Control Using Local Distance Measurements Cannot Prevent Incoherence in Platoons
Carolina Lidström, Anders Rantzer and I have posted a manuscript on the topic of incoherence phenomena in vehicle platoons.
Fig 1: Illustration of Incoherence phenomenon. The figure consists of 100 lines, each of which shows how the position of a single vehicle in a platoon evolves over time. A macroscopic behaviour in which the overall length of the platoon appears to oscillate emerges.
The use of automatic control has the potential to make road use significantly more efficient. This has been demonstrated in practice in, for example, the platooning of convoys of trucks . However, when local control strategies are used to arrange a platoon into a string like formation, there is mounting evidence that a poorly regulated accordion like motion, such as that in Fig 1, will emerge.
Questions as to the fundamental nature of such macroscopic phenomena are then of paramount importance when considering the control architectures that should be employed in these applications. For example it has been observed that if measurements of the absolute positions of the vehicles are available, controllers that eliminate undesirable macroscopic behaviours can be designed. However this clearly comes at a cost, both financially and in the complexity of the controller. The question is then, is this really necessary?
Fig 2: A Platoon of n vehicles
In this paper we prove that the emergence of incoherent behaviours in platoons are an inevitable consequence of using local distance measurements to design the control. More specifically we demonstrate that no controller, irrespective of its dynamical complexity, sparsity, or linearity, can prevent the appearance of macroscopic behaviours in the platoon if only noisy measurements of the distances between neighbouring vehicles are available.