Morning commutes are stressful enough without having to deal with autonomous vehicles tailgating you or drifting into your lane.
Systems such as Lane Keeping (LK), which directly corrects a vehicle’s direction to prevent it from drifting into a different lane, and Adaptive Cruise Control (ACC), which adjusts the speed of the vehicle to ensure a safe driving distance from a slower vehicle in front, exist today and iterations of these features have existed for quite some time. Both features are essential for fully autonomous vehicles to truly become a reality. However, when both are activated simultaneously, there’s no guarantee that they will work correctly – until now.
Jessy Grizzle, the Elmer G. Gilbert Distinguished University Professor and Jerry W. and Carol L. Levin Professor of Engineering and Director of Michigan Robotics, and a team of scientists from several universities have created a solution to ensure that controllers associated with LK and ACC behave in a formally correct way when both are activated. Their research follows the work of Prof. Necmiye Ozay and her team, who created the first-ever solution to this problem. The project was part of a NSF Frontier Grant, a large four-million-dollar multi-university projects.
“We were taking on these challenge problems that were much more realistic than what the other groups were doing, and we had this large grant to allow collaboration with many levels of expertise,” Grizzle says. “It was amazingly cool.”
Their research is published as “Correctness Guarantees for the Composition of Lane Keeping and Adaptive Cruise Control.” The paper won the IEEE Transactions on Automation Science and Engineering (T-ASE) Googol Best New Application Paper Award for 2019 awarded by the IEEE Robotics and Automation Society. The award recognizes the Best New Application Paper of T-ASE published in the previous calendar year. The winner is decided based on the significance of new applications, technical merit, originality, potential impact on the field, and clarity of presentation of the research.
The researchers present a modular correct-by-construction control approach with correctness guarantees for the simultaneous operation of LK and ACC, where the longitudinal force and steering angle are generated by solving quadratic programs. The team enforces safety constraints by confining the states of the vehicle within determined controlled-invariant sets.
The methods presented can be applied to other safety control problems. The team tested their algorithms on the Khepera robot (a miniature programmable mobile robot) and the Robotarium testbed (a remotely accessible swarm robotics research testbed). The team had written a preceding paper dealing with the fundamentals of the issue, which was specifically applied to walking robots.
“It was to help robots have guarantees so that they wouldn’t fall as often,” Grizzle says, “but the methods still have a long way to go.”
The team hopes to test their algorithms on a full-sized vehicle. The paper’s lead author is Xiangru Xu, a former postdoctoral researcher working with Grizzle who is now an assistant professor at the University of Wisconsin-Madison.
Ames, Aaron D., Xiangru Xu, Jessy W. Grizzle, and Paulo Tabuada. “Control barrier function based quadratic programs for safety-critical systems.” IEEE Transactions on Automatic Control 62, no. 8 (2016): 3861-3876.
S. W. Smith, P. Nilsson, N. Ozay, “Interdependence quantification for compositional control synthesis with an application in vehicle safety systems”, Proc. IEEE Conf. Decision Control, pp. 5700-5707, Dec. 2016.