苹果采摘机器人监测系统和表面缺陷检测方法研究

Research on apple picking robot monitoring system and surface defect detection method

摘要:

针对目前苹果采摘作业多以人工采摘为主,且机械化采摘质量和效率不高等问题,提出一种苹果采摘机器人监测系统。监测系统通过种植园环境、机器人运行状态、机器人作业质量以及后台监控系统四个模块,完成从采收前环境监测到采收后质量监测的全程监控。同时,为监测采摘后苹果的质量问题,提出了一种基于Mo-M2Det的苹果表面缺陷检测方法。以减少参数量和计算量为目的,改进M2Det目标检测网络,实现轻量化和高精度检测。实验结果表明,改进的M2Det目标检测网络检测准确率达到了98.45%,且模型参数量减少了56.3%。在实际应用中,改进的轻量化网络模型部署在检测平台上具有较好的检测效果。

Abstract:

In view of the current problem that apple picking operations are mainly manual picking, and the quality and efficiency of mechanised picking are not high, an apple picking robot monitoring system is proposed. The monitoring system completes the monitoring of the whole process from pre-harvest environmental monitoring to post-harvest quality monitoring through four modules: plantation environment, robot operation status, robot operation quality and background monitoring system. At the same time, in order to monitor the quality of apples after picking, a method for detecting apple surface defects based on Mo-M2Det is proposed. With the aim of reducing the amount of parameters and calculations, the M2Det target detection network is improved to achieve lightweight and high-precision detection. The experimental results show that the detection accuracy of the improved M2Det target detection network has reached 98.45%,and the number of model parameters has been reduced by 56.3%. In practical application, the improved lightweight network model has a good detection effect on the detection platform.