Robust Neural Network for Sim-to-Real Gap in End-to-End Autonomous Driving

被引：1
|
作者：
Pareigis, Stephan [1 ]
Maass, Fynn Luca [1 ]
机构：
[1] HAW Hamburg, Dept Comp Sci, Berliner Tor 7, D-20099 Hamburg, Germany
来源：
PROCEEDINGS OF THE 19TH INTERNATIONAL CONFERENCE ON INFORMATICS IN CONTROL, AUTOMATION AND ROBOTICS (ICINCO) | 2022年
关键词：
Sim-to-Real Gap; End-to-End Learning; Autonomous Driving; Artificial Neural Network; CARLA Simulator; Robust Control; PilotNet;
D O I：
10.5220/0011140800003271
中图分类号：
TP [自动化技术、计算机技术];
学科分类号：
0812 ;
摘要：
A neural network architecture for end-to-end autonomous driving is presented, which is robust against discrepancies in system dynamics during the training process and in application. The proposed network architecture presents a first step to alleviate the simulation to reality gap with respect to differences in system dynamics. A vehicle is trained to drive inside a given lane in the CARLA simulator. The data is used to train NVIDIA's PilotNet. When an offset is given to the steering angle of the vehicle while the trained network is being applied, PilotNet will not keep the vehicle inside the lane as expected. A new architecture is proposed called PilotNet., which is robust against steering angle offsets. Experiments in the simulator show that the vehicle will stay in the lane, although the steering properties of the vehicle differ.
引用
收藏
页码：113 / 119
页数：7
相关论文
共 50 条
  • [1] Sim-to-real reinforcement learning applied to end-to-end vehicle control
    Kalapos, Andras
    Gor, Csaba
    Moni, Robert
    Harmati, Istvan
    [J]. 2020 23RD IEEE INTERNATIONAL SYMPOSIUM ON MEASUREMENT AND CONTROL IN ROBOTICS (ISMCR), 2020,
  • [2] End-to-end autonomous driving based on the convolution neural network model
    Zhao, Yuanfang
    Chen, Yunli
    [J]. 2019 ASIA-PACIFIC SIGNAL AND INFORMATION PROCESSING ASSOCIATION ANNUAL SUMMIT AND CONFERENCE (APSIPA ASC), 2019, : 419 - 423
  • [3] End-to-end Learning Approach for Autonomous Driving: A Convolutional Neural Network Model
    Wang, Yaqin
    Liu, Dongfang
    Jeon, Hyewon
    Chu, Zhiwei
    Matson, Eric T.
    [J]. PROCEEDINGS OF THE 11TH INTERNATIONAL CONFERENCE ON AGENTS AND ARTIFICIAL INTELLIGENCE (ICAART), VOL 2, 2019, : 833 - 839
  • [4] Segmented Encoding for Sim2Real of RL-based End-to-End Autonomous Driving
    Chung, Seung-Hwan
    Kong, Seung-Hyun
    Cho, Sangjae
    Nahrendra, I. Made Aswin
    [J]. 2022 IEEE INTELLIGENT VEHICLES SYMPOSIUM (IV), 2022, : 1290 - 1296
  • [5] Learning for Attitude Holding of a Robotic Fish: An End-to-End Approach With Sim-to-Real Transfer
    Zheng, Junzheng
    Zhang, Tianhao
    Wang, Chen
    Xiong, Minglei
    Xie, Guangming
    [J]. IEEE TRANSACTIONS ON ROBOTICS, 2022, 38 (02) : 1287 - 1303
  • [6] Sim-to-Real Deep Reinforcement Learning for Safe End-to-End Planning of Aerial Robots
    Ugurlu, Halil Ibrahim
    Xuan Huy Pham
    Kayacan, Erdal
    [J]. ROBOTICS, 2022, 11 (05)
  • [7] An End-to-End Deep Neural Network for Autonomous Driving Designed for Embedded Automotive Platforms
    Kocic, Jelena
    Jovicic, Nenad
    Drndarevic, Vujo
    [J]. SENSORS, 2019, 19 (09)
  • [8] NEAT: Neural Attention Fields for End-to-End Autonomous Driving
    Chitta, Kashyap
    Prakash, Aditya
    Geiger, Andreas
    [J]. 2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021), 2021, : 15773 - 15783
  • [9] End-to-End Autonomous Driving: An Angle Branched Network Approach
    Wang, Qing
    Chen, Long
    Tian, Bin
    Tian, Wei
    Li, Lingxi
    Cao, Dongpu
    [J]. IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, 2019, 68 (12) : 11599 - 11610
  • [10] Multimodal End-to-End Autonomous Driving
    Xiao, Yi
    Codevilla, Felipe
    Gurram, Akhil
    Urfalioglu, Onay
    Lopez, Antonio M.
    [J]. IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, 2022, 23 (01) : 537 - 547
12345