Trailblazing AI research

With the largest global share of research papers in the field of AI, China is embracing all things ‘smart’: innovations from ‘smart manufacturing’ to ‘smart cities’ which are fuelling the fastest digital economy growth in the world.

Just as impressive is the three-year growth of Shanghai Research Institute for Intelligent Autonomous Systems (SRIIAS). Apart from contributing to national infrastructure projects, SRIIAS’s AI focus has advanced the understanding of complex systems which are intrinsically difficult to model.

“AI embodies the collective wisdom of humanity to better serve our needs,” says Jie Chen, the university president who launched SRIIAS in 2018.

Speaking as the editor-in-chief of Autonomous Intelligent Systems, the open-source journal co-launched by Tongji University and Springer Nature in 2021, Chen highlighted the SRIIAS’s founding goal to “promote and disseminate scientific discovery, knowledge, technology and applications in the fields related to autonomous intelligent systems”.

These systems are designed to become increasingly integrated with the latest technologies to make uncertainties manageable. Driverless transportation and unmanned aerial vehicles (UAV) are obvious applications for such systems. As Chen explained, innovations by SRIIAS also target the management of the COVID-19 pandemic and digital infrastructure to accelerate health and urban transformation, as well as port and transport automation to shape global logistics.

Reaching far, deep and fast

In public health emergencies, such as pandemics, quick data access to travel history, time spent in risky areas, and relationships with potential carriers have proven crucial at checkpoints. But, performing these checks manually is time-consuming, potentially leaving crowds of people mustered for long periods.

Efficient and autonomous decision-making based on multi-source big data, the Novel Coronavirus Pneumonia-Automatic Identification System (Tongji NCP-AIS) developed by SRIIAS, was implemented at Tongji University gates in the early 2020s to facilitate pandemic tracking on campus and in the neighbourhood.

“While Tongji NCP-AIS reduces the workload of our administrators and promotes safety, we foresee its further potential to reduce transmission risks with minimal human intervention required,” Jie Chen says.

Adapting UAVs to replace human labour in uncharted locations where visibility is low, and in the absence of GPS, can save lives. A great example of UAV benefit is in the manual inspection of hydropower tunnels, which is time-consuming, costly, and imprecise, while posing health and safety risks, such as falls, structural collapse, hazardous gas leaks, or explosions.

The AI-empowered UAV designed by the SRIIAS team, deployed with sensors, can be used to map out trajectories and capture in real time a detailed 3D map of its surroundings. The team’s designs provide accurate measurements for the navigation of environments which are too confined for human access, or impossible to walk, with up to a 700-metre drop, for example.

This potential for SLAM (Simultaneous Localization and Mapping), taken further with a swarm of drones, also brought success in surveying topological details of million-hectare forests through autonomous, collaborative mapping and navigation. Commercialization initiatives for SRIIAS’s UAVs have so far yielded more than 100 million Chinese yuan, incubated new start-ups, and expanded the scope of application potentially to forestry, thermal power and other large-scale infrastructure inspection.

Another research priority for Chen is smart cities, especially the use of digital twins which can virtually mirror the ebbs and flows of city life, such as traffic patterns, in real time. Chen is confident that SRIIAS is positioned to redefine urban communities and their surroundings, and to enable better decision-making for sustainable city design and management. Recently, the SRIIAS team collaborated with Shanghai Municipal People's Government to develop a guide for the digital transformation, planning and construction for new towns. This will be a blueprint for intelligent urban construction of five new towns in Shanghai in the next five years.

The SRIIAS team was also one of the first in China to independently design and build connected and automated vehicles (CAVs), featuring simulation technologies which promise greater precision in simulating route options and improving channel assignment for inter-vehicle communications.

Details of vehicle design are just as important for the SRIIAS team, such as nonlinear feedback control for trajectory tracking, and integrated energy control systems, to enhance energy efficiency and safety. They have enabled designs of the L4 robotaxi with 9,000 km test miles. According to the Society of Automotive Engineers, these self-driving taxis do not require human input in most circumstances, but still allow the option for manual override.

Safer and more efficient freight terminals are also possible through work by the SRIIAS team, which, together with the Hong Kong University of Science and Technology, has investigated the use of automated guided vehicles (AGVs). Apart from high-precision positioning in environments without GPS, these AGVs are designed with light detection and ranging (LiDAR) and visual sensors, as well as SLAM navigation and deep neural network, to flexibly deployed at unmanned ports.

Ambitious blueprints

At SRIIAS, world-leading AI research infrastructure is also being developed, including next-generation computing and algorithms, autonomous aviation and marine systems, as well as other specializations.

Its academic network is also growing in size and reputation. In 2021, five SRIIAS professors were elected to become members of two national academies in China – the Chinese Academy of Sciences and the CAE. Globally, SRIIAS has deepened connections with the University of Edinburgh, the University of California, Berkeley, Max Planck Institute, acatech (the German Academy of Science and Engineering), and many other leading academic institutions to foster international exchange.

Drawing on his research experience as a member of the CAE and a fellow of the Institute of Electrical and Electronics Engineers, Chen emphasizes cross-disciplinary work to cover three key issues in AI research: autonomous and sensing systems, emergent intelligence, as well as collaborative and swarm intelligence. His teams are working across nine research areas: metamaterials and sensors, multi-scale integration, nature-inspired computing, autonomous agents, system integration (biological, mechanical, and electronic systems), autonomy and interaction, heterogeneous systems and decision-making, multi-agent collaboration, as well as brain-like intelligence and bionics.

Since 2020, SRIIAS has been leading a municipal project in which universities across Shanghai collaborate on AI theory and practice.

“We have witnessed tremendous growth in our first three years, and SRIIAS will continue to play a pivotal role to focus on fundamental AI studies and train talent to accelerate innovations in service and manufacturing industries,” says Chen. “SRIIAS sets an example of cross-disciplinary excellence not just for Tongji University, but also demonstrates how AI can empower positive changes in other fields, as well as the digital transformation of China and beyond.”