KR101818535B1 - System for predicting the possibility or impossibility of vehicle parking using support vector machine - Google Patents

System for predicting the possibility or impossibility of vehicle parking using support vector machine Download PDF

Info

Publication number
KR101818535B1
KR101818535B1 KR1020110082213A KR20110082213A KR101818535B1 KR 101818535 B1 KR101818535 B1 KR 101818535B1 KR 1020110082213 A KR1020110082213 A KR 1020110082213A KR 20110082213 A KR20110082213 A KR 20110082213A KR 101818535 B1 KR101818535 B1 KR 101818535B1
Authority
KR
South Korea
Prior art keywords
parking
vehicle
svm
unit
parking space
Prior art date
Application number
KR1020110082213A
Other languages
Korean (ko)
Other versions
KR20130019908A (en
Inventor
김소연
Original Assignee
현대모비스 주식회사
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 현대모비스 주식회사 filed Critical 현대모비스 주식회사
Priority to KR1020110082213A priority Critical patent/KR101818535B1/en
Publication of KR20130019908A publication Critical patent/KR20130019908A/en
Application granted granted Critical
Publication of KR101818535B1 publication Critical patent/KR101818535B1/en

Links

Images

Abstract

A system and method for predicting automatic parking availability are disclosed. This automatic parking availability prediction system and its method are designed to learn existing vehicle position data and parking space data using a machine learning technique (Support Vector Machine) and to automatically predict whether or not the vehicle can be parked based on the learning data do. Thereby enhancing the performance of the automatic parking process.

Description

Technical Field [0001] The present invention relates to a system for predicting automatic parking availability using a machine learning technique,

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to vehicle-related technology, and more particularly to a technique for parking assistance.

1 is a block diagram of a conventional automatic parking system.

As shown, the automatic parking system is largely composed of application software and base software. The application software includes a system management unit 10 and an HMI (Human Machine Interface) 20. The system management unit 10 performs a function of integrally managing each block constituting the automatic parking system, that is, the vehicle position tracking unit 11, the parking space determination unit 12, and the parking path generation / tracking unit 13 . The vehicle position tracking unit 11 receives the vehicle information, determines the state and position of the vehicle, and transmits the state and position to the parking path generation / tracking unit 13. The parking space determination unit 12 determines the parking space by using the information received from the ultrasonic sensor, calculates the parking target point of the vehicle, and transmits the calculated parking target point to the parking path generation / tracking unit 13. The parking path generation / follow-up unit 13 calculates the parking induction locus by receiving the vehicle state and position information transmitted from the vehicle position tracking unit 11 and the parking target point information transmitted from the parking space determination unit 12, The parking space follow-up is induced by using the calculated parking induction locus and the obstacle information around the vehicle transmitted from the space determining unit 12. [ As shown in FIG. 2, such a method can not avoid errors in forming a parking path.

According to the conventional automatic parking system, since the vehicle can not follow the target trajectory, an accumulated error occurs at the time of parking completion. Also, since the driver controls the vehicle speed, it is impossible to directly control the vehicle speed in the automatic parking system. Also, when the initial position of the vehicle is not aligned with the parking line, an alignment error occurs at the time of parking completion. Then, the stop position by the driver becomes inconsistent with the initially calculated target reference point, and the parking alignment state becomes poor due to the change of the vehicle speed (rapid acceleration / rapid braking). In addition, since the conventional automatic parking system controls the parking using the determined vehicle position and the spatial position value, it is difficult to determine whether or not the parking system can be parked in the conventional system because various variables occurring in the actual parking environment can not be considered. And the number of mappings of the vehicle position and the parking space are numerous as shown in the example of Fig.

It is an object of the present invention to provide a technical solution for determining whether or not automatic parking can be performed in advance in order to reduce an error on a parking locus.

According to an aspect of the present invention, there is provided an automatic parking availability prediction system, comprising: a vehicle position recognition unit that recognizes a vehicle position by acquiring vehicle information including a steering angle and a wheel pulse; A parking space determination unit for receiving an ultrasonic sensing signal from the ultrasonic sensor to determine a parking space, and a control unit for receiving vehicle position information from the vehicle position recognition unit and receiving a parking space calculation result from the parking space determination unit, And a parking availability prediction control unit for predicting whether or not the vehicle can be parked based on the position information and the parking space calculation result. Here, the parking availability prediction control unit predicts whether or not the vehicle can be parked through a machine learning technique (Support Vector Machine) with the received vehicle position information and the parking space calculation result.

Further, the parking availability prediction control unit generates the machine learning technique module using the vehicle position information and the parking space calculation result as learning data of the machine learning technique, Predicts whether or not the vehicle can be parked based on the machine learning technique module with the actual parking data composed of the result of the spatial calculation.

According to another aspect of the present invention, there is provided a method for predicting automatic parking availability, the method comprising: recognizing a vehicle position by obtaining vehicle information including a steering angle and a wheel pulse; And estimating the availability of the vehicle using the machine learning technique with the recognized vehicle location information and the determined parking space calculation result.

The present invention creates the effect of improving the performance of the automatic parking system by predicting whether or not the parking can be automatically performed by learning the existing vehicle position data and parking space data using the machine learning technique. Further, the present invention minimizes the system change by additionally applying the automatic parking availability prediction system to the existing automatic parking system. Further, the present invention reduces the complexity by extracting feature vectors of vehicle data and classifying them using a machine learning technique.

1 is a block diagram of a conventional automatic parking system;
Fig. 2 is a reference diagram for explaining an occurrence of an error in forming a parking locus according to Fig. 1; Fig.
FIG. 3 is a reference diagram for explaining the relationship between the vehicle position and the number of parking spaces in the case of FIG. 1; FIG.
4 is a block diagram of an automatic parking system according to an embodiment of the present invention;
5 is a block diagram of an automatic parking availability prediction system according to an embodiment of the present invention.
6 is a flowchart illustrating an automatic parking availability prediction operation according to an exemplary embodiment of the present invention.
7 is a block diagram of a parking availability prediction control unit according to an embodiment of the present invention.
8 is a flowchart illustrating an operation of the parking availability prediction control unit according to an embodiment of the present invention.

BRIEF DESCRIPTION OF THE DRAWINGS The foregoing and further aspects of the present invention will become more apparent from the following detailed description of preferred embodiments with reference to the accompanying drawings. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

4 is a block diagram of an automatic parking system according to an embodiment of the present invention.

The automatic parking system includes an automatic parking availability prediction system 100, a vehicle path generation / tracking unit 200, and a vehicle control unit 300, as shown in FIG. . The automatic parking availability prediction system 100 according to an aspect of the present invention receives the sensed values from the vehicle position recognition unit 110 and the parking space determination unit 120 and outputs the sensed values to a support vector machine An alarm unit 140 for generating an alarm when the vehicle is judged to be abnormal parking by the parking availability prediction control unit 130, . The vehicle path generation / follow-up unit 200 is configured to perform a process of generating and following a vehicle path, and the path generation and tracking method itself is already known. The vehicle path generating / tracking unit 200 performs a process of generating and following a vehicle path when it is predicted that the vehicle can be parked by the automatic parking availability prediction system 100. The vehicle control unit 300 is a configuration for overall control of the parking process of the vehicle. The vehicle screen input / output unit 400 is an input / output interface for displaying video and text information for parking the vehicle on the screen.

5 is a block diagram of an automatic parking availability prediction system according to an embodiment of the present invention.

The automatic parking availability prediction system 100 includes a vehicle position recognition unit 110, a parking space determination unit 120, and a parking availability prediction control unit 130. Furthermore, the automatic parking availability prediction system 100 may further include an alarm unit 140. The vehicle position recognition unit 110 receives the vehicle information from the sensor unit 150 and recognizes the position of the vehicle based on the received vehicle information. In one embodiment, the vehicle position recognition unit 110 receives vehicle information via CAN (Controller Area Network) communication. The vehicle information includes at least some information of a steering angle, a wheel pulse, a yaw rate, and temperature information. The parking space determination unit 120 determines the parking space by using the sensing value of the ultrasonic sensor of the sensor unit 150 that measures the space in which the vehicle can park on the side of the vehicle, do. In one embodiment, the parking space determination unit 120 receives the sensing value of the ultrasonic sensor through LIN (Local Interconnect Network) communication for serial communication between the vehicle electronic systems. The automatic parking availability prediction control unit 130 receives the sensed value from the vehicle position recognition unit 110 and the parking space determination unit 120 and predicts whether or not the vehicle can be parked using a machine learning technique (SVM) Thereby determining whether or not the vehicle can be parked. The alarm unit 140 plays a role of generating an alarm when the parking availability prediction control unit 130 predicts that the parking is not possible.

6 is a flowchart illustrating an automatic parking availability prediction operation according to an exemplary embodiment of the present invention.

In S100, it is possible to measure the vehicle position using the sensed values of the steering angle, the wheel pulse, the yaw rate, and the temperature of the sensor, and the parking space of the vehicle can be measured using the sensing value of the ultrasonic signal sensor. The ultrasonic signal sensor part measures the space in which the vehicle can park on the side, front and rear, and judges whether or not the vehicle can park using the measured values.

In S200, the parking availability prediction control unit 130 receives the vehicle position information and the parking space calculation result, and determines whether or not the vehicle can be parked. The parking space can be searched for under the conditions satisfying the minimum search distance, the distance from the obstacle, the allowable steering angle, the minimum length of the parking space, and the forward limit distance after the search is completed. The automatic parking availability prediction control unit 130 generates an SVM module using the sensed value as learning data of the SVM, and then predicts whether or not the actual vehicle can be parked when parking the vehicle.

In S300, when the state of the vehicle is not possible, the alarm unit 140 generates a stop signal to the system and informs the driver that parking is not possible. When the position of the vehicle is recognized, it is judged whether or not the vehicle can park in the searched parking space. At this time, if the sensor unit measures the position change signal value of the vehicle, it is determined that parking is possible and the process returns to the first step. If the position change signal value is not measured, the system is commanded to inform the driver that parking is not possible.

7 is a block diagram of a parking availability prediction control unit according to an embodiment of the present invention.

The parking availability prediction control unit 130 predicts whether parking is possible by applying the SVM technique. Specifically, the sensor uses the values sensed from the steering angle signal, the wheel pulse signal, the yaw rate signal, the temperature signal, and the ultrasonic sensor signal for parking space recognition for vehicle position recognition. The automatic parking data feature vector extracting unit 131 extracts feature vectors from the training data. The automatic parking data feature vector processing unit 132 normalizes the automatic parking data feature vector and constructs the learning data by adding class data indicating whether parking is possible or not. The SVM module 133 includes an SVM learning module 133a for normalizing the automatic parking data feature vector and generating a support vector through SVM learning, and an SVM learning module 133a for estimating whether or not the automatic parking is possible by comparing the similarity between the support vector and the real- And an SVM classification module 133b for classifying whether or not parking is possible.

The automatic parking availability predicting unit 134 predicts whether or not parking is possible by actually inputting the actual vehicle data into the SVM classification module 133b. The automatic parking availability prediction result application unit 135 predicts whether or not the recognized vehicle can be parked in the measured parking space, and determines that parking is possible when the vehicle is in the standard. If it is out of the standard, it predicts that it is impossible to park and informs the alarm unit 140. At this time, if there is a signal for changing the position of the vehicle from the sensor, for example, a turn signal, the driver recognizes that the lane has been changed, and attempts to predict whether or not the vehicle can be parked again. If there is no signal to change the position of the vehicle, instruct the system not to park and notify the driver.

8 is a flowchart illustrating an operation of the parking availability prediction control unit according to an exemplary embodiment of the present invention.

The automatic parking data feature vector extracting unit 131 extracts feature vectors received from the vehicle position recognizing unit 110 and the parking space determining unit 120 based on the sensed values at the sensor unit at step S210. The data set of the extracted feature vector includes a vehicle speed, a gear position, a steering, a uss signal, and the like. The automatic parking data feature vector processing unit 132 adds class data indicating whether or not parking is possible to the learning data as shown in Table 1 below (S220) (S230). For example, the learning data for classifying the parking feasible feature vectors is obtained by attaching a class value to a feature vector belonging to a class that can be parked, and attaching a class value of -1 to a feature vector belonging to a class that can not be parked Then, SVM is created by using these as inputs.

Data value Class vehicle speed, gear position, steering, uss signal, ..., uss signal +1 (Parking available) vehicle speed, gear position, steering, uss signal, ..., uss signal -1 (no parking available)

The SVM module 133 learns the SVM according to each class to generate the SVM classification module 133b (S240) (S250). More specifically, the preprocessed data is used as learning data of the SVM. And we design a total of two SVMs by generating SVM that can classify binary into each class. The SVM learning then depends on the learning data set, the internal kernel functions used in learning, and the C parameter, the adjustment parameter. The kernel function of SVM converts the nonlinear input value of higher order into linear, thereby increasing the operation speed of SVM. Therefore, the SVM classification module 133b is designed by setting optimal parameters through experiments.

When the actual parking availability is predicted, the automatic parking availability predicting unit 134 automatically predicts the test data by inserting the test data into the SVM classification module 133b generated in the learning step (S260) (S270) (S280). The SVM classification module 133b outputs a value of +1 or -1, and classifies the feature vector into a class corresponding to the SVM according to the output value. The automatic parking availability prediction result application unit 135 determines that parking is possible if the vehicle is in the standard (when the +1 value is output), and returns to the step of receiving the sensor value. When the parking is not possible (when a value of -1 is outputted), the alarm unit 140 is informed to the system that a stop signal is generated and the driver is informed to try to park again. At this time, if there is a signal for changing the position of the vehicle from the sensor, for example, a turn signal, the driver recognizes that the lane has been changed and tries to park again.

The present invention has been described with reference to the preferred embodiments. It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. Therefore, the disclosed embodiments should be considered in an illustrative rather than a restrictive sense. The scope of the present invention is defined by the appended claims rather than by the foregoing description, and all differences within the scope of equivalents thereof should be construed as being included in the present invention.

100: Automatic parking availability prediction system
110: vehicle position recognizing unit 120: parking space judging unit
130: Parking availability prediction control section 140: Alarm section
150: sensor unit 200: vehicle path generation / tracking unit
300: vehicle control unit 400: vehicle screen input /

Claims (7)

  1. A vehicle position recognizing part for recognizing a vehicle position by acquiring vehicle information including a steering angle and a wheel pulse;
    A parking space determination unit for receiving an ultrasonic sensing signal from at least one ultrasonic sensor mounted on a vehicle to determine a parking space; And
    A parking space calculation unit for receiving vehicle position information from the vehicle position recognition unit and receiving a parking space calculation result from the parking space determination unit and for predicting whether or not the vehicle can be parked based on the received vehicle position information and the result of the parking space calculation; A possibility prediction control section,
    Wherein the parking availability prediction control section comprises:
    An automatic parking data feature vector extracting unit for extracting a feature vector from the vehicle location information and the parking space calculation result;
    An automatic parking data feature vector processing unit attaching class data for classifying whether or not parking is possible to the extracted automatic parking data feature vector; And
    An SVM learning module for normalizing the automatic parking data feature vector and generating a support vector by learning a support vector machine (SVM) And a SVM classification module for classifying whether or not parking is possible,
    When the actual vehicle data is actually input, the SVM classification module predicts whether or not the parking is possible, outputs +1 when the parking is possible, and outputs -1 when the parking is impossible
    Wherein the automatic parking availability prediction system comprises:
  2. delete
  3. delete
  4. delete
  5. The method according to claim 1,
    An alarm unit for alerting the driver if the parking availability prediction control unit predicts that parking is impossible;
    Further comprising: an automatic parking availability prediction system.
  6. Acquiring vehicle information including a steering angle and a wheel pulse to recognize a vehicle position, receiving an ultrasonic sensing signal from at least one ultrasonic sensor mounted on the vehicle, and determining a parking space; And
    And predicting whether or not the vehicle can be parked using the machine learning technique with the recognized vehicle location information and the determined parking space calculation result,
    The method of claim 1,
    Extracting a feature vector from the vehicle position information and the result of the parking space calculation;
    Attaching class data for classifying whether or not parking is possible to the extracted automatic parking data feature vector;
    Generating SVM (Support Vector Machine) classification module through SVM learning according to each class;
    When the actual vehicle data is actually input, the SVM classification module predicts whether or not the vehicle can be parked, outputs +1 when the parking is possible, and outputs -1 when the parking is impossible,
    The SVM classification module predicts whether or not the vehicle is parked by comparing the similarity between the support vector generated by the SVM learning module and the actual parking data feature vector,
    Wherein the automatic parking availability prediction method comprises:
  7. delete
KR1020110082213A 2011-08-18 2011-08-18 System for predicting the possibility or impossibility of vehicle parking using support vector machine KR101818535B1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
KR1020110082213A KR101818535B1 (en) 2011-08-18 2011-08-18 System for predicting the possibility or impossibility of vehicle parking using support vector machine

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
KR1020110082213A KR101818535B1 (en) 2011-08-18 2011-08-18 System for predicting the possibility or impossibility of vehicle parking using support vector machine

Publications (2)

Publication Number Publication Date
KR20130019908A KR20130019908A (en) 2013-02-27
KR101818535B1 true KR101818535B1 (en) 2018-02-21

Family

ID=47897942

Family Applications (1)

Application Number Title Priority Date Filing Date
KR1020110082213A KR101818535B1 (en) 2011-08-18 2011-08-18 System for predicting the possibility or impossibility of vehicle parking using support vector machine

Country Status (1)

Country Link
KR (1) KR101818535B1 (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR102044193B1 (en) * 2013-03-28 2019-12-02 현대모비스 주식회사 System and Method for alarming collision of vehicle with support vector machine
KR101906952B1 (en) 2013-11-08 2018-10-11 한화지상방산 주식회사 Method for generating optimized parking path of manless driving vehicle, and manless driving vehicle adopting the method
US9666074B2 (en) 2014-08-21 2017-05-30 Ford Global Technologies, Llc Method and system for vehicle parking
US9981657B2 (en) * 2016-04-14 2018-05-29 Ford Global Technologies, Llc Autonomous vehicle parking and transition to manual control

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2008087635A (en) * 2006-10-02 2008-04-17 Denso Corp Operation support device
JP2009151378A (en) * 2007-12-18 2009-07-09 Honda Motor Co Ltd Device for determining availability of parking for vehicle
JP2009205191A (en) * 2008-02-26 2009-09-10 Hitachi Ltd Parking space recognition system

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2008087635A (en) * 2006-10-02 2008-04-17 Denso Corp Operation support device
JP2009151378A (en) * 2007-12-18 2009-07-09 Honda Motor Co Ltd Device for determining availability of parking for vehicle
JP2009205191A (en) * 2008-02-26 2009-09-10 Hitachi Ltd Parking space recognition system

Also Published As

Publication number Publication date
KR20130019908A (en) 2013-02-27

Similar Documents

Publication Publication Date Title
JP5407764B2 (en) Driving assistance device
EP1754621B1 (en) Driver assistance system
US20100104199A1 (en) Method for detecting a clear path of travel for a vehicle enhanced by object detection
KR20090088210A (en) Method and apparatus for detecting target parking location by using two reference point and parking assist system using same
JP4400316B2 (en) Driving intention estimation device, vehicle driving assistance device, and vehicle equipped with vehicle driving assistance device
DE112009005424B4 (en) Object detection device and object detection method
JP4023228B2 (en) In-vehicle obstacle detection device
DE112010003874T5 (en) vehicle control
US8781720B2 (en) Vehicle travel track estimator
US9393958B2 (en) Method and system for validating information
US8812226B2 (en) Multiobject fusion module for collision preparation system
US7532109B2 (en) Vehicle obstacle verification system
US9501932B2 (en) Vehicular environment estimation device
DE102009016568B4 (en) Trajectory estimator for a moving object
EP1776602A1 (en) Method and device for operating a sensor system
WO2014192137A1 (en) Movement trajectory prediction device, and movement trajectory prediction method
US8175797B2 (en) Vehicle drive assist system
JP4790454B2 (en) Image recognition device, vehicle control device, image recognition method, and vehicle control method
US20070285305A1 (en) Obstacle detection system for vehicle
JP6196044B2 (en) Traffic signal recognition system and method
CN102439644A (en) Vehicle surrounding monitor device and method for monitoring surroundings used for vehicle
JP2015082324A (en) Composite confidence estimation for predictive driver assistant system
JP5835490B2 (en) Collision prediction device
WO2013141226A1 (en) Device and method for identifying travel section line
CN104240535A (en) Method for coordinating the operation of motor vehicles

Legal Events

Date Code Title Description
A201 Request for examination
E902 Notification of reason for refusal
E701 Decision to grant or registration of patent right
GRNT Written decision to grant