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
  • G06V20/586—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of parking space

Definitions

• the invention relates to the field of parking assistance for a motor vehicle, and more particularly to the determination of a type of parking space.
• a drawback of this solution is that the type of maneuver to be carried out to park the vehicle has not been determined.
• the automatic parking system can implement a parking space maneuver to park the vehicle on several parking spaces in battle. This results in awkward parking of the vehicle.
• Document EP 2 982 572 describes a device for determining the type of a parking space analyzing a marking on the ground delimiting the parking space. Such a solution does not bring full satisfaction insofar as it only works if the parking spaces are well delimited by a marking on the ground and if the marking on the ground is not erased.
• Another solution is to determine the type of a parking space using ultrasonic sensors.
• ultrasound detects objects at a very short distance. As a result, a free parking space is often exceeded when the type of this space is determined.
• the invention aims to overcome the aforementioned drawbacks.
• the invention aims to enable a type of parking space to be determined sufficiently early and even in the absence of markings on the ground delimiting the parking spaces.
• the detection phase is implemented by means of a front camera of the vehicle.
• the front camera of a vehicle is particularly suitable for recognizing targets corresponding to private or commercial vehicles. We can then determine the type of parking space based on vehicles already parked. The presence of markings on the ground delimiting the parking spaces is therefore not necessary. Targets can be located well in front of the vehicle. The type of parking space can therefore be determined well before the vehicle leaves a free space.
• data related to the target is determined, the data related to the target comprising a consecutive target detection count signal, a position of the target, a variable representative of the precision of the position of the target, a speed of the target, an orientation of the target and a variable representative of the precision of the target orientation, the method further comprising a filtering phase between the detection phase and the evaluation phase, the filtering phase comprising at least one filtering step chosen from among an imprecise position filtering step , an imprecise orientation filtering step, a step of filtering moving vehicles and a step of filtering ghost detections.
• the expression “consecutive detection counting signal” is understood to mean a signal which is incremented on each detection of the target and is reset when the target is no longer detected during an implementation.
• the term “filter” will be used in accordance with its usual definition, namely to submit to sorting, and that after filtering, a filtered target can be kept or rejected.
• the target is rejected if the variable representing the accuracy of the position exceeds a position threshold and the target is kept otherwise.
• the target is rejected if the variable representing the orientation accuracy exceeds an orientation threshold and the target is kept otherwise.
• imprecise position and orientation filtering steps reject targets for which data detection is imprecise. Preferably, these steps are only implemented at the start of the detection of a target and are no longer implemented afterwards.
• the variables representative of the accuracy of the position, of the respective orientation are standard deviations of position, of orientation.
• the target is rejected if its speed exceeds a speed threshold and the target is kept smon.
• This filtering step makes it possible to reject the targets corresponding to moving vehicles in order to base the determination of the type of parking space only on parked vehicles.
• this step is applied at all times as long as the target is detected in order to eliminate targets which were initially kept but which are then set in motion, such as vehicles initially stopped at a stop signal or at a traffic light. signaling.
• the target is rejected if the counting signal for the consecutive detection of the target is less than a counting threshold and the target is kept otherwise.
• the ghost detection filtering stage rejects ghost detections carried out by the front camera.
• this step is only implemented for targets located in the field of the front camera of the vehicle and is not implemented for other targets.
• this step is only implemented at the start of the detection of a target and is no longer implemented afterwards.
• a position of the target is determined, the method further comprising a filtering phase between the detection phase and the evaluation phase in which one rejects the target if the distance between the position of the target and the vehicle exceeds a distance threshold and the target is kept otherwise.
• This filtering makes it possible to limit the length of the zone in which targets are considered. This is advantageous since a target distant from the vehicle has a higher probability of being of a different type than that of parking spaces close to the motor vehicle.
• an orientation of the road is determined at the target level and the orientation of the target is relatively assessed relative to the road on the basis of the orientation of the road at the level of the target.
• Using the road orientation at the target level allows a simple assessment of the target orientation in order to determine the type of parking space when the road has a bend.
• the orientation of the road at the target level is determined by taking account of the mark on the ground, if the road does not have any mark on the ground and if the target is behind the vehicle or at the same level as the vehicle, a past trajectory of the vehicle is determined and the orientation of the road at the target is evaluated on the basis of the trajectory passed at the target and , if the road does not have any markings on the ground and if the target is in front of the vehicle, a probable trajectory of the vehicle is determined and the orientation of the road is evaluated at the level of the target on the basis of the probable trajectory at closest point to the target.
• a plurality of targets are detected and, during the determination phase, whatever a target, the target is associated with at least a first term and a second term chosen from: - a niche type of space term,
• a first sum of the first terms of the different targets and a second sum of the second terms of the different targets are calculated, and it is determined whether one of the first and second sums is greater than the other, and preferably if one of the first and second sums is greater than the other added with a strictly positive shift term.
• the type of parking space corresponding to the highest sum has a high probability of being the type of parking space in the vehicle area.
• offset is used to adjust confidence in determining the type of parking space.
• the target is associated with at least one coefficient and, before the calculation of the sums, the terms associated with the target are weighted by the coefficient associated with the target.
• the coefficient may be representative of a distance from the target from the vehicle.
• the coefficient is representative of taking into account a road marking on the ground for the evaluation of the orientation of the target relative to the road.
• the coefficient can be representative of a lateral deviation of the target from a trajectory of the vehicle
• a computer program comprising a code configured to, when executed by a processor or an electronic control unit, implement the method as defined above.
• a device for determining a type of parking space for a motor vehicle, which can be installed in a motor vehicle, said device comprising:
• the detection module being able to be in information link with a front camera of the motor vehicle
• an evaluation module capable of evaluating an orientation of the target relatively with respect to a road
• a determination module capable of determining a type of parking space taking into account an orientation evaluated by the evaluation module.
• FIG. 1 diagrammatically represents a device according to one aspect of the invention
• FIG. 2 is a graph illustrating a map containing term values of the type of parking space as a function of an orientation of the target
• FIG. 3 is a flow diagram showing a process according to another aspect of the invention.
• a motor vehicle 2 has been schematically represented.
• the vehicle 2 is linked to a direct orthonormal vector base 3 conventionally used in automotive design.
• the orthonormal base 7 comprises a vector X, a vector Y and a vector Z perpendicular to the vectors X and Y.
• the vehicle 2 comprises an on-board computer 4 in connection with information with a front camera 6.
• the camera 6 also equips an advanced driving assistance system for the vehicle 2. Such a system is also known by the Anglo-Saxon name "Advanced Driver-Assistance System” or by the corresponding acronym "ADAS".
• the function of the camera 6 is to collect a plurality of images of the environment of the vehicle 2.
• the camera 6 detects images in a field of vision situated in front of the vehicle 2, at a maximum distance of 60 m and on an angular opening contained in a range between 50 ° and 100 ° around the direction of the vector X.
• image is understood in accordance with its usual definition in the optical field, namely as designating a representation of a material object given by an optical system.
• the camera 6 is capable of isolating, from the images collected, targets corresponding to passenger vehicles and commercial vehicles. In particular, the camera 6 does not detect as targets pedestrians, trucks, coaches, motorcycles, bicycles or even street furniture which may be located in its field of vision. For each target, the camera 6 is capable of determining data linked to the target comprising:
• the orientation of a target relative to the vehicle 2 corresponds to the angle between a longitudinal direction of the target and the vector X.
• the camera 6 is also configured to detect the presence of a marking on the ground.
• the camera 6 detects a marking on the ground delimiting on the road a space intended for traffic and a marking on the ground delimiting parking spaces.
• the marking on the ground delimiting a space intended for circulation can include a center line, a line of emergency stop band or a line of delimitation of traffic lane.
• the marking on the ground delimiting parking spaces may include lines forming a "T" in the corners of the parking spaces.
• other cameras can be provided, such as cameras with a wide angle field which make it possible to construct a 360 ° view around the vehicle. Such cameras are commonly designated under the Anglo-Saxon name “Around View Monitoring" or under the corresponding acronym "AVM”.
• the vehicle 2 includes a device for detecting a free parking space (not shown).
• a device for detecting a free parking space (not shown).
• Such a device known per se, is capable of supplying the on-board computer 4 with information on the existence of a free space for parking the vehicle 2.
• the vehicle 2 comprises a device 8.
• the function of the device 8 is to determine a type of parking space. More particularly, the device 8 allows the type of a parking space to be determined even when the parking space is not delimited by a marking on the ground.
• the vehicle 2 can also include a device for determining a type of parking space on the basis of a marking on the ground (not shown).
• the combination of the two devices increases the robustness of the determination when the parking space is delimited by a ground marking and makes it possible to determine when the location is not delimited by a ground marking.
• the device 8 is configured to determine a type of parking space corresponding to a parking space with a parking space, a parking space in battle and a parking space on the cob.
• a type of parking space corresponding to a parking space with a parking space, a parking space in battle and a parking space on the cob.
• the device 8 comprises a detection module 10 in direct or indirect information link with the on-board computer 4 and the camera 6.
• the module 10 collects the data related to the different targets detected by the camera 6.
• the module 10 also collects odometric data from vehicle 2. More specifically, module 10 collects a position of vehicle 2 in a reference frame, an orientation of vehicle 2 in the reference frame, a steering wheel angle, a speed of vehicle 2 and a speed of the heading angle of vehicle 2.
• the detection module 10 collects precision information on the position of the target.
• the position accuracy information is a standard deviation of position.
• This data is directly delivered by the image processing algorithms integrated in the camera 6.
• the standard deviation of the target position corresponds to a standard deviation of the target positions over several acquisitions by the camera 6.
• the detection module 10 also collects information on the precision of the orientation of the target, in this case a standard deviation of the orientations of the target. This data is also delivered by the image processing algorithms integrated in the camera 6.
• the standard deviation of the orientation of the target corresponds to a standard deviation of the orientations of the target over several acquisitions by the camera 6.
• the module 10 includes a counter 1 1. Whatever a target, the counter 1 1 determines a consecutive detection count signal associated with the target.
• the device 8 comprises a filter module 12.
• the function of the module 12 is to filter the targets detected by the camera 6 and collected by the module 10. To this end, the module 12 is in information link with the module 10.
• the module 12 is provided with hardware and software means for implementing steps for filtering the targets collected by the module 10 on the basis of the data received by the module 10.
• the device 8 comprises an evaluation module 14.
• the function of the module 14 is to evaluate, for each target target i collected by the module 10 and not rejected by the module 12, an orientation 0 £ / rte of the target with respect to the road. To do this, the module 14 receives in particular the respective orientations of the targets collected by the module 10. The module 14 also receives the odometric data received by the module 10 and a signal representative of the detection and the form of a floor marking delimiting the space intended for circulation.
• the device 8 comprises a determination module 16.
• the function of the module 16 is to determine, on the basis of the orientations evaluated by the module 14, the type of parking replacement considered.
• the module 16 includes a cartography 18 in which the values of a term of type of slot location, of a term of type of location in battle and of a term of type of location are stored. on the cob depending on an orientation of a target relative to the road.
• the map 18 is shown in FIG. 2.
• the map 18 comprises a first curve 20, in fine dotted lines, corresponding to the values of the term of type of slot location.
• the second curve 22, in thick dotted lines corresponds to the values of the term type of spike location.
• the third curve 24, in solid lines corresponds to the values of the term type of location in battle.
• the curves 20, 22 and 24 are formed by a plurality of affine functions whose images are between - 1 and 1. Regardless of a target, when the mapping 18 issues a slot type term in niche, battle or spike equal to 1, then the target is parked in niche, in battle or spike, respectively.
• the device 8 it is possible to implement a method such as that shown in FIG. 3.
• the method comprises a first phase P l, a second phase P2, a third phase P3 and a fourth phase P4.
• the method is intended to be implemented regularly, for example every second.
• Phase P l comprises a first step E ll during which the module 10 detects targets by means of the front camera 6.
• Each target is denoted target i, where i is an integer.
• a target target i its position relative to vehicle 2 is denoted Xi / veh
• the standard deviation of its positions is denoted a xi
• its speed is denoted v
• the standard deviation of its orientation is noted a gi
• the counting signal associated with this target is noted ⁇ £ .
• ten targets are detected:
• - target l is a vehicle parked in battle, in front and 15 m from vehicle 2,
• - target_2 is a vehicle parked in battle, behind and 15 m from vehicle 2,
• - target_3 is a vehicle parked in battle, in a curve, in front and 25 m from vehicle 2,
• - target_4 is a vehicle parked on the cob, in a curve, behind and 25 m from vehicle 2,
• - target_5 is a vehicle whose x ⁇ rte position is detected with poor detection quality
• - target_6 is a vehicle whose orientation 0 ⁇ rte is detected with poor detection quality
• - target_7 is a vehicle moving in opposite direction relative to vehicle 2
• - target_8 is a parked vehicle that has not been detected two implementations of the process before,
• - target_9 is a vehicle parked forward 45 m from vehicle 2
• step E ll is followed by step E 12.
• the front camera 6 and the module 10 detects the data linked to the target target_i, i.e. the position Xi / veh , the standard deviation a xi , the speed v the orientation 0 i / veh and the standard deviation a gi .
• the signal ⁇ £ delivered by the counter 11 is also collected.
• a monitoring step E 13 whatever a target target i detected during a previous implementation of the method but not detected during the current implementation of the method, we collect previous data being the data related to the target target i detected during the previous implementation.
• updated data x £ / eu £ l , a xi , v 0 i / veh and a gi are determined .
• the previous data is updated taking into account data linked to the vehicle, such as the vehicle travel speed, steering wheel angle, or vehicle drift angle.
• the data a xi and a gi are identical to the corresponding previous data.
• the updated data is stored for a step of monitoring a subsequent implementation of the method.
• Phase P2 includes a step E21 during which the standard deviation a xi is compared to a threshold threshold_l. If the standard deviation a xi exceeds the threshold threshold l, the target target i is rejected during step E21.
• the threshold threshold l is between 2 m and 4 m, preferably substantially equal to 3 m. In this case, the detection of the positions x ⁇ / veh of the target target_5 is of poor quality so that the standard deviation s c 5 exceeds the threshold threshold_l. The target target_5 is rejected during step E21.
• Phase P2 includes a step E22.
• Step E22 implements a filtering of the remaining targets which is similar to the filtering of step E21 taking into account the standard deviation a gi .
• the threshold threshold_2 is between 2 ° and 6 °, and preferably substantially equal to 4 °. In this case, the detection of the orientations of the target target_6 is of poor quality so that the standard deviation s q 6 exceeds the threshold threshold_2. Consequently, the target target_6 is rejected during step E22.
• Phase P2 includes a step E23.
• step E23 whatever i is different from 5 or 6, the speed v t is compared to a threshold threshold_3 and, if the speed v t exceeds the threshold threshold_3, the target target_i is rejected during the step E23.
• the threshold threshold_3 is between 8 km / h and 4 km / h, and preferably substantially equal to 6 km / h.
• phase P2 comprises a step E24 during which we compare, whatever i is different from 5, 6 or 7, the signal ⁇ £ with a threshold of counting threshold_4.
• the target target i is rejected during step E24.
• the threshold threshold_4 is equal to 5.
• the target target_8 was not detected in the second implementation of the method immediately prior to the current implementation. Consequently, the signal ⁇ 8 is equal to 2. The target target_8 is then rejected during step E24.
• Phase P2 includes a step E25 during which, whatever i is different from 5, 6, 7 or 8, the distance between the target target i and the vehicle 2 is determined. The distance determined is then compared to a threshold threshold_5 .
• the threshold threshold_5 is equal to 30 m. If the determined distance exceeds the threshold threshold_5, the target target i is rejected during step E25. In this case, the distance determined for the target target_9 is 45 m. This distance exceeds the threshold threshold_5. The target target_9 is rejected during step E25. Likewise, the distance for the target target l O is 60 m. The target target l O must be rejected during step E25. At the end of step E25, phase P2 is terminated.
• Phase P3 comprises a first step E3 1 for evaluating the orientation of the targets guarded and located on a portion of road in a straight line on which the vehicle 2 is traveling.
• step E3 whatever i is, it is determined whether the target target i is situated along a straight line corresponding to the longitudinal direction of the vehicle 2. If this is the case, the orientation 0 £ / rte is equal to orientation 6 ⁇ te .
• the targets target_l and target_2 were not rejected during phase P2 and are parked along a portion of a straight line on which vehicle 2 is traveling.
• the orientation of target targets 1 and target 2 is defined as follows: (0)
• Phase P3 comprises a second step E32 during which it is determined whether there are targets guarded and parked along a portion of road forming a curve. If the answer is "no", phase P3 is finished. In this case, the targets target_3 and target_4 are parked in curves. The answer to step E32 is therefore "yes".
• step E33 is applied during which it is determined whether the curves on which the targets are located have markings on the ground. Camera 6 and module 10 can be used for this purpose. If the answer to step E33 is "yes”, step E34 is applied. If the answer to step E33 is "no”, step E35 is applied.
• step E34 an orientation of the road is determined on the basis of the ground marking. More precisely, whatever a target target i located in a curve, the orientation of the road corresponds to the orientation of the road at the point closest to the target target i relative to the vector base 3. We then pass at a step E38 which will be described later.
• step E35 a past trajectory of the vehicle 2 is collected.
• the past trajectory corresponds to the last thirty meters traveled by the vehicle.
• a memory not shown incorporated in the on-board computer 4 memorizing the trajectory of the vehicle 2.
• Step E35 is followed by a step E36 during which a probable trajectory of the vehicle 2 is anticipated.
• the speed v 2 / rte of the vehicle 2 the angle at the steering wheel is collected a 2 of vehicle 2 and the drift angle b 2 of vehicle 2.
• the probable trajectory is defined by assuming that the speed v 2 / rte , the angle a 2 and the angle b 2 remain constant over the set of the likely trajectory.
• step E37 an orientation of the road is defined at the target level as being a tangent to the past trajectory or to the probable trajectory of vehicle 2 at the point closest to the target.
• the target target_4 was not rejected during phase P2 and is parked in a curve behind the vehicle 2.
• the orientation of the road at the level of target target_4 is directed by the tangent to the trajectory passed to the point closest to the target target_4.
• the target target_3 is located on a curve in front of the vehicle 2 and devoid of markings on the ground.
• the orientation of the road at the target target_3 is directed by the tangent to the probable trajectory at the point closest to the target target_3.
• step E38 we calculate, whatever i corresponds to a target target i kept during phase P2 and situated along a curve, the orientation 0 £ / rte .
• the orientation of the road is used at the level of the target determined during step E34 or during step E37.
• the targets target_3 and target_4 are located on curves without marking on the ground. The orientation determined during step E37 is therefore used.
• phase P3 is finished.
• phase P3 implements two different modes of evaluation of the orientation of the targets relative to the road, depending on whether the target is located on a portion of a straight line on which the vehicle is traveling, or not. . It is of course without departing from the scope of the invention to envisage using only one of these two modes of evaluation.
• phase P3 only includes step E3 1. Such a variant is advantageous in that its design is simpler and that it works on straight lines as in a conventional parking lot.
• phase P3 only includes steps E33 to E38. If such a variant is more complicated, it makes it possible to take account of the orientation of the road in order to take into account targets situated in bends and to take a deviation of the orientation of the vehicle relative to the road.
• Phase P4 includes a first step E41 of reading in the mapping 18. More precisely, whatever i corresponding to a target target i not rejected during phase P2, the input of the mapping 18 orientation 0 £ / rte and we read a term t_Ci on the curve 20, a term t_bi on the curve 22 and a term t_ei on the curve 24.
• the term t_Ci is equal to 1 if the target target_i is stationed in slot and at - 1 otherwise.
• the terms t_bi and t_ei are equal to 1 if the target target i is stationed in battle and on the cob, respectively. In this case, the target targets l, target_2 and target_3 are stationed in battle while the target target_4 is stationed on the cob.
• Phase P4 includes a step E42 of calculating a coefficient e di representative of a distance from the target relative to the vehicle.
• the coefficient e di is determined as a decreasing function of the distance between the target target i and the vehicle 2.
• the coefficient e di is equal to 1 if the distance between vehicle 2 and target target i is less than 20 m and 0.25 if the distance between target target i and vehicle 2 is more than 20 m. In this case :
• Phase P4 includes a step E43 of determining a coefficient c_0i representative of taking into account a road marking on the ground for the evaluation of the orientation of the road at the level of the target.
• the coefficient c_0i is equal to 0.25 if the orientation of the road at the level of the target target i has been determined during from step E37 or equal to 1 otherwise. In this case :
• Phase P4 includes a step E44 during which a coefficient c_ei representative of a lateral deviation of the target from a trajectory of the vehicle is determined.
• a lateral difference is determined between the trajectory of the vehicle 2 and the position of a target. If, whatever i corresponds to a target target i not rejected during phase P2, the lateral deviation for the target target i is less than a threshold threshold_6, the coefficient c_ei is equal to 1. Otherwise, the coefficient c_ei is equal to 0.25.
• the threshold threshold_6 is equal to 1.5 times the length of the vehicle 2.
• the trajectory of the vehicle 2 considered may be the past trajectory and / or the probable trajectory determined during steps E35 and E36.
• the target targets l, target_2 and target_3 are parked along the road while the target target_4 is parked in MMMans a back alley.
• the lateral distance between the trajectory of the vehicle 2 and the target target_4 therefore exceeds the threshold threshold 6:
• Phase P4 comprises a step E45 1 during which the terms t_bi, t_ei, t_Ci are weighted with the different coefficients c_di, c_0i and c_ei, and a step E452 during which the weighted terms are summed. More precisely, during steps E45 1 and E452, three sums are calculated
• Phase P4 includes a step E46 of determining the type of parking space.
• step E46 the calculated sums are compared. More precisely, for each sum ⁇ b, ⁇ e or ⁇ c, we calculate a shifted sum ⁇ b ⁇ e 'or ⁇ c' by adding a term offset t decaiage strictly positive. In the example illustrated, the term t decaiage is equal to 1. If there is one of the three sums ⁇ b, ⁇ e or ⁇ c greater than the offset sums calculated from the other two sums, then the type of parking space corresponds to the parking space of this sum:
• the location is ear type.
• the sum ⁇ b is greater than each of the sums ⁇ e and ⁇ c plus the term t decaiage .
• the parking space is therefore of the battle type.
• the invention makes it possible to determine the type of a parking space reliably long before the vehicle exceeds a free parking space which operates even in the absence of markings on the ground delimiting the parking spaces. parking.

Landscapes

• Engineering & Computer Science (AREA)
• Physics & Mathematics (AREA)
• General Physics & Mathematics (AREA)
• Multimedia (AREA)
• Theoretical Computer Science (AREA)
• Traffic Control Systems (AREA)
• Control Of Driving Devices And Active Controlling Of Vehicle (AREA)

Applications Claiming Priority (2)

Publications (1)

Family

ID=65200955

Family Applications (1)

Country Status (7)

Families Citing this family (1)

Family Cites Families (30)

• 2018
  • 2018-07-31 FR FR1857135A patent/FR3084628B1/fr active Active
• 2019
  • 2019-07-08 EP EP19737527.2A patent/EP3830740A1/de active Pending
  • 2019-07-08 CN CN201980047639.6A patent/CN112513872A/zh active Pending
  • 2019-07-08 US US17/264,069 patent/US11816903B2/en active Active
  • 2019-07-08 WO PCT/EP2019/068259 patent/WO2020025260A1/fr unknown
  • 2019-07-08 KR KR1020217004780A patent/KR20210034626A/ko unknown
  • 2019-07-08 JP JP2021504764A patent/JP7480115B2/ja active Active

Also Published As

Similar Documents

Legal Events