Department of Electronic & Computer Engineering & Cheng Kar-Shun Robotics Institute Joint Seminar   - Distributed Optimization Based, Platoon Centered Car Following Control of a Connected and Autonomous Vehicle (CAV) Platoon  

CAV platooning technologies have received considerable attention in the past few years, driven by next generation smart transportation systems. This talk presents our recent results on fully distributed optimization based car following control via a platoon centered MPC approach for a CAV platoon. Both linear and nonlinear vehicle dynamics are considered. Under the linear vehicle dynamics, the underlying MPC optimization problem is a convex problem with a densely coupled central objective function and locally coupled constraints. This problem is converted to a locally coupled convex optimization problem via a decomposition method for the objective function. Operator splitting techniques are exploited to develop fully distributed algorithms, and control design and stability analysis is performed. Under the nonlinear vehicle dynamics, the underlying MPC optimization problem is nonconvex and densely coupled. We convert it to a locally coupled, albeit nonconvex, problem, for which a sequential convex programming based fully distributed algorithm is developed. The closed loop dynamics under the nonlinear vehicle dynamics is a time-varying nonlinear system subject to non-vanishing external disturbances. Further the right-hand side of this nonlinear system has no closed form expression when the MPC horizon is greater than one. To analyze the closed loop stability, we apply various tools from global implicit function theorems, stability theory of linear time-varying systems, and Lyapunov theory to establish local input-to-state stability uniform in all small coefficients pertaining to nonlinear dynamics.  Numerical tests of different types of CAV platoons in multiple traffic scenarios demonstrate the effectiveness of the proposed distributed optimization algorithms and control schemes.