Adaptive generative adjustable electric fence method and internal obstacle detection

Yixiao Liu,Zihao Tian,Lixin Tian,Zhiyuan Liu

DOI: https://doi.org/10.1016/j.trc.2024.104601

IF: 9.022

2024-04-14

Transportation Research Part C Emerging Technologies

Abstract:It has become a general consensus to design virtual parking stations in dockless bike-sharing systems. In recent years, electric fence technology has been established as a "station" for dockless bike-sharing to regulate users' random bicycle parking behavior. However, existing research has rarely considered the problems encountered in real-world deployments of electric fences, such as the static capacity that may result in insufficient capacity. In addition, electric fences can reduce their capacity if they contain obstacles. In this paper, we design an adaptive generative adjustable electric fence system, which can generate electric fences based on constructing an adaptive learning loop of users parking bicycles with system learning feedback. Meanwhile, we define the grid area density to regulate the dynamic change of electric fence capacity. It can dynamically detect and regulate the effect of an obstacle on the capacity of electric fence generation based on the density of parking in the electric fence, regardless of whether the obstacle is permanent or temporary. The method is tested on both extreme human-designed terrain and realistic terrain, and the electric fence generated by the method achieves good adaptation and capacity regulation even when there are bicycles whose GPS positions are incorrectly located on obstacles.

transportation science & technology