Target Distance Calculation Method Using Image Segmentation

摘要

Today we have made remarkable progress in the field of hardware and software. We can design many systems that provide high performance at low costs. Those systems replace humans working in places where death is likely to occur with robots. This paper proposes a system in which a robot moves to given GPS points, then searches for targets and calculates the distances. We used GPS sensors and robot-mounted sensors to move the robot to a given GPS point. When it reaches the point, it scans its surroundings and searches for humans as targets. It uses a stereo camera to take photos and the image segmentation to detect targets. The robot calculates the distance to the target and estimates the target's GPS coordinates and returns it to the operator. The results showed the calculated distance from 50cm to 200cm with an error rate of 2-4%.