Liu Zhanwen | Standing tall and shouldering the mission of intelligent transportation Fu Fanghua

As a female young scientist born in the 1980s who was born at the beginning of reform and opening up and has the honor to witness the development of many scientific and technological fields in my country from following, running side by side, to leading in some areas, Liu Zhanwen, a professor at the School of Information Engineering of Chang'an University, has a strong sense of collective honor and scientific research mission. She has been engaged in research in artificial intelligence, deep learning, computer vision, collaborative perception performance testing and other fields for a long time. In recent years, she has carried out work on full-factor perception of complex dynamic traffic environments, perception and testing of vehicle-road collaborative group intelligence fusion, and digital twin research of intelligent networked highways and urban intersections around autonomous driving and networked transportation. At present, vehicle-road collaboration has entered a new stage of development, and intelligent and networked technologies are accelerating their evolution. The traditional transportation theory system, technology system, and industrial system are undergoing drastic changes driven by big data technology. What lies before her are unprecedented opportunities and challenges, and even more so the responsibility of the times. Responding to the call, "taking the lead and playing the leading role", building a strong transportation country, and contributing to the progress of intelligent networked technology in the transportation field are the sincere voices of Liu Zhanwen and her team members in the new era and new wave.

▲Liu Zhanwen

Accumulating strength and then developing new navigation mark

After graduating from high school, Liu Zhanwen came to Xi'an to study and has lived in Xi'an for more than 20 years. She affectionately calls this ancient city her "second hometown". Xi'an's rich historical and cultural heritage has nurtured her and made her unconsciously calm down and polish herself. She said that it was Xi'an that taught her that only with sufficient accumulation can one achieve great success.

After graduating from Northwestern Polytechnical University in 2006, Liu Zhanwen entered Chang'an University, where she obtained a master's and doctoral degree, and stayed on to teach due to her outstanding performance. During her doctoral studies, she was good at using traditional machine learning methods to detect vehicle targets in low-contrast night traffic images. She was attracted by the powerful feature extraction capabilities of deep learning when she came into contact with it. After work, she has been keeping up with the forefront of artificial intelligence technology, maintaining a keen academic sense, and studying robust perception algorithms for various traffic targets in complex dynamic traffic scenes.

It was during this gradual accumulation that Liu Zhanwen strengthened her determination to continue to work in the field of intelligent transportation. Compared with solving specific engineering problems, she is more eager to tackle the scientific problems behind the core key technologies. In the winter of 2018, Liu Zhanwen went to the Institute of Advanced Transportation Technology at the University of California, Berkeley for a one-year visit and exchange. Her foreign co-supervisor was engaged in research related to the control of autonomous truck platoons, while she mainly focused on the research of the full-factor perception of the traffic environment and its uncertainty of the lead vehicle and the following vehicle.

▲Liu Zhanwen took a group photo with team members

During this period, Liu Zhanwen's visiting scholar life was filled with advanced scientific research concepts, academic reports from top scientists, and academic exchanges from multicultural backgrounds. She also participated deeply in the international cooperation projects of foreign mentors, and followed the team to the Saudi Arabia National Laboratory for phased project exchanges and results reports. "This year has become an important turning point in my scientific research career. I have learned how to build an excellent scientific research team and have become more aware that the essence of scientific research lies in the combination of theory and practice. In particular, scientific research should focus on practical applications and effects, and be able to abstract common scientific problems from practical application problems. This aspect is the direction I have been working hard on so far." Therefore, after returning to China at the end of 2019, facing the major needs of the strategy of building a strong transportation country, breaking through the core key technologies and scientific issues of vehicle-road collaborative perception, providing technical support for the pilot demonstration project of vehicle-road collaboration and autonomous driving, and promoting the implementation of scientific research results have become a new beacon in the scientific research journey of Liu Zhanwen and his team.

Working together to achieve new achievements

Liu Zhanwen is also the director of the Construction Office of the Shaanxi Provincial Internet of Vehicles and Intelligent Vehicle Testing Technology Engineering Research Center, and the director of the Traffic Information Detection and Control Engineering Laboratory of Chang'an University, and is mainly responsible for the platform construction of the laboratory. She innovatively fed back the scientific research results of the team's accumulated vehicle-road cooperative autonomous driving perception and testing to teaching, and created a first-class virtual simulation experiment course, so that students can grow simultaneously in theoretical learning and hands-on practice. The relevant scientific research and teaching achievements won three first prizes in the Shaanxi Provincial Science and Technology Progress Award and one special prize in the Shaanxi Higher Education Teaching Achievement Award in 2019. She guided students to participate in high-level subject competitions and won more than ten national awards. In the interview, Liu Zhanwen mentioned more than once the importance of "doing scientific research in a lofty manner", hoping to grow into a unique and innovative researcher who integrates knowledge and practice. Keeping up with the international frontier, exploring original innovative results, based on national strategic needs, and making tangible contributions to the development of intelligent networked transportation are the original intentions of her and her team.

One of the biggest challenges currently facing the field of vehicle-road collaboration and autonomous driving is the complexity of actual application scenarios. Researchers need to first solve the problem of traffic environment perception in various dynamic traffic scenarios. Therefore, facing the major needs in the field of vehicle-road collaboration and autonomous driving, Liu Zhanwen led the team to focus on the research direction of trusted perception and testing of vehicle-road collaboration in inconsistent scenarios, and conquered the core key technologies of multi-source heterogeneous data fusion for vehicle-road-cloud integrated information sharing, spatiotemporal feature registration between multi-modal and multi-scale data frames, real-time tracking of multiple targets in inconsistent scenarios, and virtual-real interaction testing and evaluation of complex vehicle-road systems, making useful attempts to solve the trusted perception and performance testing in the field of vehicle-road collaboration and autonomous driving. The relevant achievements have improved the safety of vehicle-road collaboration and autonomous driving, optimized the road traffic system, and further promoted the innovative development of my country's intelligent transportation digital economy.

Among them, the "Multimodal Perception Enhanced Target Trajectory Calculation and Scene Flow Digital Twin" project, funded by the National Natural Science Foundation of China in 2022, is committed to solving the current complex dynamic environment with strong coupling, strong randomness, and strong nonlinearity of multiple targets. Single intelligent perception is prone to information overload and perception failure, especially in severe occlusion, beyond visual range or blind spots, which leads to the inability to make correct decisions and precise control. Based on the two modal data of traffic environment video and radar point cloud collected by the roadside smart base station of the cooperative unit Wanjie Technology, Liu Zhanwen led the team to carry out systematic research on multimodal data fusion, target trajectory extraction, identification and prediction, and scene flow digital twin problems, solving three key problems: difficulty in multimodal frame-to-frame spatiotemporal feature registration, lack of traffic semantics in trajectory prediction, and simulated scene flow evolution deviating from the real dynamic target trajectory. The multimodal perception-enhanced target trajectory identification, trajectory prediction, scene flow estimation and digital twin they proposed in this project can describe the meso- and micro-traffic operation situation in a detailed, accurate and comprehensive manner, break through the bottleneck of the global perception technology of intelligent roadside equipment in the vehicle-road collaboration industry, and accelerate the testing and verification of key technologies of intelligent roadside perception in vehicle-road collaboration demonstration applications.

In March 2023, the results of the 14th Shaanxi Youth Science and Technology Award were announced, and Liu Zhanwen was successfully selected. When she learned the news, Liu Zhanwen was having a weekly meeting with graduate students. Looking at the young faces in front of her who were reporting on their research, her thoughts were flying and her emotions were mixed. "The past few years of the epidemic have been very difficult. My students and I have encouraged each other to tackle key projects and deal with complicated affairs, and we have witnessed each other's growth, transformation, and maturity. In this era of innovation and change, I know very well that this award is an honor we have won through concerted efforts with the support of a large platform and a large team."

The Internet of Vehicles and Intelligent Vehicle Testing Technology Research Institute where Liu Zhanwen works has more than ten key scientific research platforms and has built the only "autonomous driving closed field test base" recognized by the Ministry of Transport in domestic universities. It has formed an interdisciplinary innovative research team led by her mentor Professor Zhao Xiangmo, which consists of more than 50 young and middle-aged academic backbones and more than 300 master and doctoral students. In Liu Zhanwen's view, each of the above achievements is the result of everyone's joint efforts.

Now, after several years of accumulation, Liu Zhanwen has a clear academic development direction and a stable team. In response to the actual needs of the sustainability and input-output ratio of the vehicle-road collaborative demonstration project, relying on the demonstration project of a strong transportation country - the Qinling Tunnel Group Safety Prevention and Control System, and the national key R&D project - Smart City Vehicle-City Interaction and Demonstration, in the future, she and her team will be committed to creating full-time and full-domain integrated perception and digital twins for key sections, providing a reliable data foundation and decision-making support for the safety of smart transportation operations.