Multimodal sensing and active continuous closed‐loop feedback for achieving reliable manipulation in the outdoor physical world

The use of field robots can greatly decrease the amount of time, effort, and associated risk compared to if human workers were to carryout certain tasks such as disaster response. However, transportability and reliability remain two main issues for most current robot systems. To address the issue of transportability, we have developed a lightweight modularizable platform named AeroArm. To address the issue of reliability, we utilize a multimodal sensing approach, combining the use of multiple sensors and sensor types, and the use of different detection algorithms, as well as active continuous closed‐loop feedback to accurately estimate the state of the robot with respect to the environment. We used Challenge 2 of the 2017 Mohammed Bin Zayed International Robotics Competition as an example outdoor manipulation task, demonstrating the capabilities of our robot system and approach in achieving reliable performance in the fields, and ranked fifth place internationally in the competition.     

An energy-efficient cloud system with novel dynamic resource allocation methods

The Journal of Supercomputing - In the last decade, cloud computing has brought enormous changes to people’s lives. Cloud computing gives a client-driven computational model. In this case,...     

Collection efficiency of liquid‐based samplers for fungi in indoor air

This study assessed the collection efficiency (CE) of two popularly used sampling devices (BioSampler and Coriolis sampler) for fungal aerosols. Phosphate‐buffered saline (PBS) supplemented with or without surfactant (Tween‐20, Tween‐80, or Triton X‐100) and antifoam agent was prepared and used as collection liquids. The agar impactor (BioStage) was simultaneously operated with liquid‐based samplers to collect fungi from seven sites located at a university building, public library, and animal farming. Fungal concentrations determined by liquid samplers were divided by those by BioStage, and the ratio values represented CE. Results indicate that the CE of BioSampler was superior to that of Coriolis (P = 0.0001) and the PBS containing surfactant collected fungi better than that without surfactant (P < 0.0001), whereas antifoam agent showed no influence (P = 0.8). Moreover, fungal concentrations determined by BioSampler with surfactant‐added PBS were statistically indifferent from those by BioStage (P > 0.05) with a Spearman correlation coefficient of 0.81‐0.83 (P < 0.01). In addition to sampler and collection liquid, sampling location was also identified as a significant CE factor (P = 0.006), implying potential influences by fungal genera in the studied fields. Overall, BioSampler with surfactant‐supplemented PBS (eg, Triton X‐100) is recommended considering the great CE and compatibility with a variety of analytical assays.     

基于注意力机制与文本信息的用户关系抽取

随着社交媒体的发展,用户之间的关系网络对于社交媒体的分析有很大的帮助。因此,该文主要研究用户好友关系检测。以往的关于用户好友关系抽取的研究主要基于社交媒体上的结构化信息,比如其他好友关系,用户的不同属性等。但是,很多时候用户本身并没有大量的好友信息存在,同时也不一定有很多确定的属性。因此,我们希望基于用户发表的文本信息来对用户关系进行预测。不同于以往的潜在好友推荐算法,该文提出了一种基于注意力机制以及长短时记忆网络(long short-term memory,LSTM)的好友关系预测模型,将好友之间的评论分开处理,通过分析用户之间的评论来判断是否具备一定的好友关系。该模型将好友双方信息拼接后的结果作为输入,并将注意力机制应用于LSTM的输出。实验表明,用户之间的评论对于好友关系预测确实有较大的实际意义,该文提出的模型较之于多个基准系统的效果,取得了明显的提升。在不加入任何其它非文本特征的情况下,实验结果的准确率达到了77%。