Bryan Reimer Presents on AI and Autonomous Cars for the Future Networked Car Symposium
by Adam Felts
AgeLab Research Scientist Bryan Reimer was the keynote speaker for the 2022 Future Networked Car Symposium, an event that brings together representatives of the automotive, information and communications technology industries, along with government leaders and regulators, to discuss the status and future of vehicle communications and automated driving from both technical and regulatory viewpoints.
Dr. Reimer presented on the potential for advances in artificial general intelligence to produce a new generation of automated vehicles. To begin with, he noted that roadway safety is a global, undertreated public health crisis, with over 1.25 million fatalities worldwide every year. The fully automated vehicle represents a “holy grail” to address these harms. Vehicle automation has been a technological dream for the better part of the last century, and in the last decade, AI has produced new hopes for the potential to create a truly automated car.
However, an MIT survey found that many drivers may not be interested in the dreamed-of self-driving car. Instead, more drivers want a car that helps them drive, rather than a car that drives for them. There is a gap between the desires of consumers and drivers and the interests and projects of today’s future-focused autonomous vehicle manufacturers.
What’s more, today’s AI remains a “brittle” technology that only will function reliably as part of a system when the other elements with which it interacts—infrastructure, driver, and other supporting technologies—are harmonized with it. Over the coming years, drivers or operators will continue to oversee the technology (on or off board) in order for vehicles to operate safely—and our infrastructure will need to be reimagined to accommodate the technology in order for it to function optimally.
More research is necessary to study the automated systems of today: those that involve the interplay between driver and technology. These systems tend to perform less well than designs predict, likely because drivers do not have the skills to operate such automated systems, because the systems themselves require a level of oversight that is unrealistic given what we know about human behavior, and because the infrastructure on which these systems operate is less than ideal.
Dr. Reimer advocates for a greater emphasis on studying how drivers use advanced driver assistance technologies on roads. This work is currently being done, in part, by the AgeLab’s Advanced Vehicle Technology Consortium (AVT). Additionally, he says, AI needs to be deployed in cars not only to manage the task of driving, but to monitor drivers themselves to ensure that they are engaging with these systems properly and safely.
Dr. Reimer’s address—and the presentations from the second day of the conference—can be found here.