Modelling and cooperative control of vehicle interactions in urban expressway weaving segments via dynamic grouping and potential field superposition

Urban expressway weaving segments involve intensive vehicle interactions and strongly coupled risks, posing challenges to modelling and control in mixed traffic. This study develops a unified framework integrating trajectory planning and control through dynamic vehicle grouping and artificial potential field superposition (APFS). A data-driven grouping strategy calibrated with UAV trajectory data captures heterogeneous and context-dependent vehicle interactions. A probabilistic potential field based on a two-dimensional joint density characterises risk distributions shaped by driving behaviour and infrastructure constraints, while double-integral-based superposition captures nonlinear multi-vehicle coupling effects. Integrated with model predictive control (MPC), APFS establishes a coherent planning–control paradigm. Numerical experiments show that APFS–MPC reduces travel time by 11.6% on average, increases minimum TTC by 12.9%, and decreases maximum lateral acceleration by 45.2% compared with APF–MPC and NSP–APF–MPC, improving safety, traffic efficiency, and stability.