基于深度学习的四旋翼无人机单目视觉避障方法

针对无人机避障问题，提出一种基于深度学习的四旋翼无人机单目视觉避障方法。首先通过目标检测框选出目标在图像中的位置，并通过计算目标选框上下边距的长度，以此来估量出障碍物到无人机之间的距离；然后通过协同计算机判断是否执行避障动作；最后使用基于Pixhawk搭建的飞行实验平台进行实验。实验结果表明，该方法可用于无人机低速飞行条件下避障。该方法所用到的传感器只有一块单目摄像头，而且相对于传统的主动式传感器避障方法，所占用无人机的体积大幅减小。该方法鲁棒性较好，能够准确识别不同姿态的人，实现对人避障。     

基于动态距离的模糊社区识别算法

社区识别技术是公共安全领域潜在危害行为预警预测和已发生危害行为追踪溯源的基础，针对传统社区识别算法将社区视作单一集合而无法描述社区主次成员的问题，提出一种基于动态距离的模糊社区识别算法。该算法将传统的单一社区结构划分为核心区域和边际区域，以边际区域来描述社区的模糊区间。该算法首先将网络设想为动态演变模型，网络中的任意节点均会与其他节点产生互动，互动将改变各节点间距离，距离也反过来影响互动。在阈值的界定下，受到多个社区吸引的节点被划分到边际区域，最终距离分布趋于稳定，各个社区结构得以显现。通过对比实验验证了CDFDD算法在社区识别上的有效性。     

基于时空的弱点状运动目标检测技术

如何在复杂背景下持续有效地检测目标位置,一直是研究者们需要面对的主要挑战。本文在研究红外点状移动目标特征的基础上,根据目标无纹理,无形状的特性,提出一种改进的形态学目标增强算法,并利用目标连续时空不变性检测目标。首先,建立多尺度的图像金字塔,在每层上采用改进的形态学算法快速、粗糙定位小目标。然后进一步的根据目标在时空上的位置相关性,提出基于目标运动特征分析的精确检测方法。得到精确稳定的检测结果。最后实验结果表明与经典的形态学检测算法及其他算法相比,该技术能更有效地检测弱小目标,具有更高的鲁棒性。     

A Novel Approach for Bisphosphonates: Ionic Liquids and Organic Salts from Zoledronic Acid

Novel ionic liquids and organic salts based on mono- or dianionic zoledronate and protonated superbases, choline and n-alkylmethylimidazolium cations, were prepared and characterized by spectroscopic and thermal analyses. Most of the prepared salts display amorphous structures and very high solubility in water and saline solutions, especially the dianionic salts. Among the zoledronate-based ionic compounds, those containing choline [Ch] and methoxyethylmethylimidazolium [C₃OMIM] cations appear to have significant cytotoxicity against human osteosarcoma cells (MG63) and low toxicity toward healthy skin fibroblast cells. Because osteosarcoma is a bone pathology characterized by an increase in bone turnover rate, the results presented herein may be a promising starting point for the development of new ionic pharmaceutical drugs against osteosarcoma.     

Visual detection and tracking with UAVs,following a mobile object

ABSTRACT

Conception and development of an Unmanned Aerial Vehicle (UAV) capable of detecting, tracking and following a moving object with unknown dynamics is presented in this work, considering a human face as a case of study. Object detection is accomplished by a Haar cascade classifier. Once an object is detected, it is tracked with the help of a Kalman Filter (KF), and an estimation of the relative position with respect to the target is obtained. A linear controller is used to validate the proposed vision scheme and for regulating the aerial robot's position in order to keep a constant distance with respect to the mobile target, employing as well the extra available information from the embedded sensors. The proposed system was extensively tested in real-time experiments, through different conditions, using a commercial quadcopter connected via wireless to a ground station running under the Robot Operative System (ROS). The proposed overall strategy shows a good performance even under disadvantageous conditions as outdoor flight, being robust against illumination changes, image noise and the presence of other people in the scene.     

Poly(L-lysine)-poly(ethylene glycol) layers with different structure and their influence on silica suspension stability

The effects of the structure of di- and triblock copolymers of poly(L-lysine) – LYS with poly(ethylene glycol) – PEG as well as the length of nonionic fragment in the LYS-PEG macromolecule on the copolymer chains conformation in the adsorption layer formed on the colloidal silica (SiO2) surface were examined. Spectrophotometry and turbidimetry were applied for the determination of copolymer adsorbed amounts and stability coefficients of silica aqueous suspensions. The electrokinetic parameters such as solid surface charge density and zeta potential were also estimated. The adsorption of LYS-PEG was proved to be the highest at pH 10 whereas the lowest adsorption on the solid surface was found for the triblock copolymer with long fragments of LYS at the same pH value.