실시간 객체 좌표 생성을 이용한 회피 및 전역 경로 회귀 알고리즘 개발

Abstract

The use of depth cameras and machine learning has led to innovative results in a variety of areas. Especially in the field of autonomous driving, robots can navigate complex environments and perform tasks such as obstacle avoidance through improved spatial awareness. In this paper, we developed an algorithm that avoids obstacles and returns to the global path during GPS WayPoint autonomous driving by combining an artificial potential field with a real-time object coordinate allocation algorithm using an existing depth camera and GPS. In addition, several concepts were added to the potential field algorithm to prevent the path from changing rapidly during the process of returning from the local path to the global path, and were verified through empirical experiments. In this study, the coordinates of obstacles that will generate Repulsive force in the potential field were generated using a low-cost depth camera and GPS attached to the platform instead of expensive LIDAR, and Beyond simulation, we built the concepts necessary for the robot's avoidance and regression process when actually GPS waypoint autonomous driving.