In micro electrical discharge machining （micro EDM）, it is difficult for servo controlling the narrow discharge gap with the characters of non-linear and quick change. In this paper, aiming at solving the problems a...In micro electrical discharge machining （micro EDM）, it is difficult for servo controlling the narrow discharge gap with the characters of non-linear and quick change. In this paper, aiming at solving the problems above, a self-adaptive fuzzy controller with formulary rule （SAFCFR） is presented based on the dual feedbacks composed by gap electric signal and discharge-ratio statistics. To ensure the properties of self-optimizing and fast stabilization, the formulary rule was designed with a tuning factor. In addition, the fast-convergence algorithms were introduced to adjust control target center and output scale factor. In this way, the normal discharge ratio can tend to the highest value during micro-EDM process. Experimental results show that the proposed algorithms are effective in improving the servo-control performance. According to the drilling-micro-EDM experiments, the machining efficiency is improved by 20% through applying SAFCFR. Moreover, SAFCFR is a prompt way to optimize parameters of discharge-gap servo control.展开更多
目的目前已有的单目视觉SLAM(simultaneous localization and mapping)系统每次开始运行时都将初始帧而不是绝对位置设置为参考帧,不能在一个固定的坐标系中获得位姿,导致无法重用已有的建图信息,而且在复杂场景中相机容易跟踪失败,需...目的目前已有的单目视觉SLAM(simultaneous localization and mapping)系统每次开始运行时都将初始帧而不是绝对位置设置为参考帧,不能在一个固定的坐标系中获得位姿,导致无法重用已有的建图信息,而且在复杂场景中相机容易跟踪失败,需要当前帧与已有的关键帧非常相似时才能重定位并继续建图。针对这个问题,提出一种具有重新初始化、地图重用与地图恢复能力的视觉SLAM系统。方法首先,加载先验地图,通过ORB(oriented brief)特征匹配SLAM系统当前帧与先验地图关键帧,并结合重定位方法完成SLAM系统的初始化。接着,为了避免丢失地图,建立一种应对SLAM系统跟踪失败的地图保存机制,保存跟踪成功地图,并提出一种自适应快速重新初始化算法,引入灭点检测,自动选择最佳重新初始化策略,保证SLAM系统继续跟踪与建图,建立的地图称为恢复地图。最后,对于跟踪成功地图与恢复地图,采用改进的回环方法获得它们之间的转换关系,并提出一种地图恢复法,减少跟踪成功地图与恢复地图尺度不一带来的误差,确保得到的全局一致地图更加准确。结果在经过加噪处理的KITTI数据集上进行地图恢复融合的测试,实验结果表明,在KITTI00、KITTI02、KITTI05数据集下,本文提出的SLAM系统比ORB-SLAM2系统分别可以多获得39.25%、47.75%、32.46%的地图信息。在EuRo C数据集上的运行结果表明,本文提出的单目视觉SLAM系统不仅在建图精度方面与ORB-SLAM2效果相当,还在跟踪稳定性方面有显著提升。结论本文提出的SLAM系统可以在跟踪失败的情况下有效恢复地图;此外,还可以高效重用SLAM系统已有的建图结果,固定SLAM地图坐标系,提升SLAM系统运行稳定性。展开更多
基金Supported by the National High Technology Research and Development Program of China （No. 2007AA04Z346） , the National Natural Science Foundation of China （ No. 50905094） and China Postdoctoral Science Foundation （ No. 20080440378, 200902097）.
文摘In micro electrical discharge machining （micro EDM）, it is difficult for servo controlling the narrow discharge gap with the characters of non-linear and quick change. In this paper, aiming at solving the problems above, a self-adaptive fuzzy controller with formulary rule （SAFCFR） is presented based on the dual feedbacks composed by gap electric signal and discharge-ratio statistics. To ensure the properties of self-optimizing and fast stabilization, the formulary rule was designed with a tuning factor. In addition, the fast-convergence algorithms were introduced to adjust control target center and output scale factor. In this way, the normal discharge ratio can tend to the highest value during micro-EDM process. Experimental results show that the proposed algorithms are effective in improving the servo-control performance. According to the drilling-micro-EDM experiments, the machining efficiency is improved by 20% through applying SAFCFR. Moreover, SAFCFR is a prompt way to optimize parameters of discharge-gap servo control.
