WO2014003168A1 - 車両に搭載される画像解析装置 - Google Patents

車両に搭載される画像解析装置 Download PDF

Info

Publication number
WO2014003168A1
WO2014003168A1 PCT/JP2013/067818 JP2013067818W WO2014003168A1 WO 2014003168 A1 WO2014003168 A1 WO 2014003168A1 JP 2013067818 W JP2013067818 W JP 2013067818W WO 2014003168 A1 WO2014003168 A1 WO 2014003168A1
Authority
WO
WIPO (PCT)
Prior art keywords
vehicle
speed
learning
acceleration
vanishing point
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.)
Ceased
Application number
PCT/JP2013/067818
Other languages
English (en)
French (fr)
Japanese (ja)
Inventor
広樹 中野
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Denso Corp
Original Assignee
Denso Corp
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Denso Corp filed Critical Denso Corp
Priority to US14/411,113 priority Critical patent/US20150294453A1/en
Publication of WO2014003168A1 publication Critical patent/WO2014003168A1/ja
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/60—Control of cameras or camera modules
    • H04N23/64—Computer-aided capture of images, e.g. transfer from script file into camera, check of taken image quality, advice or proposal for image composition or decision on when to take image
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/70—Determining position or orientation of objects or cameras
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10004—Still image; Photographic image
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10016—Video; Image sequence
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30244—Camera pose
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30248—Vehicle exterior or interior
    • G06T2207/30252—Vehicle exterior; Vicinity of vehicle
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30248—Vehicle exterior or interior
    • G06T2207/30252—Vehicle exterior; Vicinity of vehicle
    • G06T2207/30256—Lane; Road marking

Definitions

  • the present invention relates to an image analysis apparatus mounted on a vehicle, and more particularly to an image analysis apparatus mounted on a vehicle that performs image analysis based on the position of a vanishing point.
  • Patent Document 1 An example of such a system is shown in Patent Document 1.
  • the system according to Patent Document 1 analyzes a captured image data obtained from an in-vehicle camera and calculates a position of a vanishing point (FOE: Focus of Expansion), thereby estimating the attitude of the in-vehicle camera.
  • FEE Focus of Expansion
  • a vanishing point is a point where parallel straight line groups gather in perspective and perspective views.
  • in-vehicle system it is possible to calculate, for example, the traveling state of the vehicle relative to the road and the distance to the vehicle traveling ahead by analyzing the captured image data in consideration of the attitude of the in-vehicle camera. it can.
  • the vanishing point position For example, by extracting an edge with a steep change in luminance from the captured image data, the area of the road marking line (white line, botsdots, etc.) reflected in the captured image data is estimated. Then, the intersection position of two straight lines obtained by linear approximation of the edge corresponding to the road marking line is calculated. The vanishing point candidate is calculated, for example, by the weighted time average of the intersection position.
  • vanishing point candidates for example, accuracy evaluation is performed by comparison with vanishing point positions learned in the past, and candidates with low accuracy as vanishing points are rejected. And the candidate which was not rejected is employ
  • the learned vanishing point position information is used, for example, when estimating a likely edge as a road marking line.
  • the vehicle 100 may be placed on the chassis dynamometer 200 and simulated as shown in FIG.
  • the vanishing point position is learned, and there is a possibility that the vanishing point position may be erroneously learned.
  • the vanishing point position is erroneously estimated by erroneously estimating the dirt on the wall 210 or the shadow of the surrounding building reflected on the wall 210 as a road marking line. Learning can occur.
  • an image analysis device mounted on a vehicle includes a camera, a learning unit, and a control unit.
  • the camera captures an area in front of the vehicle and generates image data representing the captured image.
  • the learning means analyzes the image data generated by the camera and learns the vanishing point position.
  • the control means controls on / off of the vanishing point position learning operation by the learning means based on the output of the inertial sensor provided in the vehicle.
  • the learning means can be configured to learn the vanishing point position based on, for example, the estimation result of the road marking line reflected in the image data.
  • the control means determines whether or not the vehicle is traveling on the road based on the output of the inertial sensor and determines that the vehicle is traveling on the road.
  • the learning operation can be switched from off to on.
  • the vanishing point position is learned while the vehicle is simulating on the chassis dynamometer.
  • the learning operation can be controlled on / off so that there is no such problem. As a result, it is possible to suppress erroneous learning of the vanishing point position due to the learning of the vanishing point position during the simulated running of the vehicle.
  • an acceleration sensor can be used, and the control means learns on the condition that the vehicle speed calculated by integrating the acceleration of the vehicle specified from the output of the acceleration sensor exceeds the reference speed, for example.
  • the operation can be switched from off to on.
  • control means can be configured to control on / off of the vanishing point position learning operation by the learning means based on the output of the wheel speed sensor provided in the vehicle in addition to the output of the inertia sensor.
  • control means may be based on an error between the vehicle acceleration calculated from the deviation of the vehicle speed specified from the output of the wheel speed sensor and the vehicle acceleration specified from the output of the acceleration sensor, or from the output of the wheel speed sensor. Based on the error between the specified vehicle speed and the vehicle speed calculated by integrating the vehicle acceleration specified from the output of the acceleration sensor, the learning operation is turned on from off on condition that the error is less than the reference. It can be configured to switch to.
  • the control means performs the learning operation on the condition that the speed of the vehicle specified from the output of the wheel speed sensor exceeds the reference speed and that the error is less than the reference.
  • control means learns on the condition that the vehicle speed specified from the output of the wheel speed sensor exceeds the reference speed and the vehicle acceleration specified from the output of the acceleration sensor exceeds the reference acceleration.
  • the operation can be switched from off to on. Also by this control method, the vanishing point position learning operation is appropriately executed.
  • FIG. 1 shows the configuration of the vehicle control system 1.
  • the vehicle control system 1 includes an image analysis device 10 as an in-vehicle electronic device, a vehicle control device 20, a wheel speed sensor 30, and an acceleration sensor 40 that implement the present invention.
  • an image analysis device 10 as an in-vehicle electronic device
  • vehicle control device 20 a wheel speed sensor 30, and an acceleration sensor 40 that implement the present invention.
  • an acceleration sensor 40 that implement the present invention.
  • each of the image analysis device 10 the vehicle control device 20, the wheel speed sensor 30, and the acceleration sensor 40 is connected to an in-vehicle network and configured to be able to communicate with each other.
  • the wheel speed sensor 30 In addition to the wheel speed sensor 30 and the acceleration sensor 40, various sensors (not shown) capable of detecting the traveling / driving state of the vehicle are connected to the in-vehicle network so as to provide the detected values.
  • the wheel speed sensor 30 outputs a vehicle speed signal corresponding to the rotation of the wheel, and indirectly detects the speed of the vehicle by detecting the rotation speed of the wheel.
  • the acceleration sensor 40 is an inertial sensor that performs measurement using inertia, and detects and outputs the acceleration of the vehicle based on the displacement of the member due to inertia, as is well known.
  • the image analysis apparatus 10 includes a camera 11, a communication interface 15, and a control unit 17.
  • the camera 11 shoots a field of view ahead of a vehicle (so-called own vehicle) on which the vehicle control system 1 is mounted, thereby generating captured image data as image data representing the captured image.
  • a vehicle so-called own vehicle
  • a stereo camera can also be used.
  • the communication interface 15 is controlled by the control unit 17, and is configured to be capable of bidirectional communication with communication nodes such as the vehicle control device 20, the wheel speed sensor 30, and the acceleration sensor 40 through the in-vehicle network.
  • control unit 17 performs overall control of the image analysis apparatus 10, and includes a CPU 17A, a ROM 17B, and a RAM 17C.
  • various functions as the image analysis device 10 are realized by the CPU 17 ⁇ / b> A executing various processes in accordance with programs stored in the ROM 17 ⁇ / b> B.
  • the RAM 17C is used as a working memory when the CPU 17A executes a program.
  • the control unit 17 executes a vanishing point learning process PR1, a learning control process PR2, a road lane marking estimation process PR3, a travel state estimation process PR4, and the like shown in FIG. 2 according to a program stored in the ROM 17B.
  • the vanishing point learning process PR1 is a process of learning the vanishing point (FOE) position in the captured image data according to a known technique.
  • the learned vanishing point position is stored in the ROM 17B as a parameter representing the camera posture.
  • the ROM 17B of this embodiment includes, for example, a flash memory that can electrically rewrite data.
  • the learning control process PR2 is a process for controlling the execution of the vanishing point learning process PR1.
  • the control unit 17 controls the start (on) / end (off) of the vanishing point learning process PR1 by executing the learning control process PR2.
  • the road lane marking estimation process PR3 is a process for estimating the area of the road lane marking shown in the captured image data.
  • edges as road lane marking candidates are extracted from the captured image data, and based on the positional relationship between the direction of these edges and the learned vanishing point, Edges that are likely to be road lane markings of the traveling road are determined. Thereby, the area of the road lane marking of the road on which the host vehicle travels is estimated.
  • the road marking line can be estimated using the vanishing point position calculated from the installation parameters of the camera 11 as an index.
  • the vanishing point position can be learned based on the road marking line estimated by the road marking line estimation process PR3. For example, the intersection point appearing on the extension line of two estimated road marking lines is detected as a vanishing point candidate, and the probability that the candidate is a vanishing point is determined by the error between the detected candidate position and the learned vanishing point position. If the error is large and the accuracy is low, this candidate is rejected. If the error is small and the accuracy is high, this candidate is adopted as the vanishing point, and the vanishing point position stored in the ROM 17B is learned and updated. To do.
  • the driving state estimation process PR4 is a process of analyzing the captured image data using the learned vanishing point position as an index to estimate the driving state of the host vehicle with respect to the road and the positional relationship with the preceding vehicle. Since this driving state estimation process PR4 is a well-known process, it will be briefly described.
  • the driving state estimation process PR4 based on road marking lines (white lines, botsdots, etc.) estimated from captured image data, A process for estimating the direction and position of the host vehicle can be given as an example.
  • the traveling state estimation process PR4 the forward vehicle reflected in the captured image data is searched and detected based on the vanishing point position, or the positional relationship between the detected forward vehicle and the own vehicle (the forward vehicle relative to the own vehicle). The process of estimating the distance etc.) can be given as an example.
  • Information regarding the traveling state of the host vehicle with respect to the road estimated by the traveling state estimation process PR4 and the positional relationship with the preceding vehicle is provided to the vehicle control device 20 through the communication interface 15 and the in-vehicle network, and is used for vehicle control.
  • vehicle control is used in a broad sense to control devices in the vehicle.
  • the vehicle control device 20 as vehicle control based on information obtained from the image analysis device 10, for example, when the host vehicle is traveling so as to cross a road marking line, or when the host vehicle is approaching a preceding vehicle
  • the learning value of the vanishing point position is used when estimating the road lane marking or when estimating the running state, it is not preferable that erroneous learning occurs.
  • the vanishing point position is learned in a situation where the vehicle 100 is simulated running on the chassis dynamometer 200 at the time of vehicle inspection, dirt on the wall 210 in front of the host vehicle captured by the camera 11 and the wall 210
  • mis-learning of the vanishing point position may occur due to the shadows of surrounding buildings reflected in.
  • the learning value of the vanishing point position is used, but it may be difficult to determine the correct edge as the road lane marking based on the vanishing point position obtained by mislearning. There is sex. Further, even though the vanishing point learning process PR1 can detect a correct vanishing point as a vanishing point candidate, the position of this candidate is different from the mis-learned vanishing point position. May not be used.
  • the process shown in FIG. 3 is executed as the learning control process PR2, so that the vanishing point learning process is performed in a situation where there is a high possibility that the vehicle 100 is simulated running on the chassis dynamometer 200. PR1 is not started and the learning operation is kept off.
  • the control unit 17 starts the learning control process PR2 shown in FIG. 3 when the ignition switch is turned on, and repeatedly executes this process at regular intervals until the ignition switch is turned off.
  • the control unit 17 determines whether or not the vehicle speed specified from the output of the wheel speed sensor 30 obtained through the in-vehicle network and the communication interface 15 is greater than zero (step S100). If it is determined that the vehicle speed is less than or equal to zero, the process proceeds to step S105, and if it is determined that the vehicle speed is greater than zero, the process proceeds to step S110.
  • step S105 the control unit 17 resets the determined flag F and the learning permission flag G to zero, and resets the acceleration integral value to zero, and then ends the learning control process PR2.
  • the determined flag F referred to here is a flag indicating whether or not a determination is made as to whether or not the vehicle is traveling on a road (in other words, whether or not the vehicle is simulating traveling). A determination is made that a determination has not been made, and a value of 1 indicates that a determination has been made.
  • the learning permission flag is a flag indicating whether or not the execution of the vanishing point learning process PR1 is permitted. A value of 0 indicates that it is not permitted (prohibited), and a value of 1 indicates permission. Indicates that The acceleration integral value is calculated by an acceleration integration process (step S120) described later.
  • step S110 the control unit 17 determines whether or not the determined flag F is set to the value 1, and determines that the determined flag F is set to the value 1 (Yes in step S110). ), The process proceeds to step S135, and if it is determined that the determined flag F is not set to 1 (No in step S110), the process proceeds to step S120. Since the determined flag F is reset to zero when the ignition switch is turned on, a negative determination is made in the first step S110 (No in step S110), and the process proceeds to step S120.
  • step S120 the control unit 17 finally resets the acceleration integral value to zero in step S105 based on the vehicle acceleration specified from the output of the acceleration sensor 40 obtained through the in-vehicle network and the communication interface 15.
  • a process of calculating an integral value of acceleration from the time is executed. Thereby, the actual speed of the vehicle after the vehicle starts running is estimated using the output of the acceleration sensor 40.
  • step S120 for example, the speed change obtained by multiplying the acceleration of the vehicle specified this time by the execution cycle of step S120 is added to the acceleration integral value calculated in the previous step S120. Thereby, an integrated value of acceleration (vehicle speed) after the vehicle speed becomes a value larger than zero is calculated.
  • step S130 compares the vehicle speed specified from the output of the wheel speed sensor 30 with a reference speed determined in advance in the design stage, and It is determined whether or not the vehicle speed exceeds the reference speed.
  • the reference speed can be determined by the designer in terms of whether or not the vanishing point position can be appropriately learned. Since the learning of the vanishing point position can be appropriately performed on a road with a good line of sight, for example, about 50 km / h can be set as the reference speed.
  • step S130 when it is determined that the vehicle speed specified from the output of the wheel speed sensor 30 does not exceed the reference speed (No in step S130), the control unit 17 proceeds to step S100, and the vehicle speed becomes the reference speed. Until it exceeds, the processing of steps S100 to S130 is repeated. On the other hand, when it is determined that the vehicle speed specified from the output of the wheel speed sensor 30 exceeds the reference speed (Yes in step S130), the control unit 17 is specified from the output of the wheel speed sensor 30 at that time. An error (absolute value) between the vehicle speed and the integrated acceleration value is calculated (step S140).
  • step S150 it is determined whether or not the error is less than a threshold value determined in advance in the design stage (step S150). If it is determined that the error is less than the threshold value (Yes in step S150), the determined flag F is set to a value of 1. In addition to setting the learning permission flag G to 1 (step S153), the process proceeds to step S160. On the other hand, if it is determined that the error is equal to or greater than the threshold (No in step S150), the control unit 17 sets the determined flag F to the value 1 while maintaining the state where the learning permission flag G is reset to the value 0. (Step S157), the process proceeds to Step S160.
  • step S150 As a threshold value used in step S150, when the vehicle 100 is in a simulated running state on the chassis dynamometer 200, a negative determination is made with a high probability in step S150, and the vehicle is not in a simulated traveling state but on the road. When the vehicle is traveling, a value that can be affirmatively determined with high probability can be obtained and determined by experiments or the like.
  • step S150 it is determined whether or not the vehicle 100 is traveling on the road by determining whether or not the error is less than the threshold. Only when the error is less than the threshold, the learning permission flag G By setting the value to 1, the execution of the vanishing point learning process PR1 is permitted, and when the error is equal to or greater than the threshold value, it is considered that the vehicle 100 is highly likely to be simulated running on the chassis dynamometer 200. Execution of the vanishing point learning process PR1 is prohibited.
  • step S160 it is determined whether or not the learning permission flag G is set to the value 1. If it is determined that the learning permission flag G is set to the value 1 (Yes in step S160), the vanishing point learning process is performed. After PR1 is started (step S170), the process proceeds to step S180, and if it is determined that the learning permission flag G is not set to 1 (No in step S160), the vanishing point learning process PR1 is not started, The learning control process is temporarily terminated.
  • the control unit 17 determines whether or not the end condition of the vanishing point learning process PR1 is satisfied.
  • the vehicle speed specified from the output of the wheel speed sensor 30 is equal to or lower than a predetermined learning end speed (for example, 50 km / h) within a speed range equal to or lower than the reference speed. Is less than or equal to the learning end speed, it can be determined that the end condition is satisfied, and if the vehicle speed is greater than the learning end speed, it can be determined that the end condition is not satisfied.
  • a predetermined learning end speed for example, 50 km / h
  • the end condition can be arbitrarily determined by the designer of the image analysis apparatus 10.
  • step S180 If it is determined that the end condition is not satisfied (No in step S180), the control unit 17 repeatedly executes the determination in step S180 until the end condition is satisfied, and if it is determined that the end condition is satisfied (step S180). After the vanishing point learning process PR1 is ended (step S190), the learning control process PR2 is once ended.
  • control unit 17 temporarily reduces the vehicle speed specified from the output of the wheel speed sensor 30 to zero. Under the assumption that there is no change in the situation of whether the vehicle is traveling on the road or simulated driving, the learning operation is turned on / off using the determination result in the previous step S150. Control.
  • step S105 the determined flag F set to the value 1 in steps S153 and 157 is set to the value 1 until the vehicle speed specified from the output of the wheel speed sensor 30 once falls to zero and the process of step S105 is executed. Therefore, in the learning control process until the vehicle speed once drops to zero, an affirmative determination is made in step S110, and the process proceeds to step S135.
  • step S135 similarly to the processing in step S130, it is determined whether or not the vehicle speed specified from the output of the wheel speed sensor 30 exceeds the reference speed. If it is determined that the vehicle speed does not exceed the reference speed (No in S135), the process proceeds to S100. If it is determined that the vehicle speed exceeds the reference speed (Yes in S135), the process proceeds to Step S160. To do. In the processes after step S160 (steps S160 to S190), only when the affirmative determination is made in step S150 in the past learning control process and the learning permission flag G is set to 1 (Yes in step S160). Then, the vanishing point learning process PR1 is started (step S170).
  • the vehicle control system 1 has been described above.
  • the camera 11 captures the front area of the host vehicle, and the captured image data generated by the camera 11 is analyzed by the control unit 17. While the vanishing point position is learned, on / off of the vanishing point position learning operation is controlled based on the output of the acceleration sensor 40 that is an inertial sensor.
  • step S150 it is determined whether or not the vehicle is traveling on the road based on the output of the acceleration sensor 40 that measures acceleration using inertia. Only when it is determined that the vehicle is traveling above, the learning operation is turned on (step S170).
  • the vanishing point position learning operation is controlled as in the present embodiment, the vanishing point position is learned while the vehicle 100 is simulating on the chassis dynamometer 200. Can be suppressed. Therefore, according to the present embodiment, it is possible to suppress the occurrence of mislearning of the vanishing point position during the simulated running of the vehicle, and the mislearning has an unfavorable effect on vehicle control and later learning of the vanishing point position. Can be suppressed.
  • the vanishing point position is learned and updated based on the road lane line information reflected in the captured image data estimated by the road lane line estimation process PR3.
  • Use vanishing point information Therefore, if the learned vanishing point position is greatly deviated from the correct position due to mislearning of the vanishing point position, the road lane marking itself cannot be accurately estimated, and the vanishing point position is learned and updated to the correct value. Time may be required, and it may be difficult to learn and update the vanishing point position to the correct value.
  • the vehicle control system 1 capable of realizing appropriate vehicle control based on vanishing point information is constructed. Can do.
  • the error is based on the error between the vehicle speed specified from the output of the wheel speed sensor 30 and the vehicle speed calculated by integrating the acceleration specified from the output of the acceleration sensor 40. Only when it is less than the reference (Yes in step S150), the learning operation is switched from off to on, so whether or not the vehicle is traveling on the road based only on the output of the acceleration sensor 40 is determined. It is possible to realize determination and on / off control with higher accuracy than the determination and on / off control of the learning operation, and to further suppress the mislearning of the vanishing point position.
  • the learning operation of the vanishing point position is likely to occur on a road with poor visibility such as a narrow street, according to this embodiment, there is a high possibility that the vehicle is traveling on a road with poor visibility.
  • the learning operation is kept off during low-speed driving.
  • the learning operation is kept off, and when the vehicle speed exceeds the reference speed (Yes in step S130).
  • the learning operation is turned off on condition that the vehicle speed specified from the output of the wheel speed sensor 30 exceeds the reference speed and that the error is less than the reference. Switched on. Therefore, according to the present embodiment, mislearning of the vanishing point position can be further suppressed.
  • the control unit 17 determines whether or not the vehicle speed specified from the output of the wheel speed sensor 30 is larger than zero, similarly to the first embodiment (step S200). If it is determined that the vehicle speed is less than or equal to zero, the process proceeds to step S205, and if it is determined that the vehicle speed is greater than zero, the process proceeds to step S210.
  • step S205 the control unit 17 resets the determined flag F and the learning permission flag G to the value zero and resets the error statistical value to the value zero, and then performs the learning control similarly to the process in step S105.
  • the process PR2 is once ended.
  • the error statistic value is calculated by an error statistic value calculation process (step S220) described later.
  • step S210 the control unit 17 determines whether or not the determined flag F is set to the value 1, and determines that the determined flag F is set to the value 1 (in step S210). Yes), the process proceeds to step S235, and if it is determined that the determined flag F is not set to 1 (No in step S210), the process proceeds to step S220.
  • the control unit 17 executes an error statistic value calculation process shown in FIG.
  • the vehicle acceleration corresponding to the rotational acceleration of the wheel is calculated by calculating the differential value of the vehicle speed specified from the output of the wheel speed sensor 30 (step S221).
  • the differential value can be obtained, for example, by dividing the deviation between the vehicle speed at the previous execution time of step S221 and the vehicle speed at the current execution time of step S221 by the execution period of step S221.
  • step S221 the control unit 17 specifies the acceleration of the vehicle from the output of the acceleration sensor 40, and calculates an error (absolute value) between this acceleration and the speed differential value calculated in step S221. Calculate (step S223). Further, using the error obtained in step S223 each time from the time when the error statistical value was finally reset to zero in step S205 as a sample group, a statistical value of error in this sample group is calculated (step S225). Specifically, in step S225, for example, calculating an average value of errors calculated so far as the error statistic value is given. In addition, as an error statistic value, a median value of errors in the sample population may be calculated, or a maximum value of errors in the sample population may be calculated. In step S205, the error statistic value is reset to zero, and the sample group used so far can be deleted.
  • step S230 determines whether or not the vehicle speed specified from the output of the wheel speed sensor 30 exceeds the reference speed, similarly to the process at step S130. If it is determined that the vehicle speed specified from the output of the wheel speed sensor 30 does not exceed the reference speed (No in step S230), the control unit 17 proceeds to step S200, and this vehicle speed is the reference speed. Steps S200 to S230 are repeatedly executed until the value exceeds. On the other hand, if it is determined that the vehicle speed exceeds the reference speed (Yes in Step S230), the process proceeds to Step S250, and the latest error statistic value calculated in Step S220 is less than the threshold value previously determined in the design stage. Judge whether there is.
  • step S250 If it is determined that the error statistical value is less than the threshold value (Yes in step S250), the determined flag F is set to 1 and the learning permission flag G is set to 1 (step S253), and then step S260. Migrate to On the other hand, if it is determined that the error statistical value is equal to or greater than the threshold value (No in step S250), the determined flag F is set to the value 1 while the learning permission flag G is reset to the value 0 (step S257), the process proceeds to step S260.
  • the threshold used in step S250 can be determined by the designer according to the same criteria as the threshold used in step S150 of the first embodiment.
  • step S260 determines whether or not the learning permission flag G is set to a value 1 (step S260), and determines that the learning permission flag G is set to a value 1 (Yes in step S260).
  • step S270 determines that the learning permission flag G is set to a value 1 (Yes in step S260).
  • step S280 the process proceeds to step S280, and if a negative determination is made (No in step S260), the learning control process is temporarily ended without starting the vanishing point learning process PR1.
  • step S280 the control unit 17 determines whether or not the end condition of the vanishing point learning process PR1 is satisfied as in the process in step S180, and determines that the end condition is not satisfied (step S280). No in S280), the determination in step S280 is repeatedly executed until the end condition is satisfied. If it is determined that the end condition is satisfied (Yes in step S280), after the vanishing point learning process PR1 is ended (step S290), The learning control process PR2 is temporarily terminated.
  • control unit 17 specifies the output from the wheel speed sensor 30 as in the first embodiment. Until the vehicle speed once drops to zero, the learning operation is controlled to be turned on / off using the determination result in the past step S250.
  • step S210 the process proceeds to step S235, and is specified from the output of the wheel speed sensor 30 as in the process in step S230. It is determined whether or not the vehicle speed exceeds the reference speed. If it is determined that the vehicle speed does not exceed the reference speed (No in Step S235), the process proceeds to Step S200, and if it is determined that the vehicle speed exceeds the reference speed (Yes in Step S235), Step The process proceeds to S260.
  • steps S260 to S290 as in the first embodiment, when a positive determination is made in step S250 in the past learning control processing and the learning permission flag G is set to a value of 1. Only (Yes in Step S260), the vanishing point learning process PR1 is started (Step S270).
  • the error between the vehicle acceleration calculated from the deviation of the vehicle speed specified from the output of the wheel speed sensor 30 and the vehicle acceleration specified from the output of the acceleration sensor 40 is performed based on the above, but the same effect as in the first embodiment can be obtained by such control as well.
  • the content of the learning control process PR2 executed by the control unit 17 is different from that of the first embodiment. Therefore, hereinafter, the learning control in the third embodiment is performed.
  • the contents of the process PR2 will be described selectively.
  • the control unit 17 in this embodiment starts the learning control process PR2 shown in FIG. 6 when the ignition switch is turned on, and repeatedly executes this process at regular intervals until the ignition switch is turned off.
  • the control unit 17 determines whether or not the vehicle speed specified from the output of the wheel speed sensor 30 is greater than zero, similarly to the process in step S100 (step S300). If it is determined that the speed is less than or equal to zero, the process proceeds to step S305, and if it is determined that the vehicle speed is greater than zero, the process proceeds to step S310.
  • step S305 the control unit 17 resets the determined flag F and the learning permission flag G to the value zero and resets the acceleration statistical value to the value zero, and then performs the learning control similarly to the process in step S105.
  • the process PR2 is once ended.
  • the acceleration statistical value is calculated in step S320 described later.
  • step S310 the control unit 17 determines whether or not the determined flag F is set to the value 1, and determines that the determined flag F is set to the value 1 (in step S310). If YES, the process proceeds to step S335, and if it is determined that the determined flag F is not set to 1 (No in step S310), the process proceeds to step S320.
  • step S320 the control unit 17 identifies the current vehicle acceleration from the output of the acceleration sensor 40, and was observed by the acceleration sensor 40 from the time when the acceleration statistical value was finally reset to zero in step S305.
  • Calculate acceleration statistics Specifically, in step S320, as an acceleration statistic value, for example, calculating an average value of acceleration from the time when the acceleration statistic value was finally reset to zero in step S305 is given as an example.
  • the acceleration statistical value the median value of acceleration may be calculated, or the maximum value of acceleration may be calculated.
  • step S320 the control unit 17 proceeds to step S330, and in the same way as the process in step S130, compares the vehicle speed specified from the output of the wheel speed sensor 30 with the reference speed, and determines the vehicle speed. Is determined not to exceed the reference speed (No in step S330), the process proceeds to step S300, and the processes of steps S300 to S330 are repeatedly executed until the vehicle speed exceeds the reference speed.
  • Step S330 If it is determined that the vehicle speed specified from the output of the wheel speed sensor 30 exceeds the reference speed (Yes in Step S330), the process proceeds to Step S350, and the latest acceleration statistical value calculated in Step S320 is stored in advance. It is determined whether or not a threshold value determined in the design stage is exceeded.
  • step S350 if it is determined that the acceleration statistical value exceeds the threshold (Yes in step S350), the determined flag F is set to the value 1 and the learning permission flag G is set to the value 1 (step S353), and then the step. The process proceeds to S360.
  • the threshold value can be determined by the designer according to the same criteria as the threshold value used in step S150 of the first embodiment.
  • step S360 determines whether or not the learning permission flag G is set to the value 1 (step S360), and determines that the learning permission flag G is set to the value 1 (Yes in step S360).
  • step S370 the process proceeds to step S380, and if a negative determination is made (No in step S360), the learning control process is temporarily ended without starting the vanishing point learning process PR1.
  • step S380 the control unit 17 determines whether or not the end condition of the vanishing point learning process PR1 is satisfied as in the process in step S180, and determines that the end condition is not satisfied (step S380). If the determination in step S380 is repeated until the end condition is satisfied and it is determined that the end condition is satisfied (Yes in step S380), the vanishing point learning process PR1 is ended (step S390). The learning control process PR2 is temporarily terminated.
  • control unit 17 specifies the output from the wheel speed sensor 30 as in the first embodiment. Until the vehicle speed once falls to zero, the learning operation is controlled to be turned on / off using the determination result in the past step S350.
  • step S310 an affirmative determination is made in step S310 and the process proceeds to step S335, and is specified from the output of the wheel speed sensor 30 as in the process in step S330. It is determined whether or not the vehicle speed exceeds the reference speed. If it is determined that the vehicle speed does not exceed the reference speed (No in Step S335), the process proceeds to Step S300, and if it is determined that the vehicle speed exceeds the reference speed (Yes in Step S335), Step The process proceeds to S360.
  • step S360 steps S360 to S390
  • the learning permission flag G is set to 1
  • the vanishing point learning process PR1 is started.
  • the speed of the vehicle specified from the output of the wheel speed sensor 30 exceeds the reference speed (Yes in step S330), and the acceleration sensor 40
  • the learning operation is switched from OFF to ON on the condition that the acceleration (acceleration statistical value) of the vehicle specified from the output exceeds the threshold (Yes in Step S350). Also by this control method, the vanishing point position learning operation can be appropriately executed.
  • the control unit 17 first determines whether or not the vehicle is in a driving state (step S400).
  • whether or not the vehicle is in operation is determined by whether or not the vehicle speed specified from the output of the wheel speed sensor 30 is greater than zero. be able to.
  • step S400 If it is determined that the vehicle is not being driven (No in step S400), the process proceeds to step S405. If it is determined that the vehicle is being driven (Yes in step S400), the process proceeds to step S410. To do.
  • the control unit 17 resets the learning permission flag G to zero and resets the acceleration integral value to zero, and then ends the learning control process PR2.
  • the acceleration integral value is a process attached to the learning control process PR2, and is calculated by an acceleration integration process executed in parallel with the learning control process PR2.
  • step S120 The contents of this acceleration integration processing are the same as in step S120 in the first embodiment, but in this embodiment, in addition to the processing loop of learning control processing PR2, the control unit 17 turns on the ignition switch. Until the ignition switch is turned off, the acceleration integration process is always repeatedly executed at a predetermined execution cycle, and the integrated value of the acceleration observed by the acceleration sensor 40 after the vehicle starts to be operated is calculated. This always estimates the actual speed of the vehicle. In step S405, the acceleration integrated value is reset to zero.
  • step S410 the control unit 17 determines whether or not the learning permission flag G is set to the value 1, and determines that the learning permission flag G is set to the value 1 (in step S410). Yes), the process proceeds to step S435, and if it is determined that the learning permission flag G is not set to 1 (No in step S410), the process proceeds to step S430.
  • step S430 it is determined whether or not the latest acceleration integration value calculated by the acceleration integration process exceeds a reference value determined in advance in the design stage.
  • the “reference value” used here can be set to the same value as the reference speed used in step S130.
  • step S430 If it is determined that the acceleration integral value is equal to or less than the reference value (No in step S430), the control unit 17 proceeds to step S400 and continues until the acceleration integral value exceeds the reference value (Yes in step S430).
  • the process proceeds to Step S450, and the learning permission flag G is set to the value 1, and then disappears.
  • the point learning process PR1 is started (step S470), and the process proceeds to step S480.
  • the control unit 17 determines whether or not the end condition of the vanishing point learning process PR1 is satisfied, similarly to the process in step S180.
  • step S480 If it is determined that the end condition is not satisfied (No in step S480), the determination in step S480 is repeatedly executed until the end condition is satisfied, and if it is determined that the end condition is satisfied (Yes in step S480), After ending the vanishing point learning process PR1 (step S490), the learning control process PR2 is once ended.
  • control unit 17 executes the processes after step S450 and once ends the learning control process, until it is determined that the vehicle is not in a driving state in the second learning control process (step S400). No), since the learning permission flag G is maintained at the value 1, an affirmative determination is made in step S410, and the process proceeds to step S435. In step S435, it is determined whether the resumption condition for the learning operation is satisfied as follows.
  • step S435 for example, similarly to the processing in step S430, when the acceleration integral value exceeds the reference value, it is determined that the restart condition is satisfied, and when the acceleration integral value is equal to or less than the reference value. It is determined that the restart condition is not satisfied.
  • a difference V1 ⁇ V2 between the latest acceleration integration value V1 calculated by the acceleration integration process and the acceleration integration value V2 at the time when the vanishing point learning process is finally ended is If the difference V1-V2 is greater than or equal to a threshold value determined in advance in the design stage, the restart condition is determined to be satisfied, and if the difference V1-V2 is less than the threshold value, the restart condition is not satisfied. It can be judged.
  • the threshold value can be set to such a value that it is determined that the restart condition is satisfied in an environment where the vehicle speed exceeds the reference speed.
  • the restart condition is satisfied when the vehicle speed specified from the output of the wheel speed sensor 30 exceeds the reference speed, as in the processes in steps S135, S235, and S335. If the vehicle speed does not exceed the reference speed, it can be determined that the restart condition is not satisfied.
  • control unit 17 determines whether or not the restart condition is satisfied. If the control unit 17 determines that the restart condition is not satisfied (No in step S435), the control unit 17 proceeds to step S400 while restarting. If it is determined that the condition is satisfied (Yes in step S435), the process proceeds to step S470, and the vanishing point learning process PR1 is started.
  • the vehicle speed calculated by integrating the acceleration of the vehicle specified from the output of the acceleration sensor 40 exceeds the reference speed ( In step S430, Yes), the learning operation is switched from OFF to ON. Also by such a control method, the learning operation of the vanishing point position can be appropriately executed while suppressing erroneous learning.
  • the present invention is not limited to the modes described in the above embodiments, and can take various modes.
  • the determination in steps S100, S200, and S300 can be replaced with the determination as to whether or not the vehicle is being driven, similar to the determination in step S400.
  • the output of the inertial sensor is used, on / off of the learning operation may be controlled by a method other than the above embodiment.
  • the acceleration integration process (step S120) is executed in parallel with the learning control process PR2, and thus based on the output of the acceleration sensor 40.
  • An example in which the actual speed of the vehicle is always estimated after the vehicle starts to be driven can be considered.
  • an embodiment in which the on / off control of the learning operation is executed without using the determination result in the past step S150 is also conceivable. That is, an embodiment in which the determination step in step S110 is deleted and the processing in steps S140 and S150 to S157 is performed again based on the acceleration integrated value at that time in the learning control process PR2 again is also conceivable.
  • the image analysis device 10 corresponds to an example of an electronic device mounted on a vehicle.
  • the vanishing point learning process PR1 executed by the control unit 17 corresponds to an example of the process realized by the learning unit
  • the learning control process PR2 executed by the control unit 17 is an example of the process realized by the control unit.
  • Vehicle control system 10: Image analysis device, 11 ... Camera, 15 ... Communication interface, 17 ... control unit, 17A ... CPU, 17B ... ROM, 17C ... RAM, 20 ... Vehicle control device, 30: Wheel speed sensor, 40 ... acceleration sensor, 100 ... vehicle, 200: Chassis dynamometer, 210 ... wall

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Image Analysis (AREA)
  • Fittings On The Vehicle Exterior For Carrying Loads, And Devices For Holding Or Mounting Articles (AREA)
  • Length Measuring Devices With Unspecified Measuring Means (AREA)
PCT/JP2013/067818 2012-06-29 2013-06-28 車両に搭載される画像解析装置 Ceased WO2014003168A1 (ja)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US14/411,113 US20150294453A1 (en) 2012-06-29 2013-06-28 Image analysis apparatus mounted to vehicle

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2012147005A JP6035904B2 (ja) 2012-06-29 2012-06-29 電子機器
JP2012-147005 2012-06-29

Publications (1)

Publication Number Publication Date
WO2014003168A1 true WO2014003168A1 (ja) 2014-01-03

Family

ID=49783302

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2013/067818 Ceased WO2014003168A1 (ja) 2012-06-29 2013-06-28 車両に搭載される画像解析装置

Country Status (3)

Country Link
US (1) US20150294453A1 (enExample)
JP (1) JP6035904B2 (enExample)
WO (1) WO2014003168A1 (enExample)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2015131887A1 (de) * 2014-03-05 2015-09-11 Conti Temic Microelectronic Gmbh Vorrichtung zur korrektur eines abstandswertes und/oder zur korrektur eines relativgeschwindigkeitswertes, fahrzeug und verfahren
WO2016047498A1 (ja) * 2014-09-24 2016-03-31 株式会社デンソー 物体検出装置
CN107787496A (zh) * 2015-05-26 2018-03-09 Plk科技株式会社 消失点修正装置及方法
CN110132280A (zh) * 2019-05-20 2019-08-16 广州小鹏汽车科技有限公司 室内场景下的车辆定位方法、车辆定位装置和车辆
WO2025147545A1 (en) 2024-01-03 2025-07-10 Juno Therapeutics, Inc. Lipid nanoparticles for delivery of nucleic acids and related methods and uses

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106461774B (zh) * 2014-02-20 2019-04-23 御眼视觉技术有限公司 基于雷达提示视觉成像的高级驾驶员辅助系统
JP6406886B2 (ja) * 2014-06-11 2018-10-17 キヤノン株式会社 画像処理装置、画像処理方法、コンピュータプログラム
JP6618603B2 (ja) * 2018-12-17 2019-12-11 パイオニア株式会社 撮影装置、制御方法、プログラム及び記憶媒体
DE102019111642B3 (de) * 2019-05-06 2020-06-04 Sick Ag Absichern der Umgebung eines Fahrzeugs
US11979655B2 (en) * 2021-09-30 2024-05-07 Gm Global Tehcnology Operations Llc Method to detect and overcome degradation image quality impacts

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2002259995A (ja) * 2001-03-06 2002-09-13 Nissan Motor Co Ltd 位置検出装置
JP2011192227A (ja) * 2010-03-17 2011-09-29 Clarion Co Ltd 車両姿勢角算出装置及びそれを用いた車線逸脱警報システム

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2008091565A1 (en) * 2007-01-23 2008-07-31 Valeo Schalter & Sensoren Gmbh Method and system for universal lane boundary detection
US8254635B2 (en) * 2007-12-06 2012-08-28 Gideon Stein Bundling of driver assistance systems

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2002259995A (ja) * 2001-03-06 2002-09-13 Nissan Motor Co Ltd 位置検出装置
JP2011192227A (ja) * 2010-03-17 2011-09-29 Clarion Co Ltd 車両姿勢角算出装置及びそれを用いた車線逸脱警報システム

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2015131887A1 (de) * 2014-03-05 2015-09-11 Conti Temic Microelectronic Gmbh Vorrichtung zur korrektur eines abstandswertes und/oder zur korrektur eines relativgeschwindigkeitswertes, fahrzeug und verfahren
JP2017516068A (ja) * 2014-03-05 2017-06-15 コンティ テミック マイクロエレクトロニック ゲゼルシャフト ミット ベシュレンクテル ハフツングＣｏｎｔｉ Ｔｅｍｉｃ ｍｉｃｒｏｅｌｅｃｔｒｏｎｉｃ ＧｍｂＨ 間隔値を補正するための、及び／或いは、相対速度を補正するための装置、車両、並びに、方法
US10060945B2 (en) 2014-03-05 2018-08-28 Conti Temic Microelectronic Gmbh Device for correcting a spacing value and/or for correcting a relative speed value, vehicle, and method
WO2016047498A1 (ja) * 2014-09-24 2016-03-31 株式会社デンソー 物体検出装置
JP2016065760A (ja) * 2014-09-24 2016-04-28 株式会社デンソー 物体検出装置
CN107787496A (zh) * 2015-05-26 2018-03-09 Plk科技株式会社 消失点修正装置及方法
CN107787496B (zh) * 2015-05-26 2021-07-13 Plk科技株式会社 消失点修正装置及方法
CN110132280A (zh) * 2019-05-20 2019-08-16 广州小鹏汽车科技有限公司 室内场景下的车辆定位方法、车辆定位装置和车辆
WO2025147545A1 (en) 2024-01-03 2025-07-10 Juno Therapeutics, Inc. Lipid nanoparticles for delivery of nucleic acids and related methods and uses

Also Published As

Publication number Publication date
JP2014010637A (ja) 2014-01-20
JP6035904B2 (ja) 2016-11-30
US20150294453A1 (en) 2015-10-15

Similar Documents

Publication Publication Date Title
JP6035904B2 (ja) 電子機器
WO2014003167A1 (ja) 車両に搭載される画像解析装置
US10074281B2 (en) Method and apparatus for determining lane identification in a roadway
US9283967B2 (en) Accurate curvature estimation algorithm for path planning of autonomous driving vehicle
JP6298772B2 (ja) 車載用制御装置、自車位置姿勢特定装置、車載用表示装置
US9740942B2 (en) Moving object location/attitude angle estimation device and moving object location/attitude angle estimation method
JP6538547B2 (ja) 道路曲率計測装置
JP6369399B2 (ja) センサ出力補正装置
KR102086270B1 (ko) 주행 제어 장치의 제어 방법 및 주행 제어 장치
JP2017521745A (ja) 車両のナビゲーションモジュールに対象物の存在を知らせる車載装置
CN108883770B (zh) 行进路推定方法及行进路推定装置
JP6941178B2 (ja) 自動運転制御装置及び方法
JP2017138660A (ja) 物体検出方法、物体検出装置、およびプログラム
JP5904226B2 (ja) 車両挙動予測装置及びプログラム
CN115143973B (zh) 地图信息判定装置、存储介质以及地图信息判定方法
CN113939826A (zh) 地图系统、地图生成程序、存储介质、车辆用装置以及服务器
JP7675359B2 (ja) 位置推定システム
JP7136050B2 (ja) 自車位置推定装置
JP7677051B2 (ja) 車両用制御装置及び車両用制御方法
JP6475140B2 (ja) 道路曲率計測装置
JP2020034498A (ja) スケールファクタ学習装置、スケールファクタ学習プログラムおよび記憶媒体
JP6075453B2 (ja) 運転状態推定装置
JP2015069288A (ja) 自車位置認識装置
CN114355333A (zh) 一种自动驾驶基于控制信息的目标筛选方法及系统、车辆
JP2006317287A (ja) 車両用現在位置決定装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 13809776

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 14411113

Country of ref document: US

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 13809776

Country of ref document: EP

Kind code of ref document: A1