FR2887669A3 - METHOD AND IMPROVED SHOCK PREDICTION SYSTEM BETWEEN A VEHICLE AND A PIETON - Google Patents

Abstract

The invention relates to an on-board system for predicting a collision between a vehicle and a pedestrian detected in its environment, comprising means for detecting pedestrians at a distance associated with means for predicting their trajectory, and means for detecting the operation of the vehicle. and driver behavior detecting means associated with vehicle trajectory prediction means, further comprising a central control unit receiving pedestrian information detected from the trajectory predicting means and vehicle information. and on the behavior of the driver coming from the means for detecting his trajectory, and delivering information for predicting a possible vehicle / pedestrian shock, notably constituted by a time before impact (TTI), of an impact zone (ZIP) and a probability of shock (Pc), to countermeasures systems which are triggered successively according to their respective thresholds.

Description

METHOD AND ON-OFF SHOCK PREDICTION SYSTEM

BETWEEN A VEHICLE AND A PIETON.

  The invention relates to an on-board collision prediction system between a vehicle and a pedestrian, considered as a vulnerable user of the road which must be secured from the vehicle. It also relates to a shock prediction method between a vehicle and a pedestrian detected in its environment, implemented by said embedded system.

  The automotive industry is increasingly interested in vulnerable road users, of which pedestrians make up the vast majority and the introduction of pedestrian impact tests, in the EURO NCAP tests, illustrates its interest in pedestrians. The first generation of active pedestrian protection systems is intended to detect a vehicle / pedestrian impact by contact sensors installed on the front face of the vehicle, in order to trigger protection means such as active bonnet or pedestrian airbags.

  Currently, apart from the detection of pedestrian impact, which consists in detecting that the obstacle actually struck by the vehicle is indeed a pedestrian, there are remote pedestrian detection systems present in the detection field of a sensor. remote (mono or stereovision camera, visible or infra red, active 3D camera, PMD, LIDAR, radar, range finder, ultrasound ...) around the vehicle.

  In particular, the patent application US Pat. No. 6,035,053, in the name of Mazda Motor Corporation, describes a collision detection system based on the notion of danger, delimited by thresholds on the position of the obstacle, and the request for EP 1 095 832, in the name of Director General of Public Works Reasearch Institute (JP), also deals with pedestrian accidents, but relying on cooperation with the intelligent road infrastructure, which would imply that electronic systems equip the edge of all roads.

  The pedestrian shock prediction itself has already been the subject of a patent application FR 03 15548 in the name of RENAULT, filed December 31, 2003. This prediction is based on a probabilistic model of trajectories.

  The object of the invention is to provide an improved system for pedestrian safety, on-vehicle system equipped, autonomous and not requiring interaction with other systems on other vehicles, pedestrian and / or infrastructure.

  For this, a first object of the invention is an on-board system for predicting a collision between a vehicle and a pedestrian detected in its environment, comprising on the one hand remote pedestrian detection means associated with means for predicting their presence. trajectory, and secondly means for detecting the operation of the vehicle and driver behavior detection means associated with means for predicting the trajectory of the vehicle, characterized in that it further comprises an electronic central unit of control receiving information on the pedestrian detected from the trajectory prediction means and information on the vehicle and on the behavior of the driver from the means for detecting its trajectory, and delivering information for predicting a possible vehicle shock / pedestrian, consisting in particular of a time before impact, an impact zone and a probability of c, to countermeasure systems which are triggered successively according to their respective thresholds.

  According to another characteristic of the onboard shock prediction system, the electronic control central unit is connected to means of watch, means of information of the driver of the vehicle and information of the pedestrian, means of avoidance of a collision, means for implementing a pre-crash with a pedestrian and pedestrian impact protection means, to which it issues trip commands.

  A second object of the invention is a method for predicting a collision between a vehicle and a pedestrian detected in its environment, implemented by the preceding device, comprising a remote pedestrian detection step followed by a step of predicting their trajectory, a step of detecting the operation of the vehicle and a step of detecting the behavior of the driver associated with a step of predicting the trajectory of the vehicle, characterized in that it further comprises a step of processing the prediction information of the vehicle. the trajectory of the detected pedestrians and the prediction of the trajectory of the vehicle, for calculating information for predicting a possible vehicle / pedestrian shock, constituted in particular by a time before impact, an impact zone and a probability shock, and issue trip orders to countermeasure systems.

  According to another characteristic of the shock prediction method, it further comprises a step of calculating a weighted shock probability P, p, obtained by multiplying the initial shock probability Pc by two weighting functions, or weight, the one pu, due to the time before impact and the other pu ,, due to the location of the impact: PCP = PC * ITT, * Z, P According to another characteristic of the shock prediction process, it performs the following steps: a step el) of calculating the weighted shock probability which a) in the case where the probability is zero, therefore no risk of shock detected, no shock zone or expected time before impact, is followed by a step e2) for maintaining the system in standby, and b) in the case where the probability is non-zero, so that a risk of shock is detected between a pedestrian and a vehicle and as the time before impact is reduced, is followed by: - a step e3) of comparison of the time before impact TTI with a first threshold S, which a) in the case where the time before impact is less than said threshold, is followed by a step e4) of protecting the pedestrian during the shock by triggering the passive protection devices, and b) in the case where the time before impact is greater than said threshold, is followed byE - a step e5) of comparison of the time before impact TTI with a second threshold S2, greater than the first threshold S, which a) in the case where the time before impact is less than said threshold S2, is followed by a step e6) of ordering a triggering order of a pedestrian pre-crash procedure by emergency braking, and pre-activation of passive pedestrian impact protection systems, and b) in the case where the time before impact is greater than said threshold S2: - step e7): comparison of the time before impact TTI with a third threshold S3, greater than the second threshold S2, which a) in the case where the time before impact is lower at said threshold S3, is followed by a step e8) of collision avoidance by calculating the optimal escape trajectory and triggering the vehicle trajectory control system for automatic modification of the vehicle trajectory, by combining actions on the steering to the left and / or the straight line, and on the speed of the vehicle by acceleration and / or braking, and b) in the case where the time before impact is greater than said threshold S3, is followed by a step e9) of ordering an order of informing the driver and / or the pedestrian by triggering an alert signal.

  Other features and advantages of the invention will become apparent on reading the description of nonlimiting exemplary embodiments, illustrated by the following figures which are: FIG. 1: the three main phases of pedestrian protection; FIG. 2: a system making it possible to know the probability of vehicle / pedestrian impact, FIG. 3: a top view of a vehicle and a pedestrian in front of the hood, FIG. 4: an example of a geometric modeling of a vehicle. frontal collision between a vehicle and a pedestrian, - figures 5 to 7: three variants of definitions of the shock zone corresponding to the description respectively of the single shock zone, the finer impact zone and the shock zone; Figure 8: the probability curve of a space shock as a function of time, Figure 9: the weighted probability of shock as a function of time before impact, figure 10: the expected shock zone, in%, recorded on the schema of a v FIG. 11: the weighted probability of shock as a function of the said shock zone in%; - FIG. 12: various means constituting a pedestrian protection system; FIG. 13: a non-limiting example of a flowchart of various steps constituting the prediction method according to the invention, - Figure 14: a non-limiting example of the different steps of this method according to the invention.

  The aim of the invention is to protect pedestrians near the traffic lanes, it consists in detecting and locating the pedestrians present among other obstacles around a vehicle, in predicting the evolution of the movement of said pedestrians in relation to vehicle, and to associate to them a probability of impact with said equipped vehicle as well as an estimate of the time before impact and the shock zone, with the objective of triggering the appropriate countermeasures ensuring the protection of pedestrians. This pedestrian protection can be broken down into three main phases, summarized in the diagram of Figure 1. A first phase P, consists of the analysis of the road scene, the detection of obstacles, and among them pedestrians, their position and the probable evolution of the movement of the identified pedestrian as well as that of the vehicle. The second phase P2 consists, from this information, in delivering a shock prediction, with a time before impact and a shock zone with a shock probability, in order to achieve a third phase P3 optimized triggering countermeasure systems.

  The prediction of a pedestrian shock, that is to say the announcement that the vehicle will soon hit a pedestrian, involves taking into account the uncertainties on the evolution of the respective positions of the vehicle and the pedestrian. An example of a system making it possible to know the probability of vehicle / pedestrian impact is given in the diagram of FIG. 2. It comprises on-board means 1 for detecting the environment of the vehicle, such as conventional sensors, connected to means 2 of the vehicle. processing this information, which on the one hand carries out the detection and classification of obstacles found in the environment, in particular pedestrians and on the other hand the estimation of their relative position and speed. This information on detected pedestrians is sent to means 3 for predicting their trajectory, which requires a pedestrian model with rectilinear motion, or uniform rectilinear movement, or uniformly accelerated movement, or probabilistic model of pedestrian trajectory for example. Pedestrian detection can be done independently of obstacle detection, which involves two different types of detection means.

  It furthermore comprises sensors 4 for operating the vehicle, notably delivering its speed, its longitudinal and transverse accelerations, its flying angle and other information relating to the behavior of the driver, as a function of the commands which he applies in particular to means 5 prediction of the trajectory of the vehicle, which require another model, with rectilinear movement, uniform rectilinear motion, uniformly accelerated movement, rear wheel model, or probabilistic model of vehicle trajectory.

  A central unit 6 compares these two probable trajectories and delivers a vehicle / pedestrian shock prediction to the countermeasure devices 7 in order to optimize their triggering with a view to protecting the pedestrian detected. Depending on the hardware architecture chosen, a part of the calculations and processes may be performed at dedicated computing units, typically in a sensor subsystem, the information processing is done in this subsystem and a higher level of information is then sent to the central unit.

  The invention relates solely to the prediction of frontal shocks between a pedestrian and the front of the vehicle which is modeled by a segment having the width dimension L of the vehicle, as shown in Figure 3, which is a top view of a vehicle. vehicle 8 and a pedestrian 9. The pedestrian is considered to be a cylinder of diameter 2R equal to the maximum width of an average pedestrian and of the same height as this average pedestrian, so that it is possible to define a zone of vehicle / pedestrian shock corresponding to an intersection between a segment representative of the front face of the vehicle and a disc representative of the pedestrian envelope, as shown in FIG. 4 which is an example of geometric modeling of a frontal collision between a vehicle and a pedestrian. For example, the diameter 2R is equal to 60 cm.

  Three situations are highlighted: There is shock when most of the pedestrian is covered by the front of the vehicle (zebra area C).

  If there is no shock when there is no overlap between the pedestrian model and the front face, then there is ambiguity shock / no shock when less than half of the circle encompassing the pedestrian is cut by the front face (dashed area A).

  Within the zone of ambiguity, it is possible to define a function, of the type gravity of the impact, which would pass continuously from 0 (no impact) to 1. The notion of shock would then correspond to the crossing a threshold to be defined. This possibility of weighting the importance or the severity of an impact can be interesting when evaluating real systems: a priori, to predict a shock taking place in the middle of the front face of the vehicle is simpler to predict. front impact taking place on one of the left or right edges.

  Three variants of definitions of the shock zone are shown in FIGS. 5 to 7, corresponding to the description of the single shock zone, the fine impact zone and the continuous shock zone, respectively. The simple shock zone Zs, without zone of ambiguity, is a rectangle of width equal to 2R and of length equal to the sum of the width L of the vehicle and the diameter 2R of the pedestrian model. The fine impact zone Zf is the combination of a rectangle, length L and width 2R, and two half-circles of radius R at each end. The continuous shock zone Zc has the same shape as the previous zone but with different concentric bands Bc corresponding to a decreasing probability of shock when one moves away from the hood of the vehicle.

  These different definitions of shock between a pedestrian and a vehicle are independent of time, so the system according to the invention calculates the first moment from which there is recovery between the vehicle and the pedestrian, which will be called moment of impact, or moment of first impact. This is a spatio-temporal rendez-vous. FIG. 8 is the probability curve Pc of a spatial shock as a function of time, ie the recovery rate between the vehicle and the pedestrian, presenting a first zone z, of absence of obvious shock between the current moment to of shock prediction and the instant t, from which the probability is no longer zero, a second zone z2 of ambiguity of this instant t, until time t; first impact from which shock is evident (zone z3).

  The vehicle / pedestrian shock prediction system according to the invention delivers information on: É the probability of impact Pc, É the time before impact TTI, É the expected impact zone ZIP.

  It is possible to take into account the information provided by the time before impact and the expected impact area to weight the probability of Pc shock and obtain a global PcP weighted impact probability. For this purpose, two weighting functions are defined, which respectively depend on the predicted impact time and impact zone. The probability of weighted shock Pcp is obtained by multiplying the initial shock probability Pc by the two weighting functions, or weights, one pu, due to the time before impact and the other pz, P due to the location of the impact: PcP = PC * pTT1 * pZ1P An example of weighting as a function of time before predicted TTI impact is given in table A in the appendix, representing the weighting function or weight pu, as a function of time before impact predicted, and is shown in Figure 9. In this example, the weight increases from 0 to 1 substantially linearly from 3 seconds before the expected impact.

  An example of weighting according to the ZIP predicted shock zone is given in Table B in the appendix, representing the weighting function or weight pz, P weighting points as a function of the predicted shock zone, and is represented on Figure 11. This function is symmetrical with respect to the middle of the front bumper, as shown in the diagram of a vehicle in Figure 10.

  The method according to the invention then uses this probability of weighted shock to determine different thresholds for triggering countermeasures, and therefore the strategy best suited to avoid shocks. To avoid a shock consists in reducing the probability of weighted shock and for that it is possible to act on the three quantities, linked together, determined previously: É the probability of simple shock, the time before anticipated impact: a reduction of the speed of the vehicle causes an increase in the time before impact, E the estimated impact zone: to ensure that the pedestrian is no longer in a zone of maximum risk, by acting on the vehicle or by warning the pedestrian by

example.

  From this information, it is also possible to control the system of countermeasures relating to avoidance, giving instructions for change of direction and pace.

  Depending on the environmental context of the vehicle and the imminence of the vehicle / pedestrian impact, several pedestrian protection strategies can be envisaged according to the different means that make up the pedestrian protection system and that intervene according to the seriousness of the situation in triggering the most suitable countermeasures under the control of the central unit Uc, as shown in the diagram of Figure 12. These means, successively triggered depending on the imminence of the shock, are means M, standby, means Mie information of the driver of the vehicle and Mir, information of the pedestrian, means Me to avoid a collision, means Mpc for implementing a pre-crash with a pedestrian and means Min, shock / pedestrian protection.

  The prediction method according to the invention therefore comprises, according to a non-limiting configuration mode represented on the flowchart of FIG. 13, the following steps, carried out by the central control unit which receives and processes information coming from different on-vehicle sensors and measurement systems: a step el) for calculating the weighted probability of shock and a) in the case where the central unit calculates a zero probability, therefore did not detect any risk of shock, the system remains in standby (step e2): no information on a risk of danger, on a shock zone or a time before impact delivered by the central unit, the system remains in this state and maintains the day before.

  b) in the case where the central unit detects a non-zero probability, so that a pedestrian would be likely to be hit by a vehicle and as the time before impact is reduced: - a step e3) of time comparison before TTI impact with a first threshold S ,, of between 200 and 500 ms for example: a) in the case where the time before impact is less than said threshold: a step e4) of protecting the pedestrian during the shock by triggering the devices of passive protection, such as active hoods, pedestrian airbags and others, b) in the case where the time before impact is greater than said threshold: - a step e5) of comparison of the time before impact TTI with a second threshold S2, greater than the first threshold , and between 700 ms and 1 second, for example: a) in the case where the time before impact is below said threshold: a step e6) of pre-pedestrian crash by triggering of the emergency braking system, and pre-activation of the systems passiv protection e pedestrian shock.

  b) in the case where the time before impact is greater than said threshold: a step e7) of comparing the time before impact TTI with a third threshold S3, greater than the second threshold, and including 1.5 and 2 seconds, for example: a ) if the time before impact is less than said threshold: a step e8) of collision avoidance by calculating the optimal escape trajectory and triggering the vehicle trajectory control system with a view to an automatic modification of the trajectory of the vehicle, by combining actions on the direction to the left and / or the right, and on the speed of the vehicle by acceleration and / or braking, b) in the case where the time before impact is greater than said threshold: a step e9) informing the driver and / or the pedestrian by triggering an alert signal.

  This warning signal for the driver can be: ^ visual: on the classic dashboard or a virtual dashboard with overprinting of the pedestrian at risk, type HUD (head up display) including, ^ sound: voice signal or sound characteristic from the relative direction of the pedestrian, haptic: vibration on the accelerator pedal, seat or steering wheel.

  And for the pedestrian, the alert can be a signal of the type: ^ sound: directional warning, signal characteristic, ^ visual: automatic call of headlights ^ communication signal with the road infrastructure, other vehicles or pedestrians.

  Thus, depending on the hazard defined by the probability of impact, the estimated impact zone and the predicted impact time, one of the four systems of countermeasures, information, avoidance, pre-crash or passive protection, will be selected. to be implemented. If this reaction is sufficient to reduce or eliminate the risk of shock, the system returns to standby mode. Otherwise, as the moment of shock is reduced, more and more automated and urgent reactions are triggered.

  To avoid an accident between a vehicle and a pedestrian, on the one hand it is not realistic to warn the driver whenever a pedestrian is detected by the remote pedestrian detection system, because he must already treat a quantity of data, and on the other hand means of alerting the driver, but also means of automatic modification of the trajectory, emergency braking or pedestrian protection systems on the front face of the vehicle, must guarantee a false alarm rate, so very low nuisance tripping, so that the system remains considered reliable for its users. It is therefore necessary to filter the false alarms and to trigger the countermeasures only when the shock is almost certain.

  According to a simple and non-exhaustive exemplary embodiment, the detection of a pedestrian and the estimation of its position and its speed are done thanks to a stereovision sensor, such as 2 video cameras on a stereo base, arranged at the level of rearview mirror inside the vehicle. Once acquired, these images are processed at the level of the calculation module of the obstacle detection and classification system, which classifies pedestrians and returns information about the pedestrians detected, as well as their positions and relative speeds.

  Vehicle status information (speed, gear ratio, acceleration information, pitch angles, roll, yaw, engine speed, etc.) and driver intentions (steering angle, positions and pressures on the pedals, indicators ...) are available on the CAN bus.

  The information on pedestrians, the vehicle, or the driver arrives within the central computing unit, where the prediction is made, according to 3 possible prediction modes, described in J. Kuchar, L. Yang. "A Review of Conflict Detection and Resolution Modeling Methods". In IEEE Transactions on Intelligent Transportation Systems, Vol. 1, No. 4, pp. 179-189. , December 2000: É nominal or deterministic prediction, for which a single trajectory is predicted, without taking into account uncertainties about the evolution of state variables. Its use is adapted to cases where the uncertainties are low and gives a very good estimate of current state variables and a prediction over a short time horizon.

  É The worst case prediction, on the contrary, assumes that mobiles can perform all kinds of maneuvers. It suffices that a single potential maneuver leads to a conflict so that the alert is given, which guarantees the best level of protection but in return generates false alarms.

  É Probabilistic prediction, in which the different possible evolutions of states are integrated and considered as uncertainties. For example, a position error can be added to the prediction according to the nominal method. Another approach consists in generating a set of possible trajectories, then assigning them probabilities of realization.

  In the standby mode represented by the step E 1, it is the so-called worst-case prediction (step E2) that operates continuously and, when it predicts a shock (step E3), more investigations are conducted by the intermediate of the nominal type prediction (step E4). When it also detects said shock in step E5, the method according to the invention controls an alert to the driver and / or the pedestrian, with a triggering of countermeasures (step E6). On the other hand, if the nominal prediction does not detect said shock, the method initiates a probabilistic type prediction (step E7), which in turn will cause an alert and countermeasures if it detects a shock well (step E8), otherwise return the process to sleep. The flowchart of FIG. 14 shows the different steps of this particular embodiment of the method according to the invention.

  For each of the three predictions, the weighting as a function of time before impact and the impact zone is calculated. If the probability of weighted shock is greater than the trigger threshold, then the countermeasures adapted, as a function of the time before impact, but also according to the impact zone, since it is conceivable to trigger only the passenger airbags on the side the impact if the impact does not take place in the middle of the bumper, is triggered.

  The invention thus relates on the one hand to a complete vehicle vehicle pedestrian shock prediction system, as well as to the various subsystems that make it possible to detect a pedestrian, to estimate its kinematic characteristics and to predict the risk of pedestrians. shock, and on the other hand an overall strategy of pedestrian protection, based primarily on the prediction of vehicle / pedestrian impact. The use of embedded communication systems can improve the performance of such an autonomous system.

  Advantageously, the invention makes it possible to have a second generation of systems, intended to predict pedestrian shocks a few moments before the actual shock, in order (in order of difficulty and increasing prediction horizon): E to prepare said protection systems sufficiently in advance to shorten the trigger time, and gain in response time, and take steps to reduce the severity of the shock (collision mitigation or collision mitigation), by initiating an emergency stop, and attempting a avoidance maneuver (collision avoidance).

  Advantageously, the invention also makes it possible to envisage preventing the driver from the presence of potentially dangerous pedestrians, the danger being synonymous with the risk of collision.

13 APPENDIX Table A

  TTI (s)] -8; -3] -1.5 -0.8 [-0.3; 0] PTTI 0 0.5 0.8 1

Table B

  ZIP (%) 0 20% 40% 50% 60% 70% Pm !, 1 1 0.85 0.6 0.15 0

Claims (1)

  An on-board system for predicting a collision between a vehicle and a pedestrian detected in its environment, comprising on the one hand means for detecting pedestrians at a distance associated with means for predicting their trajectory, and on the other hand means for detection of the operation of the vehicle and means for detecting the behavior of the driver associated with means for predicting the trajectory of the vehicle, characterized in that it further comprises an electronic central control unit receiving information on the pedestrian detected from trajectory prediction means and information on the vehicle and on the behavior of the driver coming from the means for detecting its trajectory, and delivering information for predicting a possible vehicle / pedestrian impact, consisting in particular of a time before (TTI), an impact zone (ZIP) and a shock probability (Pc), to counter-attack systems measure which are triggered successively according to their respective thresholds.   2. Onboard shock prediction system according to claim 1, characterized in that the electronic control central unit is connected to means (Mv) standby, means (M; c) information of the driver of the vehicle and (M1) pedestrian information, means (Me) for avoiding a collision, means (Mpc) for implementing a pre-crash with a pedestrian and means (Mp, P) for protecting a pedestrian shock, to which it issues trip orders.   3. A method for predicting a collision between a vehicle and a pedestrian detected in its environment, implemented by the onboard system according to one of claims 1 or 2, comprising a remote pedestrian detection step followed by a step of prediction of their trajectory, a step of detecting the operation of the vehicle and a step of detecting the behavior of the driver associated with a step of predicting the trajectory of the vehicle, characterized in that it further comprises a step of processing the information of predicting the trajectory of the pedestrians detected and predicting the trajectory of the vehicle, in order to calculate information for predicting a possible vehicle / pedestrian impact, consisting in particular of a time before impact (TTI), of an impact zone (ZIP) and a probability of shock (Pc), and issue trip orders to countermeasure systems.   4. Shock prediction method according to claim 3, characterized in that it further comprises a step of calculating a weighted impact probability (Pcp) obtained by multiplying the initial shock probability (Pc) by two functions. weighting, or weight, one (Pm) due to the time before impact and the other (pz, p) due to the location of the impact: n] PCP PC * PTTI * l "ZIP 5. Prediction process shock absorber according to claim 4, characterized in that the weighting function, or weight, (pn,) due to the time before impact increases substantially linearly from 0 to 1 from 3 seconds before the expected time of impact.   6. Impact prediction method according to claim 4, characterized in that the weighting function, or weight, (pu,)) due to the location of the intended impact is symmetrical with respect to the middle of the front bumper of the vehicle, maximum in an area centered in the middle of the bumper and decreasing to zero at both ends of the bumper.   7. Impact prediction method according to one of claims 3 to 6, characterized in that it performs the following steps: - a step el) calculation of the probability of weighted shock that a) in the case where the probability is zero, therefore no risk of shock detected, 20 no shock zone or predicted time to impact, is followed by a step e2) of standby system, and b) in the case where the probability is non-zero , so that a risk of shock is detected between a pedestrian and a vehicle and as the time before impact is reduced, is followed by: - a step e3) comparison of the time before impact (TTI) with a first threshold ( S,), which a) in the case where the time before impact is less than said threshold, is followed by a step e4) of protecting the pedestrian during the shock by triggering the passive protection devices, and b) in the case where the time before impact is greater than said threshold, is followed by: - a step e5) of comparisons time before impact (TTI) with a second threshold (S2), greater than the first threshold (S,), which a) in the case where the time before impact is below said threshold (S2), is followed by a step e6) for ordering a trip order from a pre-crash pedestrian procedure by emergency braking, and pre-activation of passive pedestrian crash protection systems, and b) in case the time before impact is greater than said threshold (S2), is followed by: - a step e7) of comparison of the time before impact (TTI) with a third threshold (S3), greater than the second threshold (S2), which: a) in the case where the time before impact is below said threshold (S3), is followed by a collision avoidance step e8) by calculating the optimal escape trajectory and triggering of the vehicle trajectory control system with a view to an automatic modification of the trajectory of the vehicle, by combining actions on the direction to the left and / or the right, and r the speed of the vehicle by acceleration and / or braking, and b) in the case where the time before impact is greater than said threshold (S3), is followed by a step e9) of controlling a driver information order and / or the pedestrian by triggering an alert signal.   8. Shock prediction method according to claim 7, characterized in that the control of an information order of the driver is performed by triggering an alert signal in the form: visual: on the classic dashboard or a virtual dashboard with overprinting of the pedestrian at risk, sound: voice signal or its characteristic coming from the relative direction of the pedestrian, haptic: vibrations on the accelerator pedal, on the seat or the steering wheel.   9. Impact prediction method according to claim 7, characterized in that the control of a pedestrian information command is carried out by triggering an alert signal in the form of: sound: directional warning, characteristic signal, ^ visual: automatic call of headlights E signal communication with the road infrastructure, other vehicles or pedestrians.