The system’s computational and physical components are designed together from the start, and will result in safer systems that are faster and less expensive to design.

By Bill Kisliuk and Matthew Chin

UCLA electrical engineering professor Paulo Tabuada will lead the development of technologies to improve safety and reliability of cars moving at high speeds. The research will have broad applications to other systems that operate and come across unsafe conditions.

“Now cars can automatically brake when their systems detect a hazard, such as a person unexpectedly darting out in the road,” Tabuada said. “The technology we will develop in this project will take the existing collision-avoidance systems to the next level by having cars automatically plan and execute collision avoidance trajectories when braking is not an option.”

These types of technologies, where both the computational and physical components are completely integrated into one engineered system are known as cyber-physical systems (CPS).
The project is part of a larger four-year, $4 million study on the frontiers of cyber-physical systems funded by the National Science Foundation, and based at the University of Michigan, with research teams from Texas A&M, and Carnegie Mellon University, as well as UCLA.

Specifically, Tabuada will develop methods to synthesize embedded control software based on motion primitives, low-level control designs, and virtual low-dimensional models crafted by researchers from the project’s partner universities.

Tabuada added that the project’s “correct-by-design” or “correct-by-construction” process, which emphasizes synthesizing software to meet sophisticated specifications from the start, rather than engaging in numerous rounds of reiterative testing, will result in safer cyber-physical systems that are faster and less expensive to design. He said ensuring the technology will work in potentially life-threatening situations is a significant challenge.

“I am charged with using physical models and low-level control designs to go from high-level requirements to software implementation,” he said. “Because this technology takes over for human inputs, it has to always work, no matter what.”

Tabuada leads UCLA’s Cyber-Physical Systems Laboratory and is a 2005 recipient of the National Science Foundation’s CAREER Award and a 2009 recipient of the American Automatic Control Council’s Donald P. Eckman Award.

Main Image: A high-fidelity, industry-level, vehicle simulator is used to show the response of the car in real scenarios.