Canonical Instabilities of Autonomous Vehicle Systems

The asymptotic limit theorems of control and information theories make it possible to explore the dynamics of collapse likely to afflict large-scale systems of autonomous ground vehicles that communicate with each other and with an embedding intelligent roadway. Any vehicle/road system is inherently unstable in the control theory sense as a consequence of the basic irregularities of the traffic stream, the road network, and their interactions, placing it in the realm of the Data Rate Theorem that mandates a minimum necessary rate of control information for stability. It appears that large-scale V2V/V2I systems will experience correspondingly large-scale failures analogous to the vast, propagating fronts of power network blackouts, and possibly less benign but more subtle patterns of individual vehicle, platoon, and mesoscale dysfunction. The central matter is the synergism between poorly-understood traffic flow dynamics and similarly cryptic multisource information network dynamics, leading to highly punctuated phase transition analogs.