A Robot Companion Algorithm for Side-by-Side Object Tracking and Following

Abstract

There have been several research on using a robot instead of a human for the task of object following and even human following from behind. However, very few numbers of research on side-by-side human-following robots have been conducted. An algorithm for a side-by-side human-following robot was designed, implemented, and tested in this paper. The concept design of the algorithm was inspired by human behaviors when 2 people walk together. It allows the robot to move alongside a person attached by a color tag. A small, smart toy car with a camera was used as the companion, and a wheelchair with a better-quality camera. The OpenCV library was used on a Raspberry Pi board. Both robots underwent 3 scenarios in the performance test: moving forward, backward, and stopping. The test was conducted to find the possibility of the algorithm, accuracy, and response time. During the test, a person, followed by the robot, moved forward, backward, and stopped repeatedly and unorderly. The results showed that the algorithm successfully served the objectives of this research in terms of the possibility. However, the toy cars and wheelchairs' succession rates were 81% and 95 % respectively. There were delays in the respondence due to the limitation of memory of the microcontroller and the quality of the real-time image from the lower quality of the camera. Future approaches could be the use of a real-time operating system (RTOS) to allow the robot to execute multiple tasks independently and smoothly and better hardware choices.