Fully Distributed Model Predictive Control of Connected Automated Vehicles in Intersections: Theory and Vehicle Experiments

We propose a fully distributed control system architecture, amenable to in-vehicle implementation, that aims safely coordinate connected and automated vehicles (CAVs) at road intersections. For purposes, we build upon model predictive approach, in which the agents solve nonconvex optimal problem (OCP) locally synchronously, exchange their optimized trajectories via vehicle-to-vehicle (V2V) communication. To accommodate fast solution of OCPs, apply penalty convex-concave procedure solves convexified version original OCP. experimental evaluation, complement controller with localization layer, being charge self-localization, an estimator, determines joint collision points other agents. Experimental tests reveal efficacy proposed architecture.