Disclosed are a method and an apparatus for estimating a lane by using a deep neural network. The method for estimating a lane by using a deep neural network according to the present invention comprises the following steps of: (a) using a plurality of sensors mounted on a vehicle to obtain sensor data for a preceding vehicle and a vehicle of a user; (b) processing the sensor data for the preceding vehicle and the vehicle of a user to select data for relative relationship with the preceding vehicle and state data for the vehicle as feature vectors to configure a training data set, wherein the training data set is mapped to a ground truth produced by a road lane model which satisfies a clothoid constraint obtained by a camera; (c) applying the training data set to a deep learning model to train the model to predict a coefficient of a road lane; and (d) applying the predicted coefficient of the road lane to a Kalman filter model which estimates lane information by using sensor data for the vehicle and removing noises to output and correct a final coefficient of the road lane. According to the present invention, a road lane can be estimated by using a relation between a preceding vehicle and the vehicle of a user without using any camera.

    심층 신경망을 이용한 차선 추정 방법 및 장치가 개시된다. 심층 신경망을 이용한 차선 추정 방법은, (a) 차량에 장착된 복수의 센서를 이용하여 선행 차량 및 자차량 관련 센서 데이터를 획득하는 단계; (b) 상기 선행 차량 및 자차량 관련 센서 데이터를 가공하여 선행 차량과의 상대적 관계 데이터와 자차량 상태 데이터를 특징 벡터들로서 선별하여 트레이닝 데이터 셋을 구성하는 단계-상기 트레이닝 데이터 셋은 카메라에 의해 획득된 클로소이드 제약을 만족하는 도로 차선 모델이 실측값으로 매핑됨; (c) 상기 트레이닝 데이터 셋을 딥러닝 모델에 적용하여 도로 차선 계수를 예측하도록 학습하는 단계; 및 (d) 상기 자차량 관련 센서 데이터를 이용하여 차선 정보를 추정하는 칼만 필터 모델에 상기 예측된 도로 차선 계수를 적용하여 노이즈를 제거하여 최종 도로 차선 계수를 출력하여 보정하는 단계를 포함한다.


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Lane estimation method and apparatus using deep neural network


    Weitere Titelangaben:

    심층 신경망을 이용한 차선 추정 방법 및 장치


    Beteiligte:
    CHUNG CHUNG CHOO (Autor:in) / KIM JIN SUNG (Autor:in) / CHOI JOO YOUNG (Autor:in)

    Erscheinungsdatum :

    2022-01-06


    Medientyp :

    Patent


    Format :

    Elektronische Ressource


    Sprache :

    Koreanisch


    Klassifikation :

    IPC:    B60W CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION , Gemeinsame Steuerung oder Regelung von Fahrzeug-Unteraggregaten verschiedenen Typs oder verschiedener Funktion / G01S RADIO DIRECTION-FINDING , Funkpeilung / G05D SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES , Systeme zum Steuern oder Regeln nichtelektrischer veränderlicher Größen / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen



    Lane estimation method and apparatus using deep neural network

    Europäisches Patentamt | 2022

    Freier Zugriff


    Vision-Based Lane-Changing Behavior Detection Using Deep Residual Neural Network

    Wei, Zhensong / Wang, Chao / Hao, Peng et al. | IEEE | 2019


    LANE ESTIMATION APPARATUS

    IRIE YOSHIRO / OKADA TOSHIKATSU / NISHIWAKI DAICHI et al. | Europäisches Patentamt | 2021

    Freier Zugriff

    Lidar-based Deep Neural Network for Reference Lane Generation

    Martinek, Philipp / Pucea, Gheorghe / Rao, Qing et al. | IEEE | 2020