论文部分内容阅读
机器人抓取模式选择主要是利用人的抓取经验来进行的,具有一定的不确定性和模糊性。本文根据这一特点,以研制的形状自适应手爪抓取模式分类为基础,在综合考虑抓取任务和物体特征的同时。采用模糊的输入方式,同时在保持分类能力不变的前提下,采用粗糙集理论从训练样本中提取和精简规则来构建模糊神经网络。利用神经网络良好的分类特性来选择合适的抓取模式,减少了网络输入,简化了网络拓扑结构,缩短了训练时间,提高了抓取的自动化水平。最后通过抓取实验验证了抓取模式选择的正确性。
Robot crawling mode selection is mainly based on the use of human crawling experience to carry out, with a certain degree of uncertainty and ambiguity. Based on this feature, this paper based on the classification of adaptive gripper gripper pattern, and considering the task of capturing and the characteristics of objects at the same time. Adopting the fuzzy input method, the fuzzy neural network is constructed by using the rough set theory to extract and reduce the rules from the training samples while keeping the classification ability unchanged. The use of neural networks to select a good classification of the appropriate crawling mode, reducing network input, simplifying the network topology, shorten the training time and improve the level of automation of the crawl. At last, the correctness of the selection of crawling mode is verified by the experiment of crawling.