低内存占用采摘作业机器人设计——基于分类器和top-k优化算法

Design for Picking Robot of Low Memory Occupation Based on the Classifier and Top-k Optimization Algorithm

为了提高采摘作业机器人的智能识别和自主作业能力,对路径规划大数据背景下的监测数据进行实时处理,并降低计算和通信过程中占用的内存,提出了一种基于K邻近和top-k的数据分类及低内存占用优化算法。利用该算法可以将机器人路径规划的各个节点数据进行筛选,依据K邻近算法对路径进行优选,调整实时监控数据的分类结果和数据处理流程,从而使数据更新可以不依赖于网络通信,只将少部分数据进行通信传输,有效减轻了通信负担。以机器人的采摘作业为实验对象,对优化算法进行了实验验证,结果表明:采摘机器人采用实时监测数据分类器可以对大量的数据进行有效地筛选,从而降低了通信负担,提高了内存的利用率和采摘作业的效率。

In order to improve the intelligent recognition and independent operation ability of picking robot, real-time monitoring data under the background of big data for path planning, and reduce the memory footprint of the computing and communication process, it proposed an optimization algorithm of K and top-k occupy adjacent data classification and low memory based on. By using this algorithm, each node can be data path planning, which were selected on the basis of K neighbor algorithm to optimize the path, adjusting the classification results of real-time monitoring data and data pro- cessing, so that the data update can not rely on the network communication, only small part of data transmission, effec- tively reduce the burden of communication. By picking robot as the experimental object, the optimization algorithm is veri- fied by experiments, the experimental results show that the picking robot real-time monitoring data classifier can effec- tively screen the large amounts of data, thereby reducing the communication burden, improve the efficiency of memory u- tilization and picking operation.