摘要
通过激光标志物的卷积神经网络(CNN)检测,与标志物中心点的奇异值分解(SVD)重构,实现了掘进机在巷道坐标系下的坐标估计。通过基于支撑向量数据描述(SVDD)的陀螺仪静止状态抖动抑制,与参考系变换,实现了机身与掘进臂的姿态检测。通过基于Open GL的图形学引擎,实现了工作面场景的实时虚拟渲染。测试结果表明:系统能够准确可靠地完成工作面场景下掘进机监测任务。
By detection of convolutional neural network( CNN) of laser marker and reconstruction of singular value decomposition( SVD) of center of markers,coordinates estimation of machine in tunnel coordinate is achieved. By jitter suppression of gyro in resting state based on supporting vector data description( SVDD) and transformation of reference system,body and robotic arm altitude estimation are achieved. Based on Open GL,real-time virtual rendering of scene of working face is achieved. Experimental results show that the system is able to monitor mining machine in working face scene accurately and reliably.
出处
《传感器与微系统》
CSCD
2016年第10期76-79,共4页
Transducer and Microsystem Technologies
关键词
卷积神经网络
奇异值分解
支撑向量数据描述
三维重构
姿态估计
convolutional neural network(CNN)
singular value decomposition(SVD)
support vector data description(SVDD)
3D reconstruction
altitude estimation