EP3931741A1 - Vehicle driving assistance by reliable determination of objects in deformed images - Google Patents
Vehicle driving assistance by reliable determination of objects in deformed imagesInfo
- Publication number
- EP3931741A1 EP3931741A1 EP20709283.4A EP20709283A EP3931741A1 EP 3931741 A1 EP3931741 A1 EP 3931741A1 EP 20709283 A EP20709283 A EP 20709283A EP 3931741 A1 EP3931741 A1 EP 3931741A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- vehicle
- determined
- deformation
- interest
- driving
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
- 238000013528 artificial neural network Methods 0.000 claims abstract description 32
- 238000000034 method Methods 0.000 claims abstract description 25
- 238000012549 training Methods 0.000 claims description 14
- 238000001514 detection method Methods 0.000 claims description 12
- 230000007613 environmental effect Effects 0.000 claims description 7
- 230000006870 function Effects 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 238000004590 computer program Methods 0.000 claims description 3
- 239000013256 coordination polymer Substances 0.000 description 10
- 238000004364 calculation method Methods 0.000 description 5
- 241001465754 Metazoa Species 0.000 description 2
- 241000894007 species Species 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/255—Detecting or recognising potential candidate objects based on visual cues, e.g. shapes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
Definitions
- TITLE ASSISTANCE IN DRIVING A VEHICLE, BY RELIABLE DETERMINATION OF OBJECTS IN DEFORMED IMAGES
- the invention relates to land vehicles with at least partially automated (or autonomous) driving and possibly of the automobile type, and more specifically the assistance in driving such vehicles.
- a land vehicle is at least partially automated (or autonomous) when it can be driven on a road in an automated (partial or total (without intervention of its driver)) during a phase of automated driving, or manually (and therefore with intervention by the driver on the steering wheel and / or the pedals) during a manual driving phase.
- the automated (or autonomous) driving of a vehicle may consist in partially or totally steering this vehicle or in providing any type of assistance to a natural person driving this vehicle. This therefore covers any automated (or autonomous) driving, from level 1 to level 5 in the OICA (International Organization of Automobile Manufacturers) scale.
- the term “land vehicle” here means any type of vehicle that can travel on a road, and in particular a motor vehicle, a utility vehicle, a moped, a motorcycle, a minibus, a coach, a storage robot in a warehouse. , or a road machine.
- At least one image acquired by at least one sensor of the vehicle in the environment in front of it, to determine if it contains at least one object (such as another vehicle or a pedestrian), then
- this type of assistance process independently detects objects and traffic lanes which it then associates by working in another landmark (typically that of the vehicle).
- the analysis and the determinations are frequently carried out by means of a neural network which has previously been the subject of an object detection training by means of a database storing images of the environment (generally associated to data that is representative of the known objects they contain).
- the images stored in this type of database are acquired by means of on-board cameras in laboratory vehicles and introducing very little or no deformation, and therefore these images can be considered as undistorted.
- most of the cameras that equip “normal” vehicles acquire images that are distorted, generally due to their wide viewing angle. Consequently, when one analyzes with the aforementioned neural network, on board a vehicle, the deformed images acquired with its on-board camera, they do not correspond exactly to the non-deformed images having been used for learning (and resulting from the database), and therefore the positions determined in these distorted images may be slightly different from the actual positions, so that in crowded environments an object can sometimes be mistaken for another neighboring object.
- a neural network which has previously been the subject of object detection training by means of an "undeformed" database (that is to say containing images (and any data representative of the objects they contain) that have not undergone deformation identical to that introduced by its on-board camera), and to apply to each image acquired (having a first deformation due to its camera), a second deformation, the reverse of the first deformation, in order to obtain an undeformed image which is then supplied to the neural network so that it detects areas of interest each containing an object.
- the first deformation is applied to each detected area of interest, then each object contained in an area of interest identical to an area of interest which has just undergone the first deformation is detected in the acquired image.
- This second solution certainly makes it possible to use the same database (not distorted) and the same neural network, but it requires a large amount of computation and a large additional memory.
- One of the aims of the invention is therefore to improve the situation.
- an assistance method intended to assist the driving of an at least partially autonomous driving vehicle, traveling on a portion of a road comprising a number N of traffic lanes each defined by two delimitations, with N> 1, and comprising at least one sensor acquiring images, having a first deformation, of an environment at least in front of the vehicle.
- a third step in which a driving instruction configured to adapt the driving of the vehicle is generated, according to the information determined for each object contained in a determined area of interest. Thanks to this application of the first and second deformations on the only areas of interest, it is possible to significantly limit the quantity of calculations to be carried out and additional memory capacity.
- the assistance method according to the invention may include other characteristics which can be taken separately or in combination, and in particular:
- the neural network can perform each analysis after having been the object of an object detection training by means of a database storing non-deformed environmental images and associated with representative data known objects which they contain and images of known objects having the first deformation and associated with data representative of their deformed objects;
- the image obtained can be analyzed with a first part of the neural network, adapted by learning to determine areas of interest, and each intermediate area of interest can be analyzed with a second part of the network of neurons, adapted by training to the determination of non-deformed objects;
- the driving instruction can be generated according to the relative position of each traffic lane determined for each determined object and the relative position of each determined object in its traffic lane.
- the invention also proposes an assistance device intended to assist the driving of an at least partially autonomous driving vehicle, traveling on a portion of a road comprising a number N of traffic lanes each defined by two delimitations, with N > 1, and comprising at least one sensor acquiring images, having a first deformation, of an environment at least in front of the vehicle.
- This assistance device is characterized by the fact that it comprises at least one processor and at least one memory arranged to perform the operations. consists in :
- - generate a driving instruction configured to adapt the driving of the vehicle, according to the information determined for each object contained in a determined area of interest.
- the invention also proposes a computer program product comprising a set of instructions which, when it is executed by processing means, is suitable for implementing an assistance method of the type of that presented above for adapt the driving of a vehicle to at least partially autonomous driving.
- the invention also proposes a vehicle, optionally of the automobile type, with at least partially autonomous driving, traveling on a portion of a road comprising a number N of traffic lanes each defined by two delimitations, with N> 1, and comprising at least at least one sensor acquiring images, having a first deformation, of an environment at least in front of the vehicle and an assistance device of the type of that presented above.
- FIG. 1 illustrates schematically and functionally a vehicle located on one of the two traffic lanes of a portion of the road and equipped with a sensor and a computer comprising an example of an assistance device according to the invention
- FIG. 2 schematically illustrates an example of an algorithm implementing an assistance method according to the invention
- FIG. 3 illustrates schematically and functionally an exemplary embodiment of an assistance device according to the invention.
- the object of the invention is in particular to provide a method of driving assistance, and an associated DA driving assistance device, intended to allow assistance in driving a vehicle V1 with automated (or autonomous driving) driving. ).
- the automated (or autonomous) driving vehicle V1 is of the automobile type. This is for example a car, as illustrated without limitation in Figure 1. But the invention is not limited to this type of vehicle. It relates in fact to any type of land vehicle with at least partially automated driving and able to travel on land traffic routes. Thus, it could also be a utility vehicle, a moped, a motorcycle, a minibus, a bus, a storage robot in a warehouse, or a storage vehicle. roads, for example.
- the first taxiway VC1 is here separated from the right RS safety railing by an emergency lane BAU.
- this first traffic lane VC1 is delimited by two delimitations di and 02, and the second traffic lane VC2 is delimited by two delimitations 02 and d3.
- the invention relates to any portion of road comprising at least one traffic lane. Therefore, the number N of lanes of a road section can take any value greater than or equal to one (ie N> 1).
- This first vehicle V1 comprises at least one CP sensor and an exemplary embodiment of a DA driving assistance device according to the invention.
- This CP sensor is responsible for acquiring digital images in the environment which is located at least in front of the first vehicle V1. Furthermore, this CP sensor introduces a first known deformation d1 in the environmental images that it acquires.
- the number of CP sensors is here equal to one (1). But this number can take any value greater than or equal to one (1), as long as this makes it possible to acquire digital images in the environment located at least in front of the first vehicle V1.
- the sensor CP can comprise at least one camera installed in a front part of the vehicle (for example on the windshield or on the interior rear-view mirror), and responsible for acquiring digital images at least in front of the first vehicle V1 and possibly on at least part of its two lateral sides.
- the first vehicle V1 can also include at least one ultrasonic sensor and / or at least one radar or lidar installed at the front or at the rear or on at least one lateral side of the first vehicle V1, and / or at least one camera installed in a rear part.
- the invention notably provides a driving assistance method intended to allow driving assistance of the first V1 vehicle.
- This (driving) assistance method can be at least partially implemented by the (driving) assistance device DA which comprises for this purpose at least one processor PR, for example a digital signal (or DSP ( “Digital Signal Processor”)), and MD memory.
- the processor PR for example a digital signal (or DSP ( “Digital Signal Processor”)
- MD memory for example a digital signal (or DSP ( “Digital Signal Processor”)
- the assistance device (driving) DA forms part of a computer CA of the first vehicle V1. But it is not compulsory. Indeed, the assistance device DA could include its own computer. Therefore, the device DA assistance can be implemented in the form of a combination of circuits or electrical or electronic components (or “hardware") and software modules (or “software”).
- the memory MD is live in order to store instructions for the implementation by the processor PR of the assistance method.
- the CA computer can be dedicated to complete control of the driving of the first V1 vehicle during each phase of automated (or autonomous) driving.
- the assistance method comprises three steps which for at least one of them requires the presence of a neural network RN which has been the subject of a prior object detection learning by means of a database which stores at least non-deformed environmental images and known object images having the first deformation d1 due to the sensor CP.
- This database can therefore be considered as a normal database (that is to say, storing at least non-deformed environmental images) to which images of known objects having the first deformation d1 have been added.
- These known objects (or obstacles) are, for example, vehicles of different types or pedestrians or animals of different species and / or sizes or inanimate objects of different types that can sometimes be encountered on a traffic lane VCk (or more generally on a section of road R).
- the RN neural network can equip a vehicle which does not include the latter.
- the neural network RN can form part of the processor PR.
- a first step 10 of the assistance method we (the PR processor and the MD memory) begin by obtaining at least one image of the environment of the first vehicle V1 acquired with the sensor CP.
- a second step 20-70 of the assistance method we (the processor PR and the memory MD) begin (s) by analyzing in a first sub-step 20 the image obtained with the neural network RN, in order to attempt of determining at least one zone of interest ZI comprising a known object (or obstacle).
- one can carry out a test to determine whether at least one zone of interest ZI, comprising an object (or obstacle) known, was determined during the analysis.
- a fourth sub-step 50 of the second step 20-70 we (the processor PR and the memory MD) analyze (s) the (each) intermediate zone of interest Z11 with the neural network RN in order to determine an undistorted OND object that it contains.
- a sixth substep 70 of the second step 20-70 one (the processor PR and the memory MD) detects (s) each deformed object OD in the image obtained in the first step 10, and detemninates (s) ) information that represents each deformed object OD.
- the determination of each OD deformed object in the image obtained in the first step 10 being very reliable, the information representing each OD deformed object is therefore also very reliable.
- the processor PR and the memory MD can determine in the image obtained the traffic lane VCk on which a determined object is located and the relative position of this determined object in its traffic lane VCk. Then, the PR processor and the MD memory can determine in the image obtained the relative position of each traffic lane determined for each object determined with respect to the traffic lane VC1 on which the first vehicle V1 is traveling. These different positions constitute information representative of each deformed object OD.
- the processor PR and the memory MD can, for example, determine each traffic lane VCk on which a determined object is located as a function of its boundaries (here di and 02, or 02 and d3) detected on the image obtained or defined by data from a road map representative of the road portion R.
- the latter (R) is determined at each instant as a function of the current position of the first vehicle V1 which is supplied by a device for assisting navigation DN which is present in the first vehicle V1 permanently (as illustrated in Figure 1 without limitation), or temporarily (when it is part of a portable equipment or a smart mobile phone ( or smartphone ”) or a tablet accompanying a passenger).
- Road mapping can be part of the aforementioned DN navigation aid device.
- the road map data is stored in a memory of the navigation aid device DN or on a storage medium (such as a CDROM, for example) installed in a player of the first vehicle V1.
- the road map data can be downloaded from a server accessible over the air via a communication module fitted to the first vehicle V1, at the request of the DA assistance device.
- a third step 80 of the assistance method we (the processor PR and the memory MD) generate (s) a driving instruction which is configured to adapt the driving of the first vehicle V1, as a function of the information determined in the sixth sub-step 70 for each object contained in a determined area of interest ZI.
- the driving instruction can be determined according to the relative position of each traffic lane VCk determined for each determined object and the relative position of each determined object in its travel lane VCk.
- the processor PR detects in the image obtained the presence of a second vehicle V2 in front of the first vehicle V1 and on the same traffic lane VC1 as that on which the latter is traveling (V1), and a third vehicle V3 in front of the first vehicle V1 and on the traffic lane VC2 which is placed to the left of that on which circulates the latter (V1).
- the analysis of successive images can, for example, make it possible to determine that the third vehicle V3 is passing the second vehicle V2.
- the instruction generated can request the adaptation of the speed of the first vehicle V1 so that it follows the second vehicle V2 at least as long as the third vehicle V3 has not finished passing it.
- the invention offers the advantages of the first and second solutions of the prior art (presented in the introductory part) but not the main disadvantages of the latter.
- the learning of the neural network RN can be done by means of a database (comprising at least non-deformed images and images of objects having the first deformation d1 induced by the sensor CP of the vehicle V1) , the weight of which is little increased compared to that of a database of the prior art, which facilitates its implementation.
- the neural network RN can perform each analysis after having been the object of an object detection training by means of a database which stores images of non-environment. deformed and associated with data (or annotations) representative of the known objects they contain and images of known objects having the first deformation d1 and associated with data (or annotations) representative of their distorted objects.
- data or annotations
- associated data can be, for example, types of object (vehicles of different types, pedestrians, animals of different species and / or sizes, inanimate objects of different types), or outlines of objects.
- the neural network RN can comprise a first part P1 adapted by training to the determination of the areas of interest, and a second part P2 adapted by training to the determination of objects. undistorted OND.
- the processor PR and the memory MD can analyze the image obtained in the first step 10 with the first part P1 of the neural network RN (in order to determine zones of interest ZI), and analyze each intermediate zone of interest Z11 with the second part P2 of the neural network RN in order to determine the non-deformed objects OND.
- the invention also proposes a computer program product comprising a set of instructions which, when it is executed by processing means of the electronic circuit (or hardware) type, such as for example the processor PR, is suitable for implementing the driving assistance method described above to assist driving the first vehicle V1.
- the assistance device DA is very schematically illustrated with only its random access memory MD and its processor PR which can comprise integrated circuits (or printed), or else several integrated circuits (or printed) connected by wired or wireless connections.
- integrated (or printed) circuit is meant any type of device capable of performing at least one electrical or electronic operation.
- the assistance device DA can also include a mass memory MM, in particular for the storage of the acquired image data and of the intermediate data involved in all its calculations and processing.
- this DA assistance device can also include an input interface IE for receiving at least the acquired image data, any data position of the first vehicle V1, and any road map data, for use in calculations or processing, possibly after having shaped and / or demodulated and / or amplified, in a manner known per se, by means of a digital signal processor PR '.
- this assistance device DA can also include an output interface IS, in particular for the transmission of the instructions that it generates.
- the driving assistance method can be implemented by a plurality of digital signal processors, random access memory, mass memory, input interface, output interface.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
FR1902003A FR3093054B1 (en) | 2019-02-27 | 2019-02-27 | ASSISTANCE IN DRIVING A VEHICLE, BY RELIABLE DETERMINATION OF OBJECTS IN DEFORMED IMAGES |
PCT/FR2020/050200 WO2020174142A1 (en) | 2019-02-27 | 2020-02-05 | Vehicle driving assistance by reliable determination of objects in deformed images |
Publications (1)
Publication Number | Publication Date |
---|---|
EP3931741A1 true EP3931741A1 (en) | 2022-01-05 |
Family
ID=66867537
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
EP20709283.4A Withdrawn EP3931741A1 (en) | 2019-02-27 | 2020-02-05 | Vehicle driving assistance by reliable determination of objects in deformed images |
Country Status (3)
Country | Link |
---|---|
EP (1) | EP3931741A1 (en) |
FR (1) | FR3093054B1 (en) |
WO (1) | WO2020174142A1 (en) |
-
2019
- 2019-02-27 FR FR1902003A patent/FR3093054B1/en active Active
-
2020
- 2020-02-05 WO PCT/FR2020/050200 patent/WO2020174142A1/en unknown
- 2020-02-05 EP EP20709283.4A patent/EP3931741A1/en not_active Withdrawn
Also Published As
Publication number | Publication date |
---|---|
WO2020174142A1 (en) | 2020-09-03 |
FR3093054A1 (en) | 2020-08-28 |
FR3093054B1 (en) | 2021-05-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2016042106A1 (en) | Localisation and mapping method and system | |
EP3980308B1 (en) | Method for calculating the lateral position of a motor vehicle | |
WO2015121546A1 (en) | Method for determining a speed limit in force on a road taken by a motor vehicle | |
EP3931056B1 (en) | Regulation of the speed of a vehicle when overtaking on a bend | |
FR3120040A1 (en) | Method for activating an obstacle avoidance system for a motor vehicle | |
EP3931741A1 (en) | Vehicle driving assistance by reliable determination of objects in deformed images | |
EP3938949B1 (en) | Training of a neuronal network, for assistance in driving a vehicle by determining lane boundaries which are difficult to observe | |
EP3877228B1 (en) | Construction by segmentation of virtual lanes on a carriageway | |
FR3092545A1 (en) | ASSISTANCE IN DRIVING A VEHICLE, BY DETERMINING THE TRAFFIC LANE IN WHICH AN OBJECT IS LOCATED | |
WO2017140969A1 (en) | Device and method for estimating the attention level of a driver of a vehicle | |
FR3082044A1 (en) | METHOD AND DEVICE FOR DETECTING THE TRACK ON WHICH A VEHICLE IS MOVING, ACCORDING TO THE DETERMINED DETERMINATIONS | |
FR3093311A1 (en) | ASSISTANCE IN DRIVING A VEHICLE, BY DETERMINING THE CONTROLLABILITY OF ITS DRIVING | |
FR3106108A1 (en) | Method and device for determining the trajectory of a road | |
FR3084628A1 (en) | METHOD FOR DETERMINING A TYPE OF PARKING LOCATION | |
WO2021249809A1 (en) | Method for calibrating a camera and associated device | |
FR3139227A1 (en) | Method and device for detecting a nearby object to control driving assistance for an autonomous vehicle. | |
FR3082935A1 (en) | ANALYSIS METHOD AND DEVICE FOR DETERMINING A CLOTHOIDE REPRESENTING AN ELEMENT OF A ROAD ON WHICH A VEHICLE IS MOVING | |
WO2019220030A1 (en) | Method and device for assisting the automated driving of a vehicle by determining the safety of a proposed trajectory | |
FR3100641A1 (en) | DETERMINATION OF REAL-TIME ENVIRONMENTAL INFORMATION AND SELECTIVE REGRESSION, FOR A SYSTEM | |
FR3036498A1 (en) | METHOD AND SYSTEM FOR ONLINE LOCATION OF A MOTOR VEHICLE | |
FR3137780A1 (en) | Method and device for determining a rear route of at least one lateral delimitation of a traffic lane | |
EP4227912A1 (en) | Method for detecting a central line of a road | |
FR3116501A1 (en) | Method and device for determining a virtual boundary line between two traffic lanes. | |
FR3100016A1 (en) | Method and system for managing a movement of a motor vehicle | |
FR3135240A1 (en) | Method and device for controlling a vehicle driving assistance system based on the detection of a traffic sign and a lane change |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: UNKNOWN |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE |
|
PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE |
|
17P | Request for examination filed |
Effective date: 20210716 |
|
AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR |
|
DAV | Request for validation of the european patent (deleted) | ||
DAX | Request for extension of the european patent (deleted) | ||
RAP3 | Party data changed (applicant data changed or rights of an application transferred) |
Owner name: STELLANTIS AUTO SAS |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: EXAMINATION IS IN PROGRESS |
|
17Q | First examination report despatched |
Effective date: 20240105 |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN |
|
18D | Application deemed to be withdrawn |
Effective date: 20240507 |