EP4337505A1

EP4337505A1 - Method for managing the longitudinal speed of an autonomous vehicle

Info

Publication number: EP4337505A1
Application number: EP22728827.1A
Authority: EP
Inventors: Marouane BENAZIZ; Pedro KVIESKA; Sébastien Saliou; Antoine SIMONIN
Original assignee: Renault SAS
Current assignee: Ampere Sas
Priority date: 2021-05-11
Filing date: 2022-05-11
Publication date: 2024-03-20
Also published as: CN117320942A; WO2022238497A1; FR3122856A1

Abstract

Method for managing the longitudinal speed of an autonomous vehicle (100), the autonomous vehicle travelling on a traffic lane comprising stop signage located in front of the autonomous vehicle, the autonomous vehicle being equipped with a first means of detecting (1) a first range (P1) and a second means of detecting (2) a second range (P2), the first range (P1) being greater than the second range (P2). Said method is characterized in that it comprises - a first step (E1) of detecting the stop signage by the first detection means (1), and of implementing a first deceleration logic of the autonomous vehicle, - a second step (E2) of detecting the stop signage by the second detection means (2), and of implementing a second deceleration logic of the autonomous vehicle, in that the first and second deceleration logic implement jerks, the absolute value of which is less than a first threshold (JMAX), and in that the second deceleration logic controls the stopping of the autonomous vehicle with an accuracy of around one centimeter relative to the stop signage, or around ten centimeters relative to the stop signage, or around several tens of centimeters relative to the stop signage.

Description

1

DESCRIPTION

TITLE: Method for managing the longitudinal speed of an autonomous vehicle.

The invention relates to a method for managing the longitudinal speed of an autonomous vehicle. The invention also relates to a device for managing the longitudinal speed of an autonomous vehicle. The invention also relates to a computer program implementing the mentioned method. The invention finally relates to a recording medium on which such a program is recorded.

Automated speed management systems are commonly installed on current vehicles, and are being upgraded to incorporate new features.

An evolution concerns the automated management of the stopping of the autonomous vehicle at the level of a road sign, for example a stop sign, or at the level of a traffic light.

This functionality requires very precisely locating the stopping point of the vehicle. To this end, the use of a front camera fitted to the vehicle allows the precise location of the stopping point and therefore the implementation of an automated management of the slowing down and stopping of the autonomous vehicle at the level of the road sign. or traffic light. However, this solution has the disadvantage of being limited by the range of a camera, which is less than 50 meters. In other words, this solution makes it possible to precisely locate a stopping point of the vehicle but only when the latter is located less than 50 meters from the vehicle. Such a delay in anticipation of the detection of an upcoming stop of the vehicle does not make it possible to guarantee braking that satisfies the driving comfort conditions of the autonomous vehicle. 2

The object of the invention is to provide a device and a method for managing the longitudinal speed of an autonomous vehicle remedying the above drawbacks and improving the devices and methods for managing the longitudinal speed known from the prior art. In particular, the invention makes it possible to produce a device and a method which are simple and reliable and which make it possible on the one hand to precisely locate a next stopping point of an autonomous vehicle, and on the other hand to control the stopping of the autonomous vehicle at this next stopping point, while guaranteeing braking comfort.

To this end, the invention relates to a method for managing the longitudinal speed of an autonomous vehicle, the autonomous vehicle circulating on a traffic lane comprising a stop sign located at the front of the autonomous vehicle, the autonomous vehicle being equipped with a first means for detecting a first range and a second means for detecting a second range, the first range being greater than the second range.

The process includes:

- a first step of detecting the stop signs by the first detection means, and implementing a first logic for slowing down the autonomous vehicle,

- a second step of detecting the stop signs by the second detection means, and implementing a second logic for slowing down the autonomous vehicle.

In addition, the first and second deceleration logic implement jerks whose absolute value is less than a first limit threshold, and the second deceleration logic controls the stopping of the autonomous vehicle with an accuracy of the order of a centimeter relative to the stop signs, or around ten 3 centimeters relative to the stop signage, or of the order of several tens of centimeters relative to the stop signage.

The first and the second deceleration logic can implement negative accelerations greater than a second limit threshold.

The first step may comprise a determination, at a first instant, of a first approximate position of the stop sign, and the second determining step may comprise a determination, at a second instant, of a second precise position of the stop sign, the second instant being strictly later than the first instant.

The first deceleration logic can initiate a first deceleration phase when the vehicle arrives at a given distance from the first approximate position of the stop sign.

The second deceleration logic can start a second phase of deceleration at the second instant, and the second phase of deceleration can have continuity of speed and acceleration with the first phase of deceleration.

The absolute jerk value at the end of the first deceleration phase may be greater than the absolute jerk value at the start of the first deceleration phase, and/or the absolute jerk value at the end of the second deceleration phase may be greater than the absolute value of the jerk at the start of the second deceleration phase.

The first deceleration phase may consist of three consecutive sub-phases, a first initial sub-phase having a first non-zero constant jerk, a first intermediate sub-phase having a zero jerk, and a first final sub-phase having a second nonzero constant jerk. Alternatively or in addition, the second phase 4 of deceleration can be composed of three consecutive sub-phases, a second initial sub-phase having a non-zero constant third jerk, a second intermediate sub-phase having zero jerk, and a final second sub-phase having a constant fourth jerk not bad.

The second jerk may be the product of the first jerk by a first multiplicative factor, in particular a first multiplicative factor whose sign is the sign of the product between, on the one hand, the difference between a first acceleration of the end of the first sub- final phase and a second acceleration of the first intermediate sub-phase, and, on the other hand, the difference between the second acceleration and a third acceleration of the start of the first initial sub-phase. Alternatively or in addition, the fourth jerk may be the product of the third jerk by a second multiplicative factor, in particular a second multiplicative factor whose sign is the sign of the product between, on the one hand, the difference between a fourth acceleration of the end of the second final sub-phase and a fifth acceleration of the second intermediate sub-phase, and, on the other hand, the difference between the fifth acceleration and a sixth acceleration of the start of the second initial sub-phase.

The invention further relates to a device for managing the longitudinal speed of an autonomous vehicle, the autonomous vehicle being equipped with a brake actuator. The device comprises hardware and/or software elements implementing the method as defined previously, in particular hardware and/or software elements designed to implement the method according to the invention, and/or the device comprising means of implement the method as defined above.

The invention further relates to an autonomous longitudinal speed management vehicle according to the invention. 5

The invention also relates to a computer program product comprising program code instructions recorded on a computer-readable medium to implement the steps of the method as defined above when said program runs on a computer. The invention also relates to a computer program product downloadable from a communication network and/or recorded on a data carrier readable by a computer and/or executable by a computer, comprising instructions which, when the program is executed by the computer, lead it to implement the method as defined previously.

The invention also relates to a data recording medium, readable by a computer, on which is recorded a computer program comprising program code instructions for implementing the method as defined previously. The invention also relates to a computer-readable recording medium comprising instructions which, when executed by a computer, lead the latter to implement the method as defined previously.

The invention also relates to a signal from a data medium, carrying the computer program product as defined previously.

The appended drawing shows, by way of example, an embodiment of a management device according to the invention and an embodiment of a management method according to the invention.

[Fig. 1] FIG. 1 represents an embodiment of an autonomous vehicle implementing a method for managing the longitudinal speed of an autonomous vehicle.

[Fig. 2] FIG. 2 is a flowchart of an embodiment of a method for managing the longitudinal speed of an autonomous vehicle. 6

[Fig. 3] Figure 3 represents the evolution of the speed of the autonomous vehicle as a function of the distance it has traveled since an instant of detection of a next stop.

[Fig. 4] FIG. 4 represents a speed profile implemented for the braking of the autonomous vehicle according to the first deceleration logic.

[Fig. 5] FIG. 5 is a flowchart of a method for calibrating a speed profile implemented for the braking of the autonomous vehicle according to the first or the second deceleration logic.

[Fig. 6] FIG. 6 represents a speed profile implemented for the braking of the autonomous vehicle according to the second deceleration logic.

The autonomous vehicle 100 can be an autonomous vehicle of any type, in particular a passenger vehicle, a utility vehicle, a truck or even a public transport vehicle such as a bus or a shuttle.

It is assumed that the autonomous vehicle 100 moves on a route comprising an upcoming stop ARR located in front of the autonomous vehicle. In the rest of the document, the term "next stop" is used to designate the sign (for example a stop sign) or a traffic light that can cause the vehicle to stop. If there are several road signs or traffic lights on the considered route, the next stop is the one that the autonomous vehicle will reach first. The term "stop position" refers to the position of the next stop. The stopping position is determined by a means of perception; the precision of the stop position therefore depends on the precision of the means of perception.

The autonomous vehicle 100 comprises a management system 10 and a brake actuator 5 and/or a control unit for a drive motor of the vehicle. The brake actuator 5 and/or the brake unit 7 control of a vehicle drive motor receive commands from the management system 10 in order to implement a slowing down of the autonomous vehicle according to a slowing down logic determined by the management system 10.

The term "deceleration logic" is used in the remainder of the document to designate a mode for determining a profile of longitudinal speed, longitudinal acceleration and longitudinal jerk making it possible to stop the autonomous vehicle at an upcoming stop, the position of the next stop being determined by a detection means. In the remainder of the document, the term “jerk” designates the derivative of the acceleration with respect to time. In particular, the term "longitudinal jerk" refers to the derivative of the longitudinal acceleration with respect to time.

The management system 10 mainly comprises the following elements:

- a first detection means 1,

- a second detection means 2,

- a trajectory planning system 4,

- a set of 6 speed, acceleration and jerk sensors,

- And a calculation unit 3 comprising a microprocessor 31, an electronic memory 32 and communication interfaces 33 allowing the microprocessor 31 to communicate with the detection means 1, 2, the trajectory planning system 4 and the set of sensors 6.

The trajectory planning system 4 determines a trajectory between a starting point and an arrival point of the autonomous vehicle 100. In the rest of the document, the term "trajectory" is used to designate the temporal evolution of a vector state defining the characteristics of the movement of the autonomous vehicle 100. In a preferred embodiment, the state vector comprises a position, in particular coordinates x, y, longitudinal and lateral speeds and/or 8 longitudinal and lateral accelerations and/or a yaw rate and/or a jerk. In the remainder of the document, the term “position” is used to designate either the x, y coordinates of the state vector, or the state vector as a whole.

The first detection means 1 comprises means for detecting a first range P1. The first detection range P1 is a long range, for example it is of the order of several hundred meters, or even of the order of a thousand meters.

In one embodiment, the first detection means 1 comprises a GPS location of the autonomous vehicle 100 on a standard definition map, called SD mapping in the rest of the document. The accuracy of the first detection means 1 is therefore determined by the accuracy of the GPS location which is of the order of a few meters.

The first detection means 1 is capable of receiving data from the trajectory planning system 4. The trajectory planning data make it possible to detect, within the range limit P1 and at a first given instant T 1 , the next stop ARR located on the planned trajectory of the autonomous vehicle 100.

The first detection means 1 further comprises means for calculating a first stopping distance DA1 separating the autonomous vehicle 100 from a first stopping position PA1 associated with the next stop ARR. The distance DA1 is a curvilinear distance corresponding to the length of the path segment delimited by the current position of the autonomous vehicle 100 and by the stop position PA1. The precision of the distance DA1 is of the order of a few meters, for example it is between 3 and 5 meters or between 1 and 10 meters. 9

As a side note, the first detection means 1 does not use high definition mapping.

The second detection means 2 comprises means for detecting a second range P2. The second detection range P2 is a limited range, in particular it is lower or significantly lower than the first detection range P1 of the first detection means 1. For example, the range P2 is of the order of a few tens of meters. On the other hand, the second detection means 2 makes it possible to locate the objects of the driving scene with a level of precision markedly higher than that of the first detection means 1 .

In one embodiment, the second detection means 2 comprises a camera, in particular a front camera, making it possible to locate the autonomous vehicle 100 relative to the elements of the driving scene with an accuracy of the order of a centimeter, or of a ten centimeters or a few tens of centimeters. In this embodiment, the range P2 of the second detection means is of the order of forty or fifty meters, in ideal meteorological and luminosity conditions. The range P2 can also be limited by the road infrastructure—for example when the road makes a sharp bend—or road traffic, for example when the autonomous vehicle 100 is behind a truck.

The second detection means 2 is capable of receiving data from the trajectory planning system 4. The trajectory planning data, compared with the images from the camera, make it possible to detect the next stop ARR within the range limit P2 and at a second given time T2.

The second detection means 2 further comprises means for calculating a second stopping distance DA2 separating the autonomous vehicle 10

100 of a stop position PA2 associated with the next stop ARR. The distance DA2 is a curvilinear distance corresponding to the length of the path segment delimited by the current position of the autonomous vehicle 100 and by the stop position PA2. The precision of the distance DA2 is of the order of ten centimeters, for example it is less than 20 centimeters, or even less than 15 centimeters or 10 centimeters, or even it can be of the order of a centimeter.

As a side note, the second means of detection does not use high definition mapping.

The 6 set of sensors provides the speed, acceleration and jerk of the autonomous vehicle 100 at all times.

In one embodiment, the microprocessor 31 makes it possible to execute software comprising the following modules, which collaborate with each other:

- a module 311 for detecting stop signage by the first detection means, and for implementing a first logic for slowing down the autonomous vehicle, which collaborates with the first detection system 1, the planning system trajectory 4, and the set of sensors 6, and

- a module 312 for detecting stop signage by the second detection means, and for implementing a second logic for slowing down the autonomous vehicle, which collaborates with the second detection system 2, the planning system trajectory 4 and the set of sensors 6.

The autonomous vehicle 100, in particular the automated longitudinal speed management system 10, preferably comprises all the hardware and/or software elements configured so as to implement the method defined in the subject of the invention or the method described below. 11

One mode of execution of the management method is described below with reference to FIG. 2. The method comprises two steps E1 and E2 which are executed successively.

In the first step E1, a stop signal is detected by means of the first detection means 1, then a first logic for slowing down the autonomous vehicle 100 is implemented.

At a first instant T1, the first detection means 1 receives position information POS1 of the autonomous vehicle 100 coming from the GPS. The position POS1 makes it possible to locate the autonomous vehicle 100 on the SD map. By combining the trajectory planning, vehicle position and SD mapping information, a search is made for the presence of stop signage on the route portion located in front of the autonomous vehicle 100 and within the range of the first detection means 1 . The closest stop signage of the autonomous vehicle 100 will be detected as being an upcoming ARR stop of the autonomous vehicle 100. The first detection means 1 thus determines a first position PA1 of the next ARR stop.

Thus, the first step E1 comprises a determination, at a first instant T1, of a first approximate position of the stop signage, that is to say a determination of the first stop position PA1.

We then continue with the implementation of a first logic of slowing down.

To do this, at each time t of reception of a GPS position, the stopping distance DA1(t) is calculated, which is the curvilinear distance separating the autonomous vehicle 100 from the position PA1 at time t. 12

In a preferred embodiment, the first deceleration logic comprises a comparison of the stopping distance DA1 (t) with a maximum distance threshold DMAX in order to determine whether the autonomous vehicle is, at time t, sufficiently close to PA1 stop position to start slowing down.

The threshold DMAX can have a constant value, for example 300 meters. Alternatively, the threshold DMAX can be a variable depending for example on the current speed of the autonomous vehicle, the latter being measured by the set of sensors 6.

The DMAX distance threshold allows the first deceleration logic to be broken down into two phases,

- a braking anticipation phase, starting at the instant T 1 of first detection of the next stop ARR, and during which the distance DA1 (t) between the autonomous vehicle 100 and the next stop ARR is strictly greater than DMAX, and

- a braking phase during which the distance DA1 (t) between the autonomous vehicle 100 and the next stop ARR is less than or equal to DMAX.

Different embodiments of the braking anticipation phase are possible.

Firstly, during the anticipation phase, the autonomous vehicle 100 could keep the trajectory initially planned by the trajectory planning system 4. In other words, the autonomous vehicle 100 could continue its route in the direction of the next stop ARR according to the curves of speeds, accelerations and jerks initially determined by the trajectory planning system 4, and this until it arrives at a distance DMAX from the next stop. 13

In an alternative embodiment, during the anticipation phase, the acceleration of the autonomous vehicle ARR could be limited to a maximum threshold. For example, the acceleration could be reduced to a range of values, called valid, less than or equal to 0 m/s ² . In this case, the range of valid values would be transmitted to the trajectory planning system 4 for the calculation of a new trajectory implementing accelerations in accordance with the range of valid values.

Other embodiments of the anticipation phase are possible, for example the implementation of a deceleration ramp simulating a release of the accelerator pedal by a human driver, the autonomous vehicle 100 simply slowing down due to an engine brake.

Whatever the embodiment of the anticipation phase, the distance DA1 (t) gradually decreases until it becomes less than or equal to the threshold DMAX. The autonomous vehicle 100 then enters the braking and/or deceleration phase of the first deceleration logic.

In the rest of the document, the braking and/or deceleration phase is called “braking phase”.

FIG. 3 comprises four graphs, G1, G2, G3, G4 representing the evolution of the speed of the autonomous vehicle 100 as a function of the distance d(t) that it has traveled since the instant T1 of detection of a next stop ARR.

Graphs G1 , G2 and G3 more specifically illustrate the implementation of the first slowdown logic:

- graph G1 illustrates the course of the anticipation phase,

the graph G2 illustrates the transition between the anticipation phase and the braking phase of the first deceleration logic, and 14

the graph G3 illustrates the course of the braking phase of the first slowing logic and the transition between the first slowing logic and the braking phase of the second slowing logic.

For the sake of clarity, the graphs G1 to G3 illustrate a situation where the position PA1 of the next stop ARR does not change during the implementation of the first deceleration logic. However, it is possible that the processing of the data received from the GPS may change the position PA1 of the next stop. In this case, the anticipation and braking phases are redefined according to the first deceleration logic.

In the graphs G1 to G3, the line 500 located at the abscissa DA1 materializes a finish line of the vehicle on the next stop ARR according to the position PA1 detected by the first detection means 1. The line 400 located at the abscissa d3 materializes the beginning of the braking phase of the first deceleration logic, lines 400 and 500 being distant from DMAX.

The curves 11, 12 in thin lines represent the planned speed profile for the autonomous vehicle 100, while the curves 15, 16 in bold lines represent the speed profile implemented by the vehicle as it moves. Thus the points MO to M3 represented on the graphs G1 to G3 materialize points of evolution of the speed of the autonomous vehicle 100 according to the curves 15, 16.

In the graph G1, the point MO represents the speed of the autonomous vehicle 100 at the instant T1 of detection of the next stop ARR. The MO point delimits the start of the braking anticipation phase. The point M1 being at a distance from the next stop DA(t) strictly greater than DMAX, it is located in the braking anticipation phase. 15

In the example illustrated, the speed of the autonomous vehicle 100 is constant during the anticipation phase. As previously described, the speed could however vary during the anticipation phase.

At a given instant T12, the autonomous vehicle has reached a point M2 located at the distance DMAX from the position PA1 of the next stop. Point M2, represented in graph G2, materializes the transition between the anticipation phase and the braking phase of the first deceleration logic. Point M2 therefore materializes the starting point of a speed profile 12 implemented in the braking phase of the first deceleration logic.

A preferred embodiment of the speed profile 12 of the braking phase of the first deceleration logic is described below with reference to Figure 4.

The speed profile 12 of the braking phase of the first deceleration logic preferably meets the following criteria:

- speed profile 12 enables the autonomous vehicle to reach position PA1 with substantially zero speed, and/or

- speed profiles 11 and 12 present continuity of speed and acceleration at point M2, and/or

- the speed profile 12 implements jerks whose absolute value is less than a first limit threshold JMAX.

In a preferred embodiment described below with reference to FIG. 4 and corresponding to the most frequent situation where the autonomous vehicle is heading towards the stopping point without needing to accelerate, the speed profile 12 responds by in addition to the following criteria:

- the speed profile 12 implements negative accelerations greater than a limit threshold AMIN, the threshold AMIN being able to be considered as a maximum deceleration in absolute value, and/or 16

- the speed profile 12 is made up of three consecutive sub-phases, an initial sub-phase 121 presenting a constant negative jerk J1, an intermediate sub-phase 122 presenting a zero jerk J2, and a final sub-phase 123 presenting a jerk constant positive J3, and/or - the positive jerk J3 of the final sub-phase 123 is greater than the absolute value of the negative jerk J1 of the initial sub-phase 121 .

The first limit threshold JMAX corresponding to a predetermined maximum jerk value, to reach an optimal target acceleration value at a given instant. This maximum jerk value is determined beforehand for the speed profile, a method allowing the determination of the maximum jerk value is for example described in the document FR3104520B1. Similarly, the limit threshold AMIN corresponds to a predetermined minimum acceleration (or maximum deceleration) value.

Advantageously, in the preferred embodiment, the speed profile 12 validates all of these criteria.

Other embodiments of the speed profile 12 can however be implemented in the management method according to the invention.

For example, when the vehicle is moving on a hill where the deceleration is strong,

- in the initial sub-phase 121 , the acceleration can increase from a very negative first value AO to a negative second value Ac greater than the first AO value,

- then, it can be constant in the intermediate sub-phase 122, - then, it can increase again towards a third value A3 in the final sub-phase 123.

In this case, the J1 jerk will be positive and of the same sign as the J3 jerk. 17

Similarly, in an alternative embodiment of the speed profile 12, one could implement at the end of the final sub-phase 123, a third negative acceleration value A3 which would be lower than the first acceleration value AO of the beginning of the initial sub-phase 121. In this case, the jerk J3 will be negative and of the same sign as the jerk J1.

As has already been specified, the position of the first stop position can change according to the data from the first detection means 1, which will require recalculating the speed profile 12 according to a new position of the autonomous vehicle 100 relative to a new position of the next stop.

The preferred embodiment of a speed profile 12 at a time t is therefore generically described below with reference to FIG. 4, between a given position M of the autonomous vehicle 100 at time t, the point M which may correspond to the point M2, and a stop position PA known at time t, the stop position possibly corresponding to the first stop position PA1. The curvilinear distance separating the positions M and PA at time t is called DREF.

The graphs G5, G6 and G7 of FIG. 4 respectively represent the temporal evolution of the jerk, of the acceleration and of the speed of the autonomous vehicle according to the speed profile 12, between the point M (at the instant T=0s ) and the PA stop position.

Velocity profile 12 is defined by a set of parameters represented on graphs G5 to G7, including

- so-called fixed parameters, which are determined by trajectory constraints at point M and at the stopping position PA, and

- so-called calibration parameters, the value of which can be modified from 18 so as to define a trajectory between the point M and the position PA which respects the comfort thresholds relating to the jerk and to the acceleration.

The fixed parameters and the calibration parameters described below together make it possible to define the speed profile 12 according to the three consecutive sub-phases 121, 122, 123 previously described.

In the preferred embodiment, the initial sub-phase 121 takes place between the instant T=0s and an instant T=T'1. During this sub-phase,

- jerk J1 is strictly negative and constant,

- the acceleration is a strictly decreasing linear function of time: it evolves between two negative values AO (at t=0s) and Ac (at t=T'1),

- the speed is therefore a strictly decreasing function of time evolving between a value V0 (at T=0s) and a value V1 (at T=T’1).

In the preferred embodiment, the intermediate sub-phase 122 takes place between the instant T=T'1 and an instant T=T'2. During this sub-phase,

- jerk J2 is zero,

- the acceleration is therefore a constant function of time: it is equal to the strictly negative value Ac between times TΊ and T'2,

- the speed is therefore a strictly decreasing linear function of the time evolving between the value V1 (at t=T'1) and a value V2 (at t=T'2).

In the preferred embodiment, the final sub-phase 123 takes place between the instant T=T'2 and an instant T=T'3. During this sub-phase,

- jerk J3 is strictly positive and constant,

- the acceleration is therefore a strictly increasing linear function of time: time evolving between the negative value Ac (at T=T'2) and a negative value A3 (at T=T'3),

- the speed is therefore a strictly decreasing function of the time evolving between the value V2 (at t=T'2) and a value V3 (at t=T'3). 19

The fixed parameters of speed profile 12 include the speed V0 and the acceleration AO of the vehicle at time T=0s. The speed and the acceleration being continuous at point M, the speed V0 is equal to the speed of the autonomous vehicle 100 measured at point M, and the acceleration AO is equal to the acceleration of the autonomous vehicle 100 measured at point M.

The fixed parameters of the speed profile also include the speed V3 and the acceleration A3 of the vehicle at the PA position. The autonomous vehicle 100 having to stop at the position PA, the parameters V3 and A3 are constants very close to 0. Their respective values can be fixed during the parameterization of the vehicle. We will return later in this document to the V3 and A3 values. The calibration parameters of the speed profile 12 comprise the parameters defining the temporal evolution of the jerk, that is to say the jerk values J1 and J3 applied respectively during the initial 121 and final 123 sub-phases. of the document, we define the multiplicative factor k such that J3=k. D1.

The speed profile calibration parameters 12 also include the minimum acceleration value Ac applied during the intermediate sub-phase 122. The instants TΊ to T′3 will be defined from the calibration parameters previously described for the speed profile 12.

A method of calibrating the speed profile 12 is described below with reference to Figure 5.

FIG. 5 represents a flowchart of five sub-steps C1 to C5 implementing the calibration method. 20

The method includes an iteration loop on sub-steps C1 to C4. The iteration loop makes it possible to determine a value for each of the calibration parameters J1 and Ac which allows the autonomous vehicle 100 to cover the distance DREF separating it from the stop position PA under jerk and acceleration conditions in accordance with the comfort criteria.

Then the sub-step C5 determines the speed profile I2 from the values determined for the calibration parameters J1, k and Ac.

Prior to the first calculation, an initial value is assigned to the calibration parameters. For example, Ac = -2m/s ² , J1 = -0.25m/s ³ and k=-3. The respective values of the parameters J1 and Ac are brought to evolve during the iterations on the sub-steps C1 to C4.

In a first sub-step C1, the sign S _k of the multiplicative factor k is determined.

The Math 1 formula makes it possible to determine the sign S _k according to the accelerations A0, A3 and Ac.

[Math 1] s _k = sign((A _s - A _c ). ( A _c - A ₀ ))

The multiplicative factor k can then be expressed according to the formula Math2 [Math 2] k — s _k . K _ca n _b where K caii _b is the absolute value of the multiplication factor. 21

Thus, depending on the accelerations AO, Ac and A3, the Mathl formula makes it possible to determine the sign of the jerks J1 and J3 so as to cause the acceleration to evolve according to the three consecutive sub-phases 121, 122 and 123 connecting the values AO, Ac and A3.

Furthermore, the value of jerk J1 chosen must make it possible to implement a speed profile 12 which,

- on the one hand, takes into account the speed values V0 and V3 and the acceleration values AO, A3 and Ac, and, - on the other hand, evolves according to the three sub-phases 121, 122, 123 previously defined .

For this, the absolute value of J1 must be less than or equal to the absolute value of a threshold Jnm, determined by the formula Math 3:

[Math 3]

If | J 11 is less than or equal to |Jlim|, then the value of J1 is not modified in sub-step C1 , otherwise J1 takes the value of Jlim.

In a second sub-step C2, the respective durations DTi and DT ₃ of the sub-phases 121 and 123 are expressed, according to the formulas Math4, as a function of the speeds V0 and V3, of the accelerations AO, A3 and Ac, and of the jerks J1 and D3.

[Math4] 22

The durations DTi and DT ₃ are then used to calculate, according to the Math5 formulas, the speeds V _t and V ₂ which are respectively the speeds at instant T and at instant T' _2. [Math5]

Finally, the duration of the sub-phase 122 is calculated according to the formula Math6

[Math6]

At the end of sub-step C2, all the parameters corresponding to a candidate speed profile 12 have therefore been determined, established from the fixed parameters and an initial value of the calibration parameters Ac, J1 and k.

It is thus possible to calculate, in the third sub-step C3, the distance traveled from the point M by the autonomous vehicle 100 according to the candidate speed profile 12 . A total distance of the profile X _T corresponding to the sum of the distances traveled during each of the phases 121, 122 and 123 is then calculated, according to the formulas Math7. [Math7]

C _T = C ^L 1+ X ₇ ₂ + X 3 23

In the third sub-step C4, the calibration parameters are optimized so that the total distance XT corresponds, with a given precision, to the curvilinear distance separating the point M from the first stop position PA; in other words, the calibration parameters are modified so that the distance XT is equal to the distance DREF with a given accuracy.

One embodiment of the optimization process is described below.

In a first sub-step C41 , the acceleration value A _c is sought by dichotomy making it possible to obtain a total distance of the profile X _T which is equal, to within a precision threshold, to the curvilinear distance DREF separating the point M from the PA stop position.

The acceleration A _c is located on an interval of values bounded by driving comfort criteria, in particular by the negative minimum value AMIN and the value 0. If an AOP value of this interval makes it possible to obtain a total distance of the profile X _T equal to DREF, then a preferential speed profile 12 is determined by the parameters AOP, J1 and k.

If an AOP value is not found, a second sub-step C42 is continued. The acceleration A _c is then fixed at the value AMIN, and an optimum jerk value is sought which makes it possible to obtain a total distance of the profile X _T equal to DREF.

The jerk J ₁ is located on an interval of values bounded by driving comfort criteria, in particular by the negative minimum value JMIN and the value 0. If a value JOP of this interval makes it possible to obtain a distance 24 total profile X _T equal to DREF, then a so-called alternative speed profile 12 is determined by the parameters AMIN, JOP and k.

If a value JOP is not found, then a third sub-step C43 is followed, consisting in producing a so-called degraded speed profile 12 by imposing Ac=AMIN and |J1 |=JMAX. In this case, the distance of the profile X _T will be greater than DMAX. But it is the shortest profile achievable while respecting the constraints of acceleration and jerk.

Optimization may require a loopback on step C1 (conditional loopback illustrated by a diamond between steps C4 and C5).

At the end of the optimization sub-step C4, the value of the calibration parameters is determined.

We then move on to a sub-step C5 of calculating the speed profile 12 determined by the calibration parameters.

As a side note, the velocity profile 1 thus calculated makes it possible to have a different entry and exit jerk, as well as a non-zero initial and final acceleration.

The profile 12 thus calculated allows the autonomous vehicle 100 to implement a comfortable slowing down for the users to stop at the position PA1 determined by the first detection means 1 .

The autonomous vehicle 100 therefore gradually approaches a first precise position PA1 of the next stop ARR according to the speed profile 12 implementing a slowdown according to the first slowdown logic. 25

At a time T2, the next stop ARR enters the range limit of the second detection means 2. We then move on to the second step E2.

In step E2, a second precise position PA2 of the next stop ARR is determined, then a second logic for slowing down the autonomous vehicle 100 is implemented.

To implement the second slowing down logic, the second detection means 2 calculates a stopping distance DA2(T2) separating the autonomous vehicle 100 from the precise stopping position PA2 at the instant T2 of detection of the next stop by the second detection means 2.

The point M3 represented by the graph G3 of FIG. 3 materializes the transition point between the first deceleration logic and the second deceleration logic. In other words, at point M3 is located both at a distance DA1 (T2) from the approximate position PA1 , and at a distance DA2 (T2) from the second stopping point.

In the graphs G3 and G4, the line 600 materializes a finish line of the vehicle on the next stop ARR according to the position PA2 detected by the second detection means 2.

The difference in distance ADA between lines 500 and 600 materializes the error in the estimation of the position of the next stop ARR induced by the lack of precision of the first means of detection 1 .

Line 700 materializes the start of the implementation of the second deceleration logic, lines 600 and 700 being distant from DA2(T2). 26

Curve 22 in a thin line represents the planned speed profile for the autonomous vehicle 100 during step E2, while curve 26 in a bold line represents the speed profile implemented by the vehicle as it moves. Thus the points M3 and M4 represented on the graphs G3 and G4 materialize points of evolution of the speed of the autonomous vehicle 100 according to the curve 22.

In the embodiment described, the second deceleration logic consists of one or more braking phases 22 determined by the same calculation method as the braking phase 12 of the first deceleration logic.

In this embodiment, the speed profile 22 of the braking phase according to the second deceleration logic meets the following criteria:

- speed profile 22 allows the autonomous vehicle to reach position PA2 with substantially zero speed, and/or

- speed profiles 12 and 22 present continuity of speed and acceleration at point M3, and/or

- the speed profile 22 implements jerks whose absolute value is less than a first limit threshold JMAX

In a preferred embodiment -described below with reference to FIG. 6 and corresponding to the most frequent situation where the autonomous vehicle is heading towards the stopping point without needing to accelerate-, the speed profile 22 also meets the following criteria:

- the speed profile 22 implements negative accelerations greater than a limit threshold AMIN, the threshold AMIN being able to be considered as a maximum deceleration in absolute value, and/or

- the speed profile 22 consists of three consecutive sub-phases, an initial sub-phase 221 having a constant negative jerk J'1, an intermediate sub-phase 222 having a zero jerk J'2, and a final sub-phase 223 exhibiting a constant positive jerk J'3, and/or 27

- the positive jerk J'3 of the final sub-phase 223 is greater than the absolute value of the negative jerk J'1 of the initial sub-phase 221 .

Advantageously, in the preferred embodiment, the speed profile 22 validates all of these criteria.

Other embodiments of the speed profile 22, not described in this document, are however taken into account in the management method according to the invention.

For example, when the vehicle is moving on a hill where the negative deceleration is strong,

- in the initial sub-phase 221 the acceleration can increase from a very negative first value A'O to a negative second value A'c and greater than the first value A'O,

- then it can be constant in the intermediate sub-phase 222,

- then it can increase again towards a third value A'3 in the final sub-phase 223.

In this case, the jerk J'1 will be positive and of the same sign as the jerk J'3.

Similarly, in an alternative embodiment of the speed profile 22, one could implement at the end of the final sub-phase 223, a third negative acceleration value A′3 which would be lower than the first value d acceleration A'O of the start of the initial sub-phase 221 . In this case, the jerk J'3 would be negative and of the same sign as the jerk J'1 .

The position of the second stop position can change according to the data from the second detection means 2, which will require recalculating the speed profile 22 according to a new position of the autonomous vehicle 100 relative to a new position of the next stop. 28

The preferred embodiment of a speed profile 22 at a time t is therefore generically described below with reference to FIG. 6, between a given position M of the autonomous vehicle 100 at time t, the point M which may correspond to the point M3, and a stop position PA known at time t, the stop position possibly corresponding to the second stop position PA2. The curvilinear distance separating the positions M and PA at time t is called DREF.

The graphs G8, G9 and G10 of FIG. 6 respectively represent the temporal evolution of the jerk, of the acceleration and of the speed of the autonomous vehicle according to the speed profile 22, between the point M (at the instant T=0s ) and the PA stop position.

The speed profile 22 is defined by a set of parameters represented on the graphs G8 to G10, including

- so-called calibration parameters, the value of which can be modified so as to define a trajectory between the point M and the position PA which respects the comfort thresholds relating to jerk and acceleration.

The fixed parameters and the calibration parameters described below together make it possible to define the speed profile 22 according to the three consecutive sub-phases 221, 222, 223 previously described.

In the preferred embodiment, the initial sub-phase 221 takes place between times T=0s and T=T”1. During this sub-phase,

- jerk J’1 is strictly negative and constant,

- the acceleration is a strictly decreasing linear function of time: it evolves between two negative values A'0 (at t=0s) and A'c (at t=T”1 ), 29

- the speed is therefore a strictly decreasing function of the time evolving between a value V'O (at T=0s) and a value V'1 (at T=T”1).

In the preferred embodiment, the intermediate sub-phase 222 takes place between the instant T=T”1 and an instant T=T”2. During this sub-phase,

- jerk J’2 is nil,

- the acceleration is therefore a constant function of time: it is equal to the strictly negative value A'c between times T”1 and T”2,

- the speed is therefore a strictly decreasing linear function of the time evolving between the value V'1 (at t=T”1) and a value V'2 (at t=T”2).

In the preferred embodiment, the final sub-phase 222 takes place between the instant T=T”2 and an instant T=T”3. During this sub-phase,

- jerk J’3 is strictly positive and constant,

- the acceleration is therefore a strictly increasing linear function of time evolving between the negative value A'c (at T=T”2) and a negative value A'3 (at T=T”3),

- the speed is therefore a strictly decreasing function of the time evolving between the value V'2 (at t=T”2) and a value V'3 (at t=T”3).

The fixed parameters of the speed profile 22 include the speed V′O and the acceleration AΌ of the vehicle at time T=0s. The speed and the acceleration being continuous at point M, the speed V'O is equal to the speed of the autonomous vehicle 100 measured at point M, and the acceleration AΌ is equal to the acceleration of the autonomous vehicle 100 measured at point M .

The fixed parameters of the speed profile also include the speed V′3 and the acceleration A′3 of the vehicle at the position PA. The autonomous vehicle 100 having to stop at the position PA, the parameters V′3 and A′3 are constants very close to 0. Their respective values can 30 be fixed when setting up the vehicle. We will return later in this document to the values V'3 and A'3.

The calibration parameters of the speed profile 22 comprise the parameters defining the temporal evolution of the jerk, that is to say the jerk values J′1 and J′3 applied respectively during the initial 221 and final 223 sub-phases .

The speed profile calibration parameters 22 also include the minimum acceleration value A'c applied during the intermediate sub-phase 222.

Instants T”1 to T”3 will be defined from the calibration parameters previously described for speed profile 22.

The method applied to calibrate the speed profile 22 is similar to the method described with reference to FIG. 5 for the calibration of the speed profile 12 of the first deceleration logic.

At the end of the calibration step, a speed profile 22 is obtained allowing the autonomous vehicle 100 to implement a comfortable slowdown for the users and to stop at the position PA2 determined by the second detection means 2 .

As illustrated by the graph G4 of FIG. 3, in particular by the point M4, the autonomous vehicle 100 therefore gradually approaches the precise position PA2 of the next stop ARR according to the speed profile 22 implementing a slowing down according to the second slowdown logic.

Thus, the autonomous vehicle 100 is first controlled according to at least one speed profile 12, implementing the first logic of 31 slowing to reach the first approximate stop position PA1 with substantially zero speed V3.

However, the at least one speed profile 12 is not intended to be followed up to position PA1. Indeed, before reaching the first stop position PA1, the autonomous vehicle 100 continues on at least one speed profile 22 implementing the second slowing down logic, the at least one speed profile 22 allowing it to reach the second precise stop position PA2 with a substantially zero speed V'3.

The choice to assign a non-zero value to V3 then V'3 is justified by the fact that, in certain embodiments of the invention, the set of sensors 6, in particular the speed sensor, measuring the current value of the speed of the autonomous vehicle does not make it possible to measure very low speeds, for example speeds below approximately 1 km/h, i.e. 0.3 m/s.

One solution consists in delegating the final management of the stop to an additional module which will implement an open-loop deceleration ramp over the last tens of centimeters of the speed profile.

In addition or alternatively, the speed profile 22 implemented in the second deceleration logic could be defined so that the vehicle reaches a position located at a very short distance upstream from the stop position PA2, for example 50cm, at very low speed (for example 1km/h) and with a moderate deceleration (for example 1m/s ² ). The last centimeters of the trajectory would then take place according to an open-loop deceleration ramp. 32

In total, the invention combines two complementary slowdown logics:

- a first slowing logic which makes it possible to anticipate the implementation of braking with a view to stopping the autonomous vehicle 100 at a position determined by a sign, then

- a second logic of slowing down which makes it possible to specify the implementation of the braking to immobilize the autonomous vehicle 100 in the position determined by the signpost.

The first slowing down logic requires a first detection means whose range is preferably at least one or several hundred meters, and whose precision can be relatively low, for example of the order of a few meters. The first means of detection can be a GPS location on a standard map.

The second detection logic requires a second detection means whose precision is high, for example having a margin of error of less than a few tens of centimeters, or even less than about ten centimeters, or even less than a centimeter, and whose range can be relatively low, for example of the order of a few tens of meters. The second detection means can be a front camera. The use of this second detection means in combination with the first detection means makes it possible to reach the stop with very high precision (of the order of a centimeter or about ten centimeters relative to the stop signage ). Indeed, the use of the first detection means alone would only allow a stop with an accuracy of the order of a few meters, which is largely insufficient and could prove to be dangerous when stopping the vehicle at an intersection. . Furthermore, the second detection means having a very low range, it would not on its own ensure the stopping of the vehicle when the latter is moving at high speeds (for example 80 km/h on a secondary road in France ). 33

Thus, by combining the first and the second slowing logic, the invention makes it possible to immobilize the autonomous vehicle 100 in a position determined by a sign, on the one hand by offering high benefits of comfort and braking precision and on the other hand by using detection means commonly installed on autonomous vehicles. The first deceleration logic making it possible to arrive close to the position determined by the road sign with a sufficiently low speed so that the second deceleration logic can ensure the complete stop of the vehicle with sufficient precision to ensure safety.

Claims

34 CLAIMS

1. Method for managing the longitudinal speed of an autonomous vehicle (100), the autonomous vehicle traveling on a traffic lane comprising a stop sign located at the front of the autonomous vehicle, the autonomous vehicle being equipped with a first detection means (1) of a first range (P1) and second detection means (2) of a second range (P2), the first range (P1) being greater than the second range (P2) , characterized in that it comprises a first step (E1) of detecting the stop signs by the first detection means (1), and of implementing a first logic for slowing down the autonomous vehicle, a second step (E2) of detecting the stop signs by the second detection means (2), and of implementing a second logic for slowing down the autonomous vehicle, in that the first and the second slowing down logic set implementation of jerks whose absolute value is less than a first se uil limit (JMAX), and in that the second slowing down logic controls the stopping of the autonomous vehicle with an accuracy of the order of a centimeter relative to the stop signage, or of the order of about ten centimeters relative to the stop signage, or of the order of several tens of centimeters relative to the stop signage.

2. Method according to the preceding claim, characterized in that the first and the second deceleration logic implement negative accelerations greater than a second limit threshold (AMIN). 35

3. Management method according to one of the claims, characterized in that the first step (E1) comprises a determination, at a first instant (T1), of a first approximate position (PA1) of the stop signs, and the second determination step (E2) comprises a determination, at a second instant (T2), of a second precise position (PA2) of the stop sign, the second instant (T2) being strictly subsequent to the first instant ( T1).

4. Management method according to the preceding claim, characterized in that the first deceleration logic initiates a first deceleration phase (12) when the vehicle arrives at a given distance (DMAX) from the first approximate position (PA1) of the signage stop (ARR).

5. Management method according to the preceding claim, characterized in that the second deceleration logic starts a second deceleration phase at the second instant (T2), and in that the second deceleration phase has a continuity of speed and acceleration with the first phase of deceleration.

6. Management method according to the preceding claim, characterized in that,

- the absolute value of the jerk at the end of the first deceleration phase is greater than the absolute value of the jerk at the start of the first deceleration phase, and/or

- the absolute value of the jerk at the end of the second deceleration phase is greater than the absolute value of the jerk at the start of the second deceleration phase.

7. Management method according to one of claims 5 or 6, characterized in that, 36

- the first deceleration phase is made up of three consecutive sub-phases, a first initial sub-phase having a first non-zero constant jerk (J1 ), a first intermediate sub-phase having a zero jerk, and a first final sub-phase having a second non-zero constant jerk (J3), and/or

- the second deceleration phase is made up of three consecutive sub-phases, a second initial sub-phase presenting a third non-zero constant jerk (J'1 ), a second intermediate sub-phase presenting a zero jerk, and a second sub-phase final phase having a fourth non-zero constant jerk (J′3).

Management method according to the preceding claim, characterized in that

- the second jerk (J3) is the product of the first jerk (J1) by a first multiplicative factor (k), in particular a first multiplicative factor (k) whose sign is the sign of the product between, on the one hand, the difference between a first acceleration (A3) of the end of the first final sub-phase and a second acceleration (Ac) of the first intermediate sub-phase, and, on the other hand, the difference between the second acceleration (Ac) and a third acceleration (AO) of the start of the first initial sub-phase, and/or in that

- the fourth jerk (J'3) is the product of the third jerk (J'1) by a second multiplicative factor, in particular a second multiplicative factor whose sign is the sign of the product between, on the one hand, the difference between a fourth acceleration (A'3) of the end of the second final sub-phase and a fifth acceleration (A'c) of the second intermediate sub-phase, and, on the other hand, the difference between the fifth acceleration (A' c) and a sixth acceleration (AΌ) of the start of the second initial sub-phase. 37 Device (10) for managing the longitudinal speed of an autonomous vehicle (100), the autonomous vehicle being equipped with a brake actuator (5), the device comprising elements (1, 2, 3, 4, 31 , 32, 33, 311, 312) hardware and/or software implementing the method according to one of claims 1 to 8, in particular hardware elements (1, 2, 3, 4, 31, 32, 33) and/ or software designed to implement the method according to one of the preceding claims, and/or the device comprising means for implementing the method according to one of the preceding claims. Autonomous vehicle (100) comprising a device (10) for managing the longitudinal speed according to the preceding claim.