In this paper,radar graph fractal method is introduced to describe wear debris boundaries. Research results show that it is a nice way to describe wear debris boundaries. Since the longest axis is selected as the firs...In this paper,radar graph fractal method is introduced to describe wear debris boundaries. Research results show that it is a nice way to describe wear debris boundaries. Since the longest axis is selected as the first coordinate axis,its center point selected as the center point of the radar graph,and the co-ordinate value of wear debris boundary selected as the measure parameter,the limitations existing in Yard fractal measure method can be avoided. For any wear debris,its radar graph fractal dimension value is one and only,and as the wear debris shape changes from round to strip,the radar graph fractal dimension value also changes from low to high,showing strong uniqueness and independence. Due to the fact that the researched wear debris is gotten in different wear states,the results also prove that radar graph fractal dimension value is correlated with frictional pairs work condition and wear state. Radar graph fractal method is compared with Yard fractal measure methods,and results show that ra-dar graph fractal dimension values gotten from different wear debris have enough value grads to avoid effect of errors,and provide higher sensitivity for wear debris shape. This paper also discusses the influencing factors for radar graph fractal method. With the increase of the decomposing degree value,the radar graph fractal dimension tends to keep stable at one certain value,showing the typical characteristic of the fractal theory. All this proves that radar graph fractal method is an effective description method for wear debris boundaries.展开更多
针对单一传感器SLAM(Simultaneous Localization And Mapping)技术在复杂环境中存在精度低、可靠性差等问题,提出一种基于因子图消元优化的激光雷达、视觉和IMU(Inertial Measurement Unit)融合SLAM算法(Multi Factor Graph fusion SLAM...针对单一传感器SLAM(Simultaneous Localization And Mapping)技术在复杂环境中存在精度低、可靠性差等问题,提出一种基于因子图消元优化的激光雷达、视觉和IMU(Inertial Measurement Unit)融合SLAM算法(Multi Factor Graph fusion SLAM with IMU as the Dominant system,ID-MFG-SLAM).首先,采用多因子图模型,提出以IMU为主系统,视觉与激光雷达为辅系统,通过引入辅系统观测因子约束IMU偏差,并接收IMU里程计因子实现运动预测与融合的全新结构.之后,为降低融合后的优化成本,加入滑窗机制并设计基于Householder变换的QR分解消元法将因子图转换为贝叶斯网络.最后,引入一种球面线性插值与线性插值之间的自适应插值算法,将激光雷达点云投影到单位球体上实现视觉特征点深度估计.实验结果表明,相比其他经典算法,该方法在复杂大、小场景中绝对轨迹误差分别可达到约0.68 m和0.24 m,具有更高的精度和可靠性.展开更多
基金Partially supported by Key Program of the National Natural Science Foundation of China (Grant No. 50535050)Major State Basic Research Development Program (973 Program, 2007CB607605)
文摘In this paper,radar graph fractal method is introduced to describe wear debris boundaries. Research results show that it is a nice way to describe wear debris boundaries. Since the longest axis is selected as the first coordinate axis,its center point selected as the center point of the radar graph,and the co-ordinate value of wear debris boundary selected as the measure parameter,the limitations existing in Yard fractal measure method can be avoided. For any wear debris,its radar graph fractal dimension value is one and only,and as the wear debris shape changes from round to strip,the radar graph fractal dimension value also changes from low to high,showing strong uniqueness and independence. Due to the fact that the researched wear debris is gotten in different wear states,the results also prove that radar graph fractal dimension value is correlated with frictional pairs work condition and wear state. Radar graph fractal method is compared with Yard fractal measure methods,and results show that ra-dar graph fractal dimension values gotten from different wear debris have enough value grads to avoid effect of errors,and provide higher sensitivity for wear debris shape. This paper also discusses the influencing factors for radar graph fractal method. With the increase of the decomposing degree value,the radar graph fractal dimension tends to keep stable at one certain value,showing the typical characteristic of the fractal theory. All this proves that radar graph fractal method is an effective description method for wear debris boundaries.
文摘针对单一传感器SLAM(Simultaneous Localization And Mapping)技术在复杂环境中存在精度低、可靠性差等问题,提出一种基于因子图消元优化的激光雷达、视觉和IMU(Inertial Measurement Unit)融合SLAM算法(Multi Factor Graph fusion SLAM with IMU as the Dominant system,ID-MFG-SLAM).首先,采用多因子图模型,提出以IMU为主系统,视觉与激光雷达为辅系统,通过引入辅系统观测因子约束IMU偏差,并接收IMU里程计因子实现运动预测与融合的全新结构.之后,为降低融合后的优化成本,加入滑窗机制并设计基于Householder变换的QR分解消元法将因子图转换为贝叶斯网络.最后,引入一种球面线性插值与线性插值之间的自适应插值算法,将激光雷达点云投影到单位球体上实现视觉特征点深度估计.实验结果表明,相比其他经典算法,该方法在复杂大、小场景中绝对轨迹误差分别可达到约0.68 m和0.24 m,具有更高的精度和可靠性.
