Deep Learning-based Target Following and Obstacle Avoidance Methods in Mobile Robots

Authors:

Abstract
In this study, we develop a companion robot function that can recognize people and obstacles, and track the human target while avoiding the obstacles. In implementing the JetBot tracking function, a deep learning-based object detection model is modified and utilized. A transfer learning algorithm with a pretrained mobile model is introduced for a feasible operation in the proposed framework with mobile robots.
Funding Information
  • National Research Foundation of Korea (NRF-2021R1F1A1050977)