Obstacle Avoidance in Mobile Robots In RGB-D Images Using Deep Neural Network and Semantic Segmentation

Al-Adhamı A. A., Cansever G.

4th INTERNATIONAL CONGRESS ON HUMAN-COMPUTER INTERACTION, OPTIMIZATION AND ROBOTIC APPLICATIONS, Ankara, Türkiye, 9 - 11 Haziran 2022, ss.1-5

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Basıldığı Şehir: Ankara
  • Basıldığı Ülke: Türkiye
  • Sayfa Sayıları: ss.1-5
  • Yıldız Teknik Üniversitesi Adresli: Evet

Özet

Abstract— Robotics today is integrated into the modern environment in automated production systems where manipulative robots are used in commercial and industrial activities to maneuver or have a variety of products available. On the other hand, in the area of ​​mobile robots, they have another method of operation such as obstacle avoidance, line trackers and mobile explorers; These robots are required to have a greater demand due to their high complexity in their operating system since they are used in research methods. For this reason, we propose a vision system whose main purpose is to detect and avoid obstacle in know environments by deploying a feature extraction and classification algorithm DNN of obstacles in RGB-D images.

Keywords—CNN, SVM, DL, ML, Robotics, RGB-D