ETRI-Knowledge Sharing Plaform

KOREAN
논문 검색
Type SCI
Year ~ Keyword

Detail

Conference Paper 자율주행을 위한 TORCS 기반 End-to-End 학습
Cited - time in scopus Share share facebook twitter linkedin kakaostory
Authors
최두섭, 안택현, 안경환, 최정단
Issue Date
2017-11
Citation
대한전자공학회 종합 학술 대회 (추계) 2017, pp.740-743
Publisher
대한전자공학회
Language
Korean
Type
Conference Paper
Abstract
In this paper, we trained a convolutional neural network (CNN) to map a single front-facing camera image from The Open Racing Car Simulator (TORCS) directly to steering commands. To this end, we developed the driving experience data acquisition system, which saves the front-facing camera images and the corresponding steering commands during racing at 10Hz. The collected data is then used to train a deep convolutional neural network. At the end of this paper, we analyze how the collected data affects the driving quality of the end-to-end driving system.
KSP Keywords
Camera Image, Car simulator, Convolution neural network(CNN), Data Acquisition(DAQ), Deep convolutional neural networks, Driving experience, Driving system, End to End(E2E), Front-facing camera, Racing car, data acquisition system