Person Following Algorithm with Pixel-Area Addition Method of Thermal Sensors for Autonomous Mobile Robots

Abstract

This paper discusses development of a small au-tonomous mobile robot without cameras for visual sensing, and newly proposes a human tracking algorithm and optimal control method using a thermal array sensor capable of acquiring 64 pixels of thermal images. This robot is equipped with two thermal array sensors, measuring a human's body temperature in real time from thermal images of 128 pixels. In this paper, the heat distribution is divided into seven areas, and we determine the direction of movement for a robot in order to generate an appropriate following motion. The proposed tracking control algorithm adds some pixels of the thermal image and selects the highest temperature area by sorting. This algorithm enables the robot to calculate the direction of movement and appropriately outputs speed commands to the four drive wheels by combining proportional control method. Finally, we conduct tracking experiments on a circular course and a straight-line course to clarify the validity of the proposed method. In this experiment, we evaluate the performance of the four-pixel addition and nine-pixel addition methods, and show their usefulness and characteristic differences.