DOI QR코드

DOI QR Code

Performance Evaluation Using Neural Network Learning of Indoor Autonomous Vehicle Based on LiDAR

라이다 기반 실내 자율주행 차량에서 신경망 학습을 사용한 성능평가

  • 권용훈 (강원대학교 컴퓨터정보통신공학과) ;
  • 정인범 (강원대학교 컴퓨터정보통신공학과)
  • Received : 2022.09.20
  • Accepted : 2022.11.17
  • Published : 2023.03.31

Abstract

Data processing through the cloud causes many problems, such as latency and increased communication costs in the communication process. Therefore, many researchers study edge computing in the IoT, and autonomous driving is a representative application. In indoor self-driving, unlike outdoor, GPS and traffic information cannot be used, so the surrounding environment must be recognized using sensors. An efficient autonomous driving system is required because it is a mobile environment with resource constraints. This paper proposes a machine-learning method using neural networks for autonomous driving in an indoor environment. The neural network model predicts the most appropriate driving command for the current location based on the distance data measured by the LiDAR sensor. We designed six learning models to evaluate according to the number of input data of the proposed neural networks. In addition, we made an autonomous vehicle based on Raspberry Pi for driving and learning and an indoor driving track produced for collecting data and evaluation. Finally, we compared six neural network models in terms of accuracy, response time, and battery consumption, and the effect of the number of input data on performance was confirmed.

클라우드를 통한 데이터 처리는 통신 과정에서 지연시간과 통신비용 증가 등 같은 많은 문제가 발생한다. 사물인터넷 분야에서는 이러한 문제를 해결하기 위해 엣지 컴퓨팅 연구가 활발히 이루어지고 있고, 대표적인 응용 분야로 자율주행이 있다. 실내 자율주행에서는 실외와 달리 GPS와 교통정보를 이용할 수 없기 때문에 센서를 활용하여 주변 환경을 인식해야 한다. 그리고 자원이 제약된 모바일 환경이기 때문에 효율적인 자율주행 시스템이 필요하다. 본 논문에서는 실내 환경에서 자율주행을 위해 신경망을 사용하는 기계학습을 제안한다. 신경망 모델은 LiDAR 센서에서 측정된 거리 데이터를 바탕으로 현재 위치에 가장 적절한 주행 명령을 예측한다. 신경망의 입력 데이터의 수에 따른 성능 평가를 수행하기 위해 6가지의 학습 모델을 설계하였다. 주행과 학습을 위해 Raspberry Pi 기반의 자율주행 차량을 제작하였고, 학습 데이터 수집과 성능평가를 위한 실내 주행 트랙을 제작하였다. 6가지의 신경망 모델들은 정확도와 응답시간 그리고 배터리 소모에 대한 성능 비교를 하였고, 입력 데이터의 수가 성능에 미치는 영향을 확인하였다.

Keywords

References

  1. W. Kang, Y. Jung, and I. Hwang, "Major technology of autonomous driving," KICS Information & Communication Magazine Open Lecture Series, Vol.35, No.1, pp.28-35, 2018.
  2. N. Abbas, Y. Zhang, A. Taherkordi, and T. Skeie, "Mobile edge computing: A survey," IEEE Internet of Things Journal, Vol.5, No.1, pp.450-465, 2018. https://doi.org/10.1109/JIOT.2017.2750180
  3. S. Jeong, "Edge computing and vehicles: Opportunities and challenges for the future," The Journal of Korean Institute of Communications and Information Sciences, Vol.46, No.5, 2021.
  4. N. El-Sheimy and Y. Li, "Indoor navigation: State of the art and future trends," Satellite Navigation, Vo1.2, No.1, pp.1-23, 2021. https://doi.org/10.1186/s43020-020-00033-9
  5. A. A. Suzen, B. Duman, and B. Sen, "Benchmark analysis of jetson TX2, jetson nano and raspberry PI using Deep-CNN," 2020 International Congress on Human -Computer Interaction, Optimization and Robotic Applications (HORA), pp.1-5, 2020.
  6. J. Noh, K.-M. Yang, M.-R. Park, J.-W. Lee, M.-G. Kim, and K.-H. Seo, "LiDAR point cloud augmentation for mobile robot safe navigation in indoor environment," Journal of Institute of Control, Robotics and Systems, Vol.28, No.1, pp.52-58, 2022. https://doi.org/10.5302/J.ICROS.2022.21.0209
  7. Q. Zou, Q. Sun, L. Chen, B. Nie, and Q. Li, "A comparative analysis of LiDAR SLAM-based indoor navigation for autonomous vehicles," IEEE Transactions on Intelligent Transportation Systems, Vol.23, No.7, pp.6907-6921, 2022. https://doi.org/10.1109/TITS.2021.3063477
  8. H. Kim and Y. S. Choi, "Development of a ROS-Based autonomous driving robot for underground mines and its waypoint navigation experiments," Tunnel & Underground Space, Vol.32, No.3, pp.231-242, 2022.
  9. Y. J. Choi, T. Rahim, I. N. A. Ramatryana, and S. Y. Shin, "Improved CNN-based path planning for stairs climbing in autonomous UAV with LiDAR sensor," 2021 International Conference on Electronics, Information, and Communication (ICEIC), pp.1-7, 2021.
  10. D. G. Park, K. R. Lee, J. W. Jang, and D. H. Kim "ROS-based control system for localization and object identification of indoor self-driving mobile robot," Transactions of the KSME A, Vol.45, No.12, pp.1149-1160, 2021.
  11. Y. Quan, L. Huang, L. Ma, Y. He, and R. Wang, "Neural Network-Based Indoor Autonomously-Navigated AGV Motion Trajectory Data Fusion," Automatic Control and Computer Sciences, Vol.55, No.4, pp.334-345, 2021. https://doi.org/10.3103/S0146411621040076
  12. D. T. Son, M. T. Anh, D. D. Tu, L. V. Chuong, T. H. Cuong, and H. S. Phuong, "The practice of mapping-based navigation system for indoor robot with RPLIDAR and raspberry Pi," 2021 International Conference on System Science and Engineering (ICSSE), IEEE, 2021.
  13. M. Liao, D. Wang, and H. Yang. "Deploy indoor 2D laser SLAM on a raspberry pi-based mobile robot," 2019 11th International Conference on Intelligent Human-Machine Systems and Cybernetics (IHMSC), Vol.2. IEEE, 2019.
  14. H. Leon Araujo et al., "Autonomous mobile robot implemented in LEGO EV3 integrated with raspberry Pi to use android-based vision control algorithms for humanmachine interaction," Machines, Vol.10, No.3, pp.193, 2022.
  15. M. Calcroft and A. Khan, "LiDAR-based obstacle detection and avoidance for autonomous vehicles using raspberry Pi 3B," 2022 UKACC 13th International Conference on Control (CONTROL), pp.24-29, 2022.
  16. Raspberry Pi [Internet], https://www.raspberrypi.org
  17. RPLIDAR A1M8 [Internet], https://www.slamtec.com