基于单目视觉的采摘机器人障碍物检测和路径规划

Obstacle Detection and Path Planning of Picking Robot Based on Monocular Vision

具有自主作业能力的采摘机器人一直是国际上研究的热点,而障碍物检测躲避能力是其重要的功能,因为在机器人识别作业区域或成熟果实后需要自主的定位和移动。为此,提出了一种基于单目视觉和人工势能场的障碍物检测和避障算法,可以有效采集和检测障碍物的信息,再依据障碍物及目标区域的距离使用人工势能场方法对路径进行优化,实现采摘机器人的自主移动。为了验证障碍物检测和避障方案的可行性,模拟采摘机器人作业环境和自主移动流程,对采摘机器人避障行为进行了测试。测试结果表明:采用单目视觉和人工势场方法可以使机器人成功的避障,并规划出效率最高的到达目标作业区域路径,对采摘机器人自主导航技术的研究具有重要的意义。

It has always been a research hotspot for the harvesting robot with autonomous operation ability in the world, whose important function is obstacle detection and avoidance ability, because the robot needs autonomous positioning and movement after identifying the operation area or mature fruit.Above it, it proposed an obstacle detection and avoidance algorithm based on monocular vision and artificial potential field.Monocular vision can effectively collect and detect the information of obstacles.Then, according to the distance between the obstacles and the target area, it used artificial potential field method to optimize the path to realize the autonomous movement of the picking robot.In order to verify the feasibility of the obstacle detection and avoidance scheme,it tested the obstacle avoidance behavior of the picking robot by simulating the working environment and autonomous movement process of the picking robot.The test results show that the robot can successfully avoid obstacles by using monocular vision and artificial potential field method,and the path to the target operation area with the highest efficiency is planned for the picking robot.The research of navigation technology is of great significance.