Model Predictive Control for Collision Avoidance of Networked Vehicles Using Lagrangian Relaxation*

:This paper focuses on designing a steering intervention system based on Model Predictive Control. The goal is to avoid collisions between networked vehicles. Since the resulting program is non-convex, it is converted to a convex Semi-Definite Program by using Lagrangian relaxation. Furthermore, the utilized prediction vehicle model is linearized in an exact way to avoid errors due to linear approximation. Initially, the whole problem is formulated. Finally, the performance of the controller is presented in Model-in-the-Loop simulations.