Event Location: NSH 1507
Bio: Serban Sabau received the M.S. in control engineering from “Politehnica” Bucharest in
2002 and the Ph.D. in electrical engineering from the University of Maryland in 2011. Before joining Stevens
in 2013 as anAssistant Professor he was a postdoctoral fellow at the University of Pennsylvania. His interests
are in numerical algorithms for distributed optimization and distributed control and in fault tolerant control
methods for driverless cars.

Abstract: On highways, all vehicles (automated or not) must travel in the string (or platoon) formation.
Surprisingly enough, this has been an open problem in controls for more than fifty years, dating back to the
work of Levine and Athans on intelligent vehicles highway systems. The reason for which automated
platooning is not a reality today is that no existing solution was deemed completely satisfactory from several
engineering standpoints. We present a novel distributed control design that vastly outperforms existing
methods with respect to the relevant quantitative metrics of safety and performance. Our method is also the
only one (we are aware of) capable to tackle systematically with several key practical necessities: it can
accommodate heterogeneous platoons by compensating the different time constants associated with the
dynamics of various types of drive trains (and hydraulic actuators), it can completely compensate the vehicle
to vehicle communications time delays and time-jittering, it retains optimality during merging/ exiting
maneuvers, requiring only “local” reconfigurations of the control scheme. As an extra bonus, the scheme is
capable of detecting in real time (while functioning on the highway) any “abnormal” behavior caused by
sensors faults or sensors spoofing and by wireless communications disruptions or hacking. Our method is
based on a novel type of structural constraints in distributed control, introduced by the speaker. We disprove
longstanding conjectures that claimed that vehicle-to-vehicle communications with more than one preceding
vehicle (beyond the direct line of sight) may be beneficial for platooning, since it provides a “preview” of
disturbances. We also show how our methods can be exploited for more general, two-dimensional
formations, appearing in flight control of aerial vehicles and other practical applications.