Z. Morley Mao receives NSF funding to build safe, resilient autonomous vehicles

The four-year grant will support research into improving the safety and reliability of tele-operated vehicles.
Prof. Z. Morley Mao
Z. Morley Mao

Z. Morley Mao, professor of Computer Science and Engineering at University of Michigan, and her collaborators at the University of Minnesota have received a four-year, $2.825-million grant from the National Science Foundation to support their project titled “Integrated Networking, Edge System and AI Support for Resilient and Safety-Critical Tele-Operations of Autonomous Vehicles.” Their work will support the next generation of safe, resilient autonomous vehicles (AVs) by exploring and enhancing tele-operations technology.

“There have been huge strides in the development of self-driving vehicles in recent years,” said Mao, “but we need to be sure that we are not sacrificing safety and reliability in the name of progress.”

Autonomous vehicles are becoming a reality, with self-driving and driver assist technology widely available in personal cars, and with robotaxi services on trial in several U.S. cities. While such advances are exciting and bode well for the future of sustainable transportation, these technologies are far from perfect.

As recent headlines remind us, errors in autonomous vehicles and the artificial intelligence (AI) tools that power them, from their tendency to miss or misinterpret stimuli in the environment to their vulnerability to cyberthreats, can be extremely dangerous and even fatal.

Mao and her team propose to address this issue by exploring tele-operations as an alternative to fully autonomous self-driving vehicles, an arrangement in which a human operator controls the vehicle remotely, either completely or partially as needed.

“The primary benefit of tele-operations in this context,” said Mao, “is that they enable more human control over systems that still need improving, ensuring greater safety as these technologies continue to develop.”

While AV tele-operations offer a promising solution, they remain largely aspirational and require further advances in 5G and next-generation networks to become feasible.

With these challenges in mind, Mao and her collaborators are planning to explore and develop a plan for the incremental adoption of tele-operations in AVs alongside the emergence of next-generation networking technologies.

Their project involves developing an interdisciplinary and comprehensive set of networking frameworks and systems architecture optimized for the types of AI programs commonly deployed in AVs. The team also plans to equip these systems with built-in mechanisms to protect against inaccurate or false AI predictions, offsetting the risk of dangerous mistakes. The next stage of their project will involve integrating human involvement into these systems to ensure seamless and safe remote operation.

These innovations will then be incorporated into a prototype AV tele-operation platform and tested in a variety of environments and commercial applications, including smart manufacturing and precision agriculture. Safety and resiliency will be at the center of these experiments, ensuring that appropriate checks are in place to ensure that remotely operated AVs are as error-free as possible.

“With this grant, we hope to bring tele-operated AVs from aspiration to reality,” said Mao, “with a particular focus on safety.”