IT201800006211A1 - METHOD OF CONTROL OF A VEHICLE EQUIPPED WITH AN AUTONOMOUS DRIVING - Google Patents

Description

DESCRIPTION

of the patent for industrial invention entitled: "METHOD OF CONTROL OF A VEHICLE EQUIPPED WITH AN AUTONOMOUS DRIVE"

TECHNIQUE SECTOR

The present invention relates to a method of controlling a vehicle equipped with autonomous driving.

ANTERIOR ART

Automation to support driving has a very long history and starts in the 1950s, with the automatic gearbox and the first cruise controls. In the 90s, the first adaptive cruise controls began to spread, which represented a real revolution: for the first time a car understood its context, trying to operate in an "intelligent" way, even if only by adjusting the speed. If shuttles and driverless trucks are already on the market, the situation is a bit different for cars: with reference to the SAE classification of driving automation levels, level 1 technologies are now widespread (such as cruise control adaptive) in which the vehicle and the driver share the responsibility of driving while technologies of level 2 (in which the vehicle drives alone, but the driver must remain alert continuously) or level 3 (in which the driver can be distracted, but must be ready to take back the controls) are marketed by some brands in some cases with restrictions on the road sections in which autonomous driving can be operated (for example motorways or roads mapped with high precision). Technologies of level 4 or 5 (in which the driver is no longer required to operate for safety purposes, or is not necessary at all) are mostly the subject of applied research.

However, it has been observed that often the driving choices (particularly, but not limited to, as regards the intensity of accelerations and decelerations) made autonomously by the vehicle (both when the vehicle replaces the human driver, and when the vehicle "corrects ”Human driver driving errors) are perceived by the human driver as wrong as they do not correspond to the human driver's view of driving a vehicle. In other words, the human driver is not able to judge the driving choices made autonomously by the vehicle in an aseptic and objective way, but is obviously bound to judge the driving choices made autonomously by the vehicle based on his own experience, sensitivity, state emotional and therefore it often happens that the human driver believes (wrongly) that the driving choices made autonomously by the vehicle are wrong.

Consequently, when the vehicle completely replaces the human driver (who therefore becomes a "passenger" who must pay attention to driving), the human driver is often not (completely) satisfied with the autonomous driving of the vehicle (for example the exact same type of driving followed by the vehicle may be judged too fast and reckless by an elderly human driver and too slow and cautious by a young human driver); similarly, when the vehicle assists the human driver in some situations while still leaving most of the control to the human driver, the human driver often perceives the intervention of the vehicle as extraneous and dissonant and therefore annoying.

DESCRIPTION OF THE INVENTION

The object of the present invention is to provide a method of controlling a vehicle equipped with autonomous driving which transmits a feeling of safety to the human driver and, at the same time, is not perceived by the human driver as invasive.

According to the present invention, a method of controlling a vehicle provided with autonomous driving is provided, according to what is claimed by the attached claims.

The claims describe preferred embodiments of the present invention forming an integral part of the present description.

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention will now be described with reference to the attached drawing, which illustrates a non-limiting embodiment thereof; in particular, the attached figure is a schematic and plan view of a road vehicle which implements the control method of the present invention.

PREFERRED FORMS OF IMPLEMENTATION OF THE INVENTION

In the attached figure, the number 1 indicates as a whole a road vehicle equipped with two front wheels 2 and two rear driving wheels 3, which receive the drive torque from a propulsion system.

The road vehicle 1 comprises a control unit 4 which is able to operate autonomous driving of level 2-5, or at least in some situations the electronic control unit 4 is able to replace the human driver by performing actions (braking, acceleration, steering) different and independent from the actions of the human driver. The control unit 4 is connected to a series of sensors that detect the status inside the vehicle 1 (for example the position of the steering wheel, the position of the accelerator pedal, the position of the brake pedal, cameras or sensors observation facing the human driver ...) and the state outside the vehicle 1 (e.g. ultrasonic sensors, perimeter cameras, front cameras, medium range radar at the corners of vehicle 1, a long range radar at the front, a scanner front laser ...).

The vehicle 1 comprises a driving station provided with a plurality of driving controls 5 which can be operated by a human driver. For example, the driving controls 5 comprise a steering wheel 6, an accelerator pedal 7, a brake pedal 8, and a transmission control 9 (which, for example, can allow selection of the type of gear desired between forward gear, reverse, park and possibly even allow you to request a gear ratio change).

In addition, the road vehicle 1 comprises actuators that allow you to control the travel of the road vehicle 1 regardless of the action on the driving controls 5 of the driving position; the actuators can mechanically act in parallel to the corresponding driving commands 5 (i.e. the action of the actuators also determines a displacement of the corresponding driving commands 5 as usually occurs for the steering wheel 6 which maintains a mechanical connection with the front steering wheels 2) or the actuators can act mechanically in a completely independent way from the corresponding driving commands 5 (i.e. the action of the actuators does not also determine a displacement of the corresponding driving commands 5 as usually occurs for the pedals 7 and 8 and for the control 9 transmission).

The control unit 4 supervises the operation of the road vehicle 1, implements an autonomous driving software, and is able to act on the actuators to control the motion of the road vehicle 1 in a way at least partially independent of the human driver (i.e. from the action of the human driver on the driving controls 5).

The control unit 4 uses a series of driving parameters to control the driving of the road vehicle 1 autonomously from the human driver; for example, these driving parameters may include: the average / instantaneous acceleration in non-hazardous situations, the average / instantaneous deceleration in non-hazardous situations, jerk ("jerk", i.e. the derivative of acceleration with respect to time third order derivative of position with respect to time) average / instantaneous, the average / instantaneous speed of cornering, and / or the differential between the actual speed and the speed limit. In other words, the variation of the driving parameters determines a predetermined driving style that is more or less fast (therefore more or less sporty) which can favor performance rather than energy efficiency or vice versa.

During a design and implementation phase of the autonomous driving software (which is used by the control unit 4), a series of possible predetermined driving styles are determined; this series of possible predetermined driving styles can contain a very variable number of predetermined driving styles: from a minimum of two up to a maximum of a few hundreds. For each predetermined driving style, the corresponding values of the driving parameters are determined, again during the design and implementation phase of the autonomous driving software, so that each predetermined driving style is characterized by its own and unique "recipe" ( “Mix”) of driving parameter values different (to a greater or lesser extent) from all other predetermined driving styles; in other words, the driving parameters are (more or less) different from a predetermined driving style to a predetermined driving style.

According to a preferred embodiment, the predetermined driving styles can be differentiated according to the context in which the driving takes place: for example, a group of predetermined driving styles for city driving, a group of predetermined driving styles can be provided for driving outside the city, a group of predetermined driving styles for driving on the highway, and a group of predetermined driving styles for driving on the track (generally the groups have a different number of predetermined driving styles as in the various contexts are generally identifiable several numbers of different predetermined driving styles).

During the design and implementation phase of the autonomous driving software (which is used by the control unit 4), an artificial neural network is created and trained that allows to classify a predetermined driving style, i.e. a sample of the driving of the human driver makes it possible to determine which predetermined driving style the human driver driving sample belongs to. The artificial neural network is a mathematical model consisting of a group of information interconnections made up of artificial neurons and processes that use a computational connectionism approach. In most cases the artificial neural network is an adaptive system that changes its structure based on external or internal information flowing through the network during the learning phase. In practical terms the neural network is a non-linear structure of statistical data organized as a modeling tool and is used to simulate complex relationships between inputs and outputs that other analytical functions cannot represent. The artificial neural network receives external signals on a layer of input nodes (processing units), each of which is connected with numerous internal nodes, organized in several levels. Each node processes the received signals and transmits the result to subsequent nodes.

During normal use of the road vehicle 1, the control unit 4 performs an initial learning phase in which the driving control of the road vehicle 1 is completely entrusted to a human driver (obviously with the exception of emergency situations in which there is an imminent threat to the safety of things and people); during this initial learning phase the control unit 4 detects and stores the actions performed by the human driver on the driving commands 5 of the vehicle 1 and / or the consequent dynamics (typically longitudinal speed, longitudinal acceleration, transverse acceleration) of the vehicle 1 for generate at least one sample of the human driver's driving (thus the driving sample consists of a temporal sequence of actions performed by the human driver on driving commands 5 of the vehicle 1 and / or a temporal sequence of vehicle dynamics data).

When the control unit 4 has at its disposal at least one sample of the driving of the human driver, the control unit 4 classifies, using the artificial neural network, the driving style of the human driver among all the possible predetermined driving styles in function the sample of the human driver's guide; that is, the control unit 4 determines, using the artificial neural network, which predetermined driving style among all the possible predetermined driving styles is most similar (corresponding) to the driving style of the human driver. Specifically, the control unit 4 provides the human driver's driving sample as input to the artificial neural network and the artificial neural network provides the predetermined driving style that is closest to the human driver's driving sample.

According to a possible embodiment, the control unit 4 could pre-process the sample of the human driver's guide to normalize and / or synthesize the sample of the human driver's guide. For example, the human driver's driving sample could be pre-processed to determine synthetic indicators that can be input to the neural network; such synthetic indicators of the human driver's driving sample could include: the minimum / average / maximum speed when driving at constant speed, the minimum / average / maximum longitudinal acceleration, the minimum / average / maximum transverse acceleration, the maximum steering angle, minimum / medium / maximum steering wheel rotation speed / acceleration 6, minimum / medium / maximum accelerator pedal pressure 7 speed / acceleration, minimum brake pedal pressure 8 speed / acceleration / average / maximum, the minimum / average / maximum distance from the vehicle in front, and / or the minimum / average / maximum distance from the vehicle following.

In other words, the control unit 4 could pre-process the sample of the human driver's driving to determine synthetic indicators which are then fed into the neural network which determines the predetermined driving style that is closest to the driver's driving sample. human.

As previously mentioned, it is preferable to differentiate the driving styles according to the context (for example urban, suburban, motorway, track); in this case, several artificial neural networks are generally used, each of which is specialized for a single context (for example urban, extra-urban, motorway, track) and also the synthetic indicators that are obtained from the sample of the human driver's guide are generally differentiated according to context (for example, transverse acceleration is essential when driving on the track and generally not very important when driving on the motorway).

Once the control unit 4 has classified the driving style of the human driver (i.e. once the control unit 4 has determined the predetermined driving style that comes closest to the standard of human driver driving), the 'control unit 4 intervenes on the driving control of the vehicle 1 autonomously from the human driver using the driving parameters that distinguish the predetermined driving style that is closest to the driving sample of the human driver (i.e. also according to the classification of the driving style of the human driver).

To sum up, the control unit 4 always intervenes in the best (most effective) way possible (completely regardless of the driving style of the human driver) when an emergency situation arises in which there may be a threat to the safety of things and people . On the other hand, when the control unit 4 can choose (for example whether to go more or less fast always respecting the speed limits and prudence, whether to accelerate / brake more or less smoothly or pronounced, whether or not to admit oversteer cornering when driving on the track ...), the control unit 4 chooses according to the driving style of the human driver (i.e. how the driving style of the human driver has been classified and therefore assimilated to one of the predetermined driving styles) in such a way as to reproduce (in an improved, or more effective and efficient way) the driving style of the human driver to ensure a reassuring and appreciated continuity between what the human driver would do and what the road vehicle 1 actually does. In other words, the intervention of the control unit 4 is "camouflaged" to be as far as possible "invisible" to the human driver who, at best, cannot distinguish between his own actions and the actions of the control unit 4. check; in this way, the human driver never feels like a stranger at the mercy of the road vehicle 1, but always feels (rightly or wrongly) in full control of the situation.

The embodiments described here can be combined with each other without departing from the scope of protection of the present invention.

The control method described above has numerous advantages.

In the first place, the control method described above allows the control unit 4 to intervene on the driving control of the road vehicle 1 with methods which are entirely similar (and therefore "congenial") to the driving style of the human driver; in this way, the human driver recognizes as his own (therefore, from his personal point of view, correct and reassuring) the autonomous driving performed by the control unit 4 both when the control unit 4 takes complete control of the road vehicle 1 , both when the control unit 4 is limited to correcting / integrating the actions of the human driver (especially in this case the human driver who is still driving is not bothered by an intervention of the control unit 4 which, although technically correct and formally flawless, it is in any case alien to one's own sensitivity).

In accordance with the control method described above, when the control unit 4 assists the human driver in driving without replacing the human driver (particularly in performance driving on the track during which the objective is to maximize performance), a possible intervention of the control unit 4 is always “respectful” of the specificities (positive and negative) of the human driver as it is in any case correlated to the driving style of the human driver; therefore the human driver may not really perceive a possible intervention of the control unit 4 (since the intervention takes place in continuity with the driving style of the human driver and therefore can be confused with the driving style of the human driver) or in any case the driver human perceives a possible intervention of the control unit 4 as a positive help to go beyond its own limits (but not beyond the limits of the road car 1, that is to safely approach the limits of the road car 1) and not as an extraneous imposition ( or like the "stick on the hands" of a good but bad teacher).

Furthermore, the control method described above is simple and inexpensive to implement, as it uses physical components that are already present in an autonomous road vehicle, therefore without any additional cost from the "hardware" point of view. Even from a "software" point of view, the control method described above does not require a large amount of computing power, nor a large memory occupation.

LIST OF REFERENCE NUMBERS OF THE FIGURES

1 vehicle

2 front wheels

3 rear wheels

4 control units

5 driving commands

6 steering wheel

7 accelerator pedal

8 brake pedal

9 transmission control

Claims (9)

CLAIMS 1) Method of controlling a vehicle (1) equipped with autonomous driving; the control method includes the steps of: detecting and memorizing, during a learning phase in which the control of the driving of the vehicle (1) is completely entrusted to a human driver, the actions performed by the human driver on the vehicle's driving commands (5) (1) and / or the resulting vehicle dynamics (1) to generate at least one sample of the human driver's driving; determining, during a design and implementation phase, a series of possible predetermined driving styles; classify the driving style of the human driver among all the possible predetermined driving styles according to the sample of the driving of the human driver; and intervene on the control of the driving of the vehicle (1) autonomously by the human driver also according to the classification of the driving style of the human driver. 2) Control method according to claim 1 and comprising the further step of determining for each possible predetermined driving style of the corresponding driving parameters which are different from predetermined driving style to predetermined driving style and are used to control the driving of the vehicle (1) autonomously from the human driver. 3) Control method according to claim 2, wherein the driving parameters which are different from predetermined driving style to predetermined driving style include: the average / instantaneous acceleration in non-hazardous situations, the average / instantaneous deceleration in situations not of danger, the average / instantaneous speed of cornering, and / or the differential between the actual speed and the speed limit. 4) Control method according to claim 1, 2 or 3 and comprising the further step of using an artificial neural network to classify the driving style of the human driver. 5) Control method according to claim 4, in which the artificial neural network is trained during the design and implementation phase. 6) Control method according to claim 4 or 5 and comprising the further steps of: pre-processing the sample of the human driver's guide to determine synthetic indicators; and provide the synthetic indicators in input to the neural network. 7) Control method according to one of claims 1 to 6 and comprising the further steps of: pre-processing the sample of the human driver's guide to determine synthetic indicators; and use the synthetic indicators to classify the driving style of the human driver among all the possible predetermined driving styles. 8) Control method according to claim 7, wherein the synthetic indicators comprise: the minimum / average / maximum speed in case of running at constant speed, the minimum / average / maximum longitudinal acceleration, the minimum / average / transverse acceleration maximum, maximum steering angle, minimum / medium / maximum steering wheel rotation speed / acceleration 6, minimum / medium / maximum accelerator pedal pressure 7 speed / acceleration, pedal pressure speed / acceleration 8 of the minimum / average / maximum brake, the minimum / average / maximum distance from the vehicle in front, and / or the minimum / average / maximum distance from the vehicle following. 9) Control method according to one of claims 1 to 8 and comprising the further steps of: determine, using the data provided by a positioning system, the context in which driving takes place; and differentiate the driving style of the human driver according to the context in which the driving takes place in such a way as to have different possible driving styles in different contexts.