On the morning of May 13th, the research group held the eighth session of its series of invited academic lectures in the Nashspace Conference Room of Tsinghua University. This activity was hosted by Li Nan, and Associate Professor Pingbo Tang from Carnegie Mellon University was invited to give a lecture titled Human-Cyber-Physical Systems for Resilient Civil Infrastructure Operations in a Changing World. In addition to the teachers and students of this research group, more than ten teachers and students from Tsinghua University, Southeast University, Tongji University, Hong Kong University of Science and Technology, Shenzhen Urban Public Safety and Technology Institute and other universities and scientific research institutions participated in this lecture online.
Dr. Tang’s research focuses on various human factor reliability problems in Human-Cyber-Physical Systems. In high intensity and low fault tolerance scenarios, the reliability of Human-Cyber interactions cannot fully meet the requirements. Therefore, for the recognition of unsafe scenes in construction, Dr. Tang proposed the principle of training set generation based on crowd evaluation and screening according to the Bayesian Network Model, which takes into consideration the difference between individual cognitive levels. Compared with the common majority vote algorithm, the performance of scene recognition is significantly improved.
Picture 1 Presentation of Report Content (I)
Accurate and comprehensive bridge inspection records can be used as a benchmark to support bridge operation and maintenance. However, in this inspection process, insufficient reliability of Human-Physical Interaction may lead to the miss of critical defects. Thus, Dr. Tang proposed to use personal characteristics analysis to predict and improve the task performance of bridge inspectors. In the study, the cognitive behavior of the test subjects in identifying bridge defects was captured, the personal behavior database was established, and artificial intelligence algorithms were trained to predict cognitive load and personal behavior fluctuations based on the complexity of the task. Finally, Dr. Tang’s research can assist bridge inspectors to improve the inspection success rate by giving corresponding feedback adjustments and reasonable task assignments.
Dr. Tang also introduced the frontier work direction of his team: Considering the reliability of Human-Human Interaction, integrating bridge inspection reports and finite element models of different ages, so as to reduce the influence of subjective defects in bridge inspection reports and generate reliable bridge state assessment. The ultimate goal of future research, he said, is to realize human-machine cooperation and let machine help people make decisions in real time.
Picture 2 Presentation of Report Content (II)
In the second part of the report, Dr. Tang shared his more than ten years' research experience with all the students present. Tang said that in the process of scientific research, it is important to cultivate one's independent research ability, start from practical engineering problems and make reasonable short-term and long-term research plans. Dr. Tang also gave specific and detailed suggestions on the presentation of reports and the writing of academic papers. The participants benefited a lot from the report and the friendly exchange and discussion with Dr. Tang inspired the students for their own research. The atmosphere of the whole exchange session was lively, and the report lasted about two and a half hours.
Picture 3 Teacher Pingbo Tang, Nan Li and students during the online lecture
