A Defensive Motion Framework for Autonomous Platoon to Handle Cut-in Maneuvers

Autonomous platooning has emerged as a prominent trend in intelligent transportation systems. However, in hybrid traffic situations where autonomous vehicles (AVs) coexist with human-driven vehicles (HDVs), the aggressive cut-in behavior of HDVs poses a significant threat to the safety, stability, and integrity of autonomous platoons. To address this challenge, this paper proposes a cut-in defensive framework for autonomous platoons, consisting of two main components. Firstly, an LSTM network combined with the stochastic forward reachable set with a certain acceptance level predicts the occupancy area of a cut-in vehicle when its intention is detected. Secondly, the cut-in defense scheme for the platoon system is developed, where a cut-in risk index for platoon gaps (CRIG) is introduced to determine the cut-in point, allocating distinct driving modes to each vehicle in the platoon. After that, a high-level cut-in risk defense planner, integrated with a low-level motion controller, governs the single vehicle in the platoon. The high-level controller employs interactive Model Predictive Control (MPC) to suppress oscillation while ensuring cut-in safety, while the low-level controller utilizes an artificial potential field (APF)-based MPC approach for vehicle control. Simulation experiments evaluate the effectiveness of the proposed framework by introducing cut-in behaviors from different parts of the platoon compared to a baseline method.