SmartPL: An integrated approach for platoons driving on mixed-traffic freeways
Date:
Li, H., You, L.*, Xie, J., Ye, Y., & Tan, X. (2024, December 2-6). SmartPL: An integrated approach for platoons driving on mixed-traffic freeways [Poster Presentation]. The 31st International Conference on Neural Information Processing (ICONIP 2024), Auckland, New Zealand. (CCF C, EI)
Abstract: Vehicle platooning is a critical technology for transportation efficiency and energy consumption reduction. Even though conventional solutions to generate simultaneous lane-changing strategies have been widely studied, when managing platoons on freeways with mixed traffic, it becomes challenging to support sequential lane-changing maneuvers, which can perform efficient lane-changing even in high traffic density. To achieve that, this paper proposes an integrated approach called SmartPL, which can: 1) maintain close and efficient driving behaviors with stable platoon inter-vehicle distance during lane changs through a hybrid control model that separately manages car-following and lane-changing behaviors; 2) produce foresighted strategies by roadside unit (RSU) deploying a dedicated neural network, which comprises a traffic information encoder, a spatial relation extractor, and a strategy maker; and 3) facilitate sequential and safe lane-changing maneuvers by implementing a safety monitor on each platoon member. Furthermore, the efficiency and effectiveness of SmartPL are evaluated in a mixed-traffic freeway simulation. SmartPL achieves superior performance in accelerating and safeguarding lane-changing maneuvers, with average improvements of 12.54%, 14.66%, and 22.94% over three state-of-the-art methods, respectively. Code is available at https://github.com/IntelligentSystemsLab/SmartPL.