode.

[0068] The calculated distance Dmax, modified by the longitudinal planner 23, is used by the latter in a sub-step of optimizing the objective function J, which delivers as output the vehicle's setpoint longitudinal acceleration (or deceleration) ax_cons, the integration of which provides the setpoint longitudinal velocity Vox. This is therefore a problem of predictive control of the vehicle's longitudinal velocity subject to given constraints. Naturally:

[0069] and

[0070] in, is the instantaneous longitudinal acceleration of vehicle 2 along the longitudinal axis (Gx) of vehicle 2, is the instantaneous longitudinal velocity of vehicle 2 along the longitudinal axis (Gx) of vehicle 2, and is the distance traveled by vehicle 2 from its initial position (e.g., the initial position set when vehicle 2 is on its initial path). These data are discretized into computational increments and expressed as 、 and to specify their values at any given computation increment k.

[0071] Predictive control consists in finding the longitudinal acceleration that minimizes the objective function J at each calculation increment k. :

[0072]

[0073] The constraints are:

[0074]

[0075]

[0076]

[0077] in:

[0078] - is the maximum acceleration value that is not exceeded,

[0079] - is the reference longitudinal speed of the vehicle 2, which corresponds to the reference speed provided by the navigation module 22 projection onto the longitudinal axis (Gx) of vehicle 2, and

[0080] - It is the maximum acceleration increment or decrement between the calculated increment k+i and the calculated increment k.

[0081] The number N of calculation increments considered by the objective function J is set by a person skilled in the art, for example after various experiments of the operation of the longitudinal planner 23 on a given braking system or powertrain.

[0082] Then, the longitudinal acceleration of the objective function J is minimized This is considered as the setpoint longitudinal acceleration (or deceleration) ax_cons of the vehicle 2 and allows the determination of the setpoint longitudinal speed Vox by integrating this setpoint. This setpoint longitudinal speed Vox is then input into the lateral planner 28.

[0083] In step 12 of controlling the steering angle of the vehicle's wheels, the lateral planner 28 uses a method for setting a yaw rate according to the yaw rate setpoint. o to control the steering angle δ of the wheels, this yaw rate setpoint being a virtual reference determined in real time by solving an optimization problem such as minimizing the quadratic distance between the previous yaw rate and the yaw rate that makes it possible to reach the new path Tr, subject to constraints related in particular to the steering system of the vehicle 2.

[0084] A paper titled “A Reference Governor approach for Lateral Control of Autonomous Vehicles,” presented by Dimitrios Kapsalis et al. at the IEEE conference ITSC2021 (IEEE is the acronym for the Institute of Electrical and Electronics Engineers, and ITSC is the acronym for the International Conference on Intelligent Transportation Systems) in September 2021, describes a similar control procedure that uses the measured longitudinal velocity of vehicle 2.

[0085] In this embodiment of the invention, the step 12 of controlling the steering angle of the wheels uses the method described in the above document, but uses a setpoint longitudinal velocity Vox instead of the measured longitudinal velocity in the lateral control loop. Thus, the yaw rate setpoint Vox o The longitudinal approach strategy of the vehicle 2 implemented by the longitudinal planner 23 is taken into account, which makes the control method 1 according to the invention more efficient.

[0086] To illustrate this improvement over the prior art, tests were conducted on two ego vehicles, one without implementing the control method 1 according to the present invention, and one on a vehicle 2 that did. During these tests, the vehicles performed obstacle avoidance maneuvers under identical initial conditions. Each ego vehicle, at a speed of 14 m / s (meters per second), encountered an obstacle positioned 17 meters away, 3.2 meters laterally from the new lane to which it must move to avoid the obstacle. The new lane corresponded to the new or target path T. The minimum safety distance was set to 3.2 meters in front of the obstacle.

[0087] Figure 4 shows a path 3(t) taken by an autonomous vehicle that does not implement the control method 1 according to the invention, and Figure 5 shows the reference yaw rate obtained from the reference yaw angle provided by the navigation module of the autonomous vehicle , the measured yaw rate of the ego vehicle and the yaw rate setpoint given to the steering system of the autonomous vehicle .

[0088] These data show that the navigation module of the autonomous vehicle gives a reference yaw angle (the measured yaw rate) that the autonomous vehicle cannot follow. Not following the reference yaw rate ), because the navigation module's request exceeds the capabilities of the ego vehicle's steering system. This is because the vehicle's longitudinal velocity is very high and affects the reference yaw rate, causing it to change too rapidly for the ego vehicle's steering system to follow. This is why the ego vehicle oscillates around its target path T and exhibits instability.

[0089] Various driving modes were tested with a vehicle 2 implementing the control method 1 according to the invention. Figure 6 Shown are the path 2_1(t) followed by vehicle 2 during the obstacle avoidance maneuver when the sporty first driving mode is selected, the path 2_2(t) followed by vehicle 2 during the obstacle avoidance maneuver when the smooth second driving mode is selected, and the path 2_3(t) followed by vehicle 2 during the obstacle avoidance maneuver when the third, feel-safer driving mode is selected.

[0090] same, Figure 7 、 Figure 8 and Figure 9 The longitudinal speed vx_1(t), the steering angle δ_1(t) of the wheels, and the lateral acceleration ay_1(t) of the vehicle 2 during the obstacle avoidance maneuver when the first sport driving mode is selected, the longitudinal speed vx_2(t), the steering angle δ_2(t) of the wheels, and the lateral acceleration ay_2(t) of the vehicle 2 during the obstacle avoidance maneuver when the second smooth driving mode is selected, and the longitudinal speed vx_3(t), the steering angle δ_3(t) of the wheels, and the lateral acceleration ay_3(t) of the vehicle 2 during the obstacle avoidance maneuver when the third, safer-feeling driving mode is selected are respectively shown. The steering angle of the wheels is Figure 8 The unit of temperature is degrees.

[0091] These data show that when the lateral planner 28 uses the setpoint longitudinal speed Vox, in particular for determining the yaw rate setpoint At 0, vehicle 2 reaches target path T without oscillations and without instability. This is because the speed of vehicle 2 decreases before the obstacle, which allows lateral planner 28 to generate a yaw rate setpoint that remains within the physical limitations of the steering system of vehicle 2. Specifically, the longitudinal speed and lateral acceleration of the vehicle decrease from the first second. These figures also show that the first driving mode corresponds to Figures 6 to 9 The fastest of the three maneuvers shown, and the third driving mode corresponds to the smoothest and most comfortable of the three.

[0092] Of course, the invention is not limited to the examples that have just been described and many modifications can be made to these examples without departing from the scope of the invention.

Claims

1. A method (1) for controlling the path of a vehicle (2) approaching an obstacle (Obs), the method comprising the step (12) of controlling the steering angle (δ) of the wheels, which step implements a yaw rate setpoint ( ) and for the calculation of the yaw rate set point ( ) to control the steering angles of the wheels, the step (12) of controlling the steering angles (δ) of the wheels using the longitudinal speed of the vehicle (2), The control method (1) is characterized in that the control method further comprises a step (10) of controlling the speed of the vehicle (2), and the control method is characterized in that the longitudinal speed of the vehicle (2) is a set point longitudinal speed (Vox) determined in the step (12) of controlling the speed of the vehicle (2).

2. The control method (1) according to claim 1, wherein: The step (10) of controlling the speed of the vehicle (2) comprises a sub-step of optimizing an objective function (J) taking into account a distance (Dmax), the calculation of which depends on the one hand on the measured distance (Dobs) between the vehicle (2) and the obstacle (Obs) and on the other hand on the yaw angle (ψ) of the vehicle (2), the optimization sub-step being able to provide a setpoint longitudinal acceleration (ax_cons) of the vehicle (2), the integration of which provides said setpoint longitudinal speed (Vox).

3. The control method (1) according to claim 2, wherein: The calculated distance (Dmax) depends on the minimum safe distance (Ds) between the vehicle (2) and the obstacle (Obs) to be circumvented.

4. The control method (1) according to any one of claims 2 and 3, wherein: At least until the yaw angle (ψ) of the vehicle (2) allows the obstacle (Obs) to be bypassed, the calculated distance (Dmax) increases as the yaw angle (ψ) increases.

5. The control method (1) according to any one of claims 2 to 4, wherein: The optimization sub-step satisfies a constraint according to which the distance traveled by the vehicle (2) during a predetermined number of calculation increments must be less than said calculated distance (Dmax).

6. The control method (1) according to any one of claims 2 to 5, wherein: According to a predefined driving mode selection, once the yaw angle (ψ) allows to circumvent the obstacle (Obs), the calculated distance (Dmax) is updated so that the calculated distance no longer depends on the measured distance (Dobs) between the vehicle (2) and the obstacle (Obs).

7. The control method (1) according to claim 5 or 6, wherein: As soon as the yaw angle (ψ) allows the obstacle (Obs) to be bypassed, the calculated distance (Dmax) is updated so as to no longer constrain the objective function (J) or to constrain the objective function according to another obstacle in the path of the vehicle (2) than the obstacle (Obs) to be bypassed.

8. The control method (1) according to claim 5 or 6, wherein: The environment of the vehicle (2) is divided into a first area (z1) not containing the obstacle (Obs) and located between the vehicle (2) and the first end of the obstacle (Obs) and a second area (z2) containing the obstacle (Obs) and located between the first end of the obstacle (Obs) and the second end of the obstacle (Obs) in a manner orthogonal to the initial path of the vehicle (2), and as long as the vehicle (2) is in the first area (z1), once the yaw angle (ψ) allows to bypass the obstacle (Obs), the calculated distance (Dmax) is updated so that the calculated distance depends only on the yaw angle (ψ) and the distance between the vehicle (2) and the target path (Tr) of the vehicle (2).

9. The control method (1) according to claim 5 or 6, wherein: The environment of the vehicle (2) is divided into a first area (z1) not containing the obstacle (Obs) and located between the vehicle (2) and the first end of the obstacle (Obs) and a second area (z2) containing the obstacle (Obs) and located between the first end of the obstacle (Obs) and the second end of the obstacle (Obs) in a manner orthogonal to the initial path of the vehicle (2). As long as the vehicle (2) is in the first area (z1) or the second area (z2), once the yaw angle (ψ) allows to bypass the obstacle (Obs), the calculated distance (Dmax) is updated so that the calculated distance depends only on the yaw angle (ψ) and the distance between the vehicle (2) and the target path (Tr) of the vehicle (2).

10. A computer program comprising program code instructions for executing the steps of the control method (1) as claimed in any one of claims 1 to 9 when said program is executed on one or more computers of a vehicle (2).

Images