FR3073482A1 - METHOD FOR THE RELIABILITY OF THE CONTROL OF AN AUTONOMOUS VEHICLE - Google Patents

Abstract

The invention relates to a method for making the control of an autonomous vehicle more reliable during an autonomous taxiing phase on a taxiway (V), said method comprising a step of generating, from captured images, a detection video signal determining the taxiway (V) on which the vehicle is traveling, and a step of generating, from the generated detection video signal, a video signal for controlling the trajectory of said vehicle on the taxiway (V). The method further comprising a step of superimposing said video signals in a convergence video signal (30) and a step of determining the convergence of the detection and control video signals on said convergence video signal (30) to detect a video signal. error in the generation of the control video signal.

Description

METHOD FOR RELIABILITY OF THE CONTROL OF A SELF-CONTAINED VEHICLE.

The invention relates, in general, autonomous driving, in particular of a motor vehicle.

The invention relates more particularly to a method of making the control of an autonomous vehicle more reliable.

In known manner, a motor vehicle includes aids, such as parking aids and driving aids. Driving aids include regulating the vehicle's longitudinal speed, braking assistance, keeping the vehicle in a lane, blind spot monitoring, driver monitoring, etc. Such aids make it possible to assist the driver when using the vehicle in order to help him in his decision-making.

Today, certain vehicles, known as "autonomous", include a computer which executes a decision state machine allowing an advanced driving aid. Indeed, the decision state machine allows the vehicle, from various sensors detecting the environment around the vehicle, to make decisions autonomously and thus allow the vehicle to drive itself. However, such an autonomous vehicle must comply with standards in order to guarantee the safety of the passengers of the autonomous vehicle. Such standards aim in particular to impose a minimum level of reliability in decision-making.

We know from document FR3047589, a method for reconstructing an aerial view of the ground from images captured by a vehicle camera in order to detect the marking lines on the ground and thus to determine when the vehicle crosses a line. This thus makes it possible to estimate the position of the vehicle with respect to the traffic lane on which it is traveling in order to allow the decision state machine to autonomously control the vehicle so that it remains in its lane.

However, such a solution has a low level of security for autonomous driving.

The invention therefore aims to solve these drawbacks by proposing a method for increasing the security level of the orders taken by the decision state machine.

To achieve this result, the present invention relates to a method for making the control of an autonomous vehicle more reliable during an autonomous driving phase on a traffic lane, said vehicle comprising a detection module suitable for capturing images of the environment around said vehicle, said method comprising a step of generating a detection video signal from the images captured by said detection module, said detection video signal determining the traffic lane on which the vehicle is traveling, and a step of generating, from the generated detection video signal, a video signal for controlling the trajectory of said vehicle on the taxiway. The method further comprising a step of superimposing said control video signal and said detection video signal in a convergence video signal and a step of determining the convergence of the detection and control video signals on said convergence video signal in order to detecting an error in the generation of the control video signal.

Thanks to the method according to the invention, the level of robustness of the control of the vehicle in autonomous driving is high thanks to the monitoring of the control signal. This monitoring is carried out easily by comparison of the control video signal with the detection video signal during the study of the convergence of these video signals. The method according to the invention thus makes it possible to determine the ability of the vehicle to be maintained in the autonomous driving phase by detecting possible errors in the video control signal. This allows safe operation of the vehicle during the autonomous driving phase.

Advantageously, the detection video signal comprises the lines for marking on the ground of a taxiway and the obstacles present on said taxiway.

Advantageously, the detection module comprises at least one camera suitable for capturing images of the ground in order to generate the detection video signal.

Advantageously, the vehicle comprises a control computer executing a state machine adapted to generate the video control signal.

Advantageously, the video detection signal includes different zones: a security zone, an out of zone and an uncertainty zone.

Advantageously, the control video signal comprises a target for the trajectory of the vehicle.

Advantageously, the step of determining convergence comprises determining the position of the trajectory target in one of said zones.

Advantageously, the method comprises a step of alerting the driver of the vehicle when the position of the trajectory target is determined outside of the area.

The invention also relates to a motor vehicle comprising a detection module suitable for capturing images of the environment around said vehicle and generating, from said captured images, a detection video signal determining the traffic lane on which circulates the vehicle, said vehicle comprising a control computer adapted to generate, from the video detection signal, a video signal for controlling the trajectory of said vehicle on the traffic lane, said vehicle further comprising a computer suitable for superimposing said signal control video and said detection video signal in a convergence video signal and for determining the convergence of the detection and control video signals on said convergence video signal in order to detect an error in the generation of the control video signal.

Other characteristics and advantages of the invention will appear on reading the detailed description of the embodiments of the invention, given by way of example only, and with reference to the drawings which show:

• Figure 1, a diagram of an image of a detection video signal generated by the method according to the invention in a motor vehicle, • Figure 2, a diagram of an image of a generated control video signal by the method according to the invention, and • FIG. 3, a diagram of an image of a convergence video signal generated from the signals of FIGS. 1 and 2.

In what follows, the embodiments described relate more particularly to an implementation of the method according to the invention in a motor vehicle. However, any implementation in a different context, in particular in any type of vehicle, is also covered by the present invention.

A motor vehicle is conventionally adapted to travel on a traffic lane V. To do this, a driver controls the longitudinal movement of the vehicle, in other words the speed of the vehicle, using brake controls and accelerator, the latter acting on the vehicle's powertrain. The driver also controls the lateral movement of the vehicle, in other words when the vehicle is turning left or right, using the steering system.

The motor vehicle according to the invention also includes an autonomous operating mode. For this purpose, the motor vehicle comprises a module for detecting the environment around the vehicle, and at least one control computer for autonomous driving of the vehicle.

The detection module includes different sensors, for example a radar, a camera, etc. Such sensors can also be sensors used for other vehicle aids, in particular parking aid cameras and the reversing radar, in order to limit the number of elements of the vehicle and thus limit their costs. These sensors make it possible to observe the lane V on which the vehicle is moving, and in particular the lane V towards the front and on the sides of the vehicle when the vehicle is moving forward. The detection module can also recover information related to the vehicle, in particular the speed of movement of the vehicle on taxiway V, the angle of the steering wheel, etc.

The detection module can thus detect obstacles present around the vehicle as well as the ground marking lines 11 delimiting the traffic lane V in which the vehicle moves, as illustrated in FIG. 1.

The detection module thus generates a video signal, designated detection signal 10, shown in Figure 1, representative of the environment around the vehicle. Such a detection signal 10 represents what the vehicle "sees" from its environment. The detection signal 10 is defined in a Cartesian coordinate system linked to the vehicle. Such a mark is fixed and common to all the computers of the vehicle in order to form a reference mark allowing a comparison of signals as will be described later. The detection signal 10 notably comprises the lines 11 of marking on the ground delimiting the traffic lane V. Such a detection signal 10 has the form of an image representing the environment around the vehicle, in particular the traffic lane V towards the front of the vehicle. This image can in particular be reconstituted from images captured by vehicle cameras as described in document FR3047589.

The control computer runs a state machine which makes it possible to autonomously control the vehicle from the detection signal 10 received from the detection module.

In other words, the state machine receives the detection signal 10 and then determines, in the same Cartesian coordinate system, the target 21 of the trajectory that the vehicle must reach, as illustrated in FIG. 2. The state machine controls then the trajectory of the vehicle in order to reach the target 21. For this purpose, the state machine generates a video signal, designated control signal 20, making it possible to control the various organs of the vehicle, in particular the powertrain, the steering, etc. . The control signal 20 has the form of an image representing the taxiway V at the front of the vehicle and a target 21 of the vehicle trajectory. The control signal 20 thus represents the target 21 of the trajectory at the output of the state machine.

The state machine includes different states in each of which the vehicle may be (position on the track, etc.). We can go from one state to another under certain conditions which depend on the detection signal 10 and each state corresponds to a control signal 20. At input, the state machine receives the detection signal 10 which corresponds to the conditions in which vehicle is located. At the output, the state machine sends the control signal 20 corresponding to the new state of the vehicle in order to control its trajectory up to the target 21.

The vehicle further comprises a computer, preferably different from the control computer, adapted to compare the detection signal 10 which enters the state machine with the control signal 20 which leaves the state machine. In other words, the computer compares the video signal at the input of the state machine to the video signal at the output of the state machine in order to detect a possible error and thus increase the reliability of the autonomous control by the state machine. Such a comparison is made easy because the two signals 10, 20 are defined in the same reference frame linked to the vehicle. Such a computer then generates a video signal, designated convergence signal 30, as illustrated in FIG. 3, making it possible to compare the signals 10, 20 and to determine the differences in order to detect any anomaly in the control signal 20, which increases the safety of autonomous driving. The comparison of the detection 10 and control 20 signals will be described in detail later. The convergence signal 30 has the form of an image comprising the lines 11 of marking on the ground of the detection signal 10 and the target 21 of the control signal 20 in order to determine the position of the target 21 relative to the lines 11 of marking .

The vehicle finally comprises display means making it possible to display a visible alert for the driver of the vehicle and / or for the users of the vehicle, in order to prevent the latter when an anomaly is detected.

The reliability process according to the invention will now be described with reference to Figures 1 to 3.

When the vehicle according to the invention is traveling on a taxiway V in autonomous mode, the cameras capture images of the ground and the radars detect if there are obstacles on the taxiway V. The detection module then generates an image of the detection signal 10, representing the lane V on which the vehicle is traveling, and the lines 11 of marking on the ground delimiting said lane V.

This detection signal 10 is sent to the state machine which receives it as input. The state machine then generates an image of the control signal 20 comprising the trajectory target 21 that the vehicle must reach. This thus allows the vehicle to control the movement of the vehicle on the taxiway V so as to follow the target 21 of trajectory.

The vehicle computer receives the detection signal 10 and the control signal 20 and superimposes the images of each of the signals 10, 20 defined in the same reference. The computer then generates an image of the convergence signal 30 from this superposition of images. This image of the convergence signal 30 includes the trajectory target 21 and the ground marking lines 11 defining three zones: a safety zone 31 situated between the lines 11, a zone known as “outside zone 32” situated outside the lines 11 and an uncertainty zone 33 located near the lines 11. The computer then determines, in a step of determining the convergence between the detection signal 10 and the control signal 20, the position of the trajectory target 21 in one of these zones 31, 32, 33 in order to determine whether the target 21 is correct, in other words, whether it is indeed on the traffic lane V between the lines 11 of marking on the ground. This verification ensures that the vehicle is intended to remain between lines 11 of taxiway V.

If the trajectory target 21 is in the safety zone 31, as illustrated in FIG. 3, the signals 10, 20 are convergent, in other words, the trajectory target 21 is in the central zone of the image and no error is then detected.

If the target 21 is outside zone 32, the target 21 is completely off-center from the image and an error is then detected by the computer which can then send an alert message to the driver of the vehicle, in particular by a human interface machine (HMI) of the vehicle, for example by an indicator light or by a display screen.

If the target 21 is in the uncertainty zone 33, the target 21 is slightly offset from the image. The computer then analyzes, in an analysis step, the residue representative of the errors generated during the reliability process. Such an analysis of the residue makes it possible to distinguish an error during the determination of the convergence, from an error during the generation of the control signal 20. Different methods of analysis of the residue can moreover be used, including methods distinct from the determination of the convergence between the detection signal 10 and the control signal 20 described above, for example by video correlation between the detection signal 10 and the convergence signal 30. This analysis makes it possible to determine whether the target 21 is located between the 10 lines 11 or not.

Monitoring the convergence between the detection signal 10 and the control signal 20 of the state machine thus makes it possible to verify the correct functioning of the sensors of the detection module and to make the representation of the environment of the vehicle more reliable. in signals 10, 20. In addition, the use of a third signal (the convergence signal 30) makes it possible to increase the sensitivity of the vehicle to errors using this double control (by the control signal 20 and by the convergence signal 30) and is more robust to noise.

Claims (10)

1. Method for making the control of an autonomous vehicle more reliable during an autonomous driving phase on a traffic lane (V), said vehicle comprising a detection module suitable for capturing images of the environment around said vehicle, said method including: - a step of generating a detection video signal (10) from the images captured by said detection module, said detection video signal (10) determining the traffic lane (V) on which the vehicle is traveling, a step of generation, from the video detection signal (10) generated, of a video control signal (20) of the trajectory of said vehicle on the traffic lane (V), a step of superimposing said control video signal (20) and said detection video signal (10) in a convergence video signal (30), and - a step of determining the convergence of the detection (10) and control (20) video signals on said convergence video signal (30) in order to detect an error in the generation of the control video signal (20). 2. Method according to claim 1, in which the detection video signal (10) comprises the lines (11) for marking on the ground of a taxiway (V) and the obstacles present on said taxiway (V). 3. Method according to one of the preceding claims, in which the detection module comprises at least one camera adapted to capture images of the ground in order to generate the detection video signal (10). 4. Method according to one of the preceding claims, in which the vehicle comprises a control computer executing a state machine adapted to generate the control video signal (20). 5. Method according to one of the preceding claims, in which the detection video signal (10) comprises different zones: a security zone (31), an out-of-zone (32) and an uncertainty zone (33). 6. Method according to one of the preceding claims, wherein the control video signal (20) comprises a target (21) of the vehicle trajectory. 7. The method of claims 5 and 6, wherein the step of determining convergence comprises determining the position of the target (21) of trajectory in one of said areas (31,32, 33). 8. Method according to the preceding claim, comprising a step of alerting the driver of the vehicle when the position of the trajectory target (21) is determined in the out-of-area area (32). 9. Motor vehicle (100) comprising a detection module suitable for capturing images of the environment around said vehicle and generating, from said captured images, a video detection signal (10) determining the traffic lane 10 (V) on which the vehicle travels, said vehicle comprising a control computer adapted to generate, from the detection video signal (10), a video control signal (20) of the trajectory of said vehicle on the taxiway (V), said vehicle further comprising a computer adapted to superimpose said control video signal (20) and said detection video signal (10) in a convergence video signal (30) and to determine the convergence of the video signals of detecting (10) and controlling (20) on said convergence video signal (30) in order to detect an error in the generation of the controlling video signal (20).