Shangzhi Xu's research while affiliated with Tongji University and other places

... Interesting results are presented in the study by Yang [21], which examines the correlation between driving style, fuel consumption, and overall emissions. The authors conducted extensive research and practical measurements in real traffic conditions. ...

... In 1998, Nagatani [1] proposed the famous one-dimensional lattice model for macroscopic traffic flow on single-lane roads. In order to accurately reproduce and simulate the nonlinear phenomena of complex traffic systems at a higher level, some scholars have extended the Nagatani model [1] by incorporating factors such as backward looking effect [2] , driver anticipation effect [3] , interruption effect [4] , passing effect [5,6] , driver delay effect [7,8] ,multiple density difference effect [9] ,flux limit effect [10] and self-stabilizing effect [11]. This has led to a series of single-lane characterization models that are closer to real traffic conditions. ...

... In the past few years, deep learning has been used for the enhancement of nighttime pedestrian detection. The Region-based Fully Convolutional network (R-FCN) [2], Fast R-CNN [13], Mask R-CNN [15,16], Residual Neural Network (ResNet), and [17], LeNet [18], were state-of-the-art frameworks that have been used for the detection of nighttime pedestrian detection. LeCun et al. [9] suggested the LeNet architecture for the detection of nighttime pedestrian detection. ...

... It is then forwarded to the SIP server. At this point, if we check the terminal state in the SCE database again, we can find that it has changed from Offline to Online, which proves that the dynamic innovation of terminal state has been successful [9,10]. ...

... The model was then converted from the microscopic to the macroscopic system [32][33][34] using the following transformations: ...

... The multiplex phenomena of the traffic flow field have been investigated following two main approaches: (1) the Eulerian approach, revealing the whole scenario of the traffic flow field based on the laminar fluid flow, and (2) the Lagrangian approach, focusing on each individual vehicle and including an offshoot widely used for simulation-based research that is termed the cellular automata (CA) traffic model. Meanwhile, to investigate the dynamical behavior of the traffic flow field, various types of traffic flow models have been proposed, including continuum models [1][2][3][4][5][6], microscopic models [7][8][9][10][11][12], lattice hydrodynamic models [13][14][15][16][17], and various CA models [18][19][20][21][22][23] while considering various effects, such as the system time delay effect, the horn effect, the heterogeneous vehicle's effect, and the CA-human-driven mixed-flow effect. Although an array of traffic models have already been devised, traffic flow analysis is a fundamental aspect of studying the complex behavior of nonlinear sciences, and the majority of the previous studies are largely conventional and straightforward, except for several admirable works [24][25][26][27][28][29][30][31]. ...

... This section introduces the lane-level vehicular movement model from three aspects, car-following models, lane-changing models, and vehicular trajectory models. Table 5 [162][163][164][165][185][186][187][188][189][190][191][192][193][194][195][196][197] summaries some applications for communication and transportation, which is also vital to autonomous driving technology. ...

... The authors also applied the HDM to IDM, and devised an anticipative and smooth braking strategy. Other relevant reference can be referred to the literature [34] . ...

... In vehicle-following models such as the IDM [13], agents interact with the vehicle in front and adjust their speed so that the time gap between them remains at the desired value. These kinds of models not only work for highway but can also be adapted for urban traffic [47]. In cellular automata models, both space and time are discretised. ...

... The traffic flow models developed so far, whether in microscopic or macroscopic approaches, have been concerned with the synchronizing state because controlling this metastable state enhances the efficiency of the traffic flow field. There are several types of traffic models that have been developed following microscopic 2-11 , macroscopic 12-20 , lattice hydrodynamic [21][22][23][24][25][26][27] , gas kinetic [28][29][30][31] , and cellular automated [32][33][34][35][36][37][38][39] approaches. The models [40][41][42] , in which the eye-tracking device's effect has been introduced for investigating driver's attention to study an actual traffic flow, revealed a new dimension of traffic flow modeling. ...