Zhao, Huijing is a tenured associate professor in the Department of Machine Intelligence, School of EECS. She obtained her B.Sc. from Peking University in 1991, and Ph.D. from the University of Tokyo in 2004 respectively. She has research interest in several areas in connection with intelligent vehicle and mobile robot, such as machine perception, behavior learning and motion planning, and she has special interests on the studies through real world data collection.

Dr. Zhao has published more than 60 research papers in recent 10 years, and many of them are published in top-tier conferences and journals in her research area, such as T-ITS, ICRA, IROS, IV and ITSC. She serves as an associate editor of the T-IV since 2016, and also of the conferences ICRA12, IV17, IROS17. She has served as a key member of the Rule and Referee Committee of "Intelligent Vehicle Future Challenge in China” 2010-2014. She received the Most Influential Paper over the Decade Award from IAPR Conference on Machine Vision and Application (2009), received the Okawa Fundation Research Grant (2011), and supervised the team "FalconView" that won the first prize (100,000 EUR) of the Valeo Innovation Challenge 2015.

As the Principal Investigator (PI) and Co-PI, Dr. Zhao has been funded in three NSFC funded projects, including two general projects "On-Road Vehicles' Behavior Analysis and Contextual Understanding" (No.61573027, 2017-2019) and “Towards 3D SLAM in an Outdoor Dynamic Urban Environment”(No.60975061, 2007-2009), and a Sino-French project "PRETIV: Multimodal Perception and Reasoning for Transnational Intelligent Vehicles" (No.61161130528, 2012-2014) co-funded by NSFC and ANR. She has also been a sub-leader of a NSFC key project “Research on the Evaluation System and Integrated Testing Environment for the Intelligent Behavior of Unmanned Ground Vehicles” (No: 90920304, 2010-2013). She has been the chair of the PSA-PKU Openlab on “Multimodal Perception and Reasoning for Intelligent Vehicles” since 2012, and has been the PI of DMAPR project of the Openlab that is now on the 2^nd stage (v1.0 2012-2015; v2.0 2016-2019). Her research achievements are summarized as follows:

1) On-road trajectory data collection for naturalistic driving behavior study: Modeling and reasoning the driving behaviors in real-world traffic is crucial for the applications such as autonomous driving, advanced driving assistance systems (ADAS) etc. One of the key issues is real-world data collection. Since 2013, based on our algorithms and systems, we have been collecting the multi-modal data that monitor simultaneously the behavior of driver, ego and environmental vehicles by driving our instrumented vehicle on the public motorways in Beijing for more than 7000km by 10 different drivers. A major feature of this activity is that on-road vehicle trajectories are collected, which is crucial for tactical behavioral studies. Some data (sole-authored by PKU) has been web published, which, to the best of our knowledge, is the first dataset containing large sets of vehicle trajectories collected on-road for driving behavior studies.

2) Driving behavior modeling and reasoning using real-world driving data: A series of studies have been conducted at both semantic and behavioral levels. Car followings and lane changes are addressed, and the behaviors are studied at both the tactical and control levels. Except that the studies are conducted on real-world data, another major feature is that the driver/ego’s behavior is analyzed within the context of environmental vehicles to reveal their interactive relations. There have been studies in literature by analyzing the behaviors at certain small highway segments (e.g. 1km) due to the difficulties in on-road vehicle trajectory collection at larger area. This series of studies are featured, and the results are important to improve the intelligence of autonomous driving and advanced driving assistance systems at real-world traffic.

3) Semantic and dynamic mapping of outdoor environments: Major challenges preventing now-a-day’s mobile robots/autonomous driving systems are that the environments are complex and changing. Accurate maps with large richness of detail have been the core to reduce the difficulties in online perception and reasoning, while developing these maps containing semantic meaning and addressing scene dynamics are still open topics. A number of studies have been conducted in this scope, e.g. SLAM with MODT (Simultaneous Localization and Mapping with Moving Object Detection and Tracking); scene understanding in a large dynamic environment; scene dynamics analysis etc. While, how can a robot learn a spatial and temporal map from its own experience of the environment? How can robots share their maps to have more consistent and global knowledge of the environment? Researches are conducted to pursue answers of these questions.