In this recent preprint, my student Jonas Hansson and I continue to describe the concept of serial consensus, which is a closed-loop design that appears like two first-order consensus protocol applied in series, rather than a traditional second-order consensus protocol. This yields surprising and huge benefits in vehicular formations where interactions are limited to being relative (eg based on radar measurements) and directed (eg measuring distances w.r.t. the vehicle ahead). Well-known issues with string stability are solved and Jonas derives hard bounds on the transients.
Do check out the paper and let us know what you think!
Emma Tegling 