Enhancing Connected Autonomous Vehicle Formations: Discrete–Offline–Online Three-Layer Architecture for Platoon Reconfiguration

The formation transformation in intelligent connected autonomous vehicles (CAVs) enhances platoon versatility and significantly improves traffic efficiency. Current formation control strategies for CAV platoons often focus on fixed formation scenarios. This paper proposes a three-layer architecture for platoon reconfiguration, encompassing discrete, offline, and online layers. CAV platoons utilize this architecture to transform their existing formation into a specified target formation from the Intelligent Transportation System (ITS). In the discrete layer, we propose a formation representation scheme and design A* and cooperative sorting algorithms to achieve the optimal intermediate formation sequence. Moving to the offline layer, we design a Signal Temporal Logic-based model predictive control algorithm (MPC). This algorithm plans continuous, dynamically feasible, and collision-free safe trajectories, which are stored in an offline trajectory database. In the online layer, we design a successive linearization-based MPC to track the offline trajectories in real-time traffic environments and accomplish the platoon reconfiguration task. We implement single-lane and multi-lane platoon reconfiguration tasks in the MATLAB platform, comparing them with two advanced platoon reconfiguration algorithms. The experimental results, demonstrating the effectiveness of the proposed approach, are presented and discussed.