1. |
- Han, Yong, et al.
(author)
-
Pedestrian trajectory prediction method based on the Social-LSTM model for vehicle collision
- 2024
-
In: Transportation Safety and Environment. - 2631-6765 .- 2631-4428. ; 6:3
-
Journal article (peer-reviewed)abstract
- Techniques for predicting the trajectory of vulnerable road users are important to the development of perception systems for autonomous vehicles to avoid accidents. The most effective trajectory prediction methods, such as Social-LSTM, are often used to predict pedestrian trajectories in normal passage scenarios. However, they can produce unsatisfactory prediction results and data redundancy, as well as difficulties in predicting trajectories using pixel-based coordinate systems in collision avoidance systems. There is also a lack of validations using real vehicle-to-pedestrian collisions. To address these issues, some insightful approaches to improve the trajectory prediction scheme of Social-LSTM were proposed, such methods included transforming pedestrian trajectory coordinates and converting image coordinates to world coordinates. The YOLOv5 detection model was introduced to reduce target loss and improve prediction accuracy. The DeepSORT algorithm was employed to reduce the number of target transformations in the tracking model. Image Perspective Transformation (IPT) and Direct Linear Transformation (DLT) theories were combined to transform the coordinates to world coordinates, identifying the collision location where the accident could occur. The performance of the proposed method was validated by training tests using MS COCO (Microsoft Common Objects in Context) and ETH/UCY datasets. The results showed that the target detection accuracy was more than 90% and the prediction loss tends to decrease with increasing training steps, with the final loss value less than 1%. The reliability and effectiveness of the improved method were demonstrated by benchmarking system performance to two video recordings of real pedestrian accidents with different lighting conditions.
|
|
2. |
- Jing, Xiao, et al.
(author)
-
Safety benefit of cooperative control for heterogeneous traffic on-ramp merging
- 2022
-
In: Transportation Safety and Environment. - : Oxford University Press (OUP). - 2631-6765 .- 2631-4428. ; 4:4
-
Journal article (peer-reviewed)abstract
- The safety of heterogeneous traffic is a vital topic in the oncoming era of autonomous vehicles (AVs). The cooperative vehicle infrastructure system (CVIS) is considered to improve heterogeneous traffic safety by connecting and controlling AVs cooperatively, and the connected AVs are so-called connected and automated vehicles (CAVs). However, the safety impact of cooperative control strategy on the heterogeneous traffic with CAVs and human-driving vehicles (HVs) has not been well investigated. In this paper, based on the traffic simulator SUMO, we designed a typical highway scenario of on-ramp merging and adopted a cooperative control method for CAVs. We then compared the safety performance for two different heterogeneous traffic systems, i.e. AV and HV, CAV and HV, respectively, to illustrate the safety benefits of the cooperative control strategy. We found that the safety performance of the CAV and HV traffic system does not always outperform that of AV and HV. With random departSpeed and higher arrival rate, the proposed cooperative control method would decrease the conflicts significantly whereas the penetration rate is over 80%. We further investigated the conflicts in terms of the leading and following vehicle types, and found that the risk of a AV/CAV followed by a HV is twice that of a HV followed by another HV. We also considered the safety effect of communication failure, and found that there is no significant impact until the packet loss probability is greater than 30%, while communication delay's impact on safety can be ignored according to our experiments.
|
|
3. |
- Zhang, Jie, et al.
(author)
-
An IDDES study of the near-wake flow topology of a simplified heavy vehicle
- 2022
-
In: Transportation Safety and Environment. - : Oxford University Press (OUP). - 2631-6765 .- 2631-4428. ; 4:2
-
Journal article (peer-reviewed)abstract
- The complex wake flow of a GTS (ground transportation system) model contributes to large percentage of the aerodynamic drag force. Therefore, predicting accurate wake flow will help carry out the drag reduction strategies. In this paper, the near-wake flow topology of the GTS was studied at Re = 2.7x10(4) to assess the capability of a hybrid RANS/LES (Reynolds-averaged Navier-Stokes/large eddy simulation) approach, known as IDDES (improved delayed detached eddy simulation). The current study also aims to understand the effects of different computational parameters, e.g. the spatial resolution, time step, residual level, discretization scheme and turbulence model, on this asymmetrical wake flow configuration. A comparison of IDDES with previous water channel tests, well-resolved LES, partially averaged Navier-Stokes and URANS (unsteady RANS) was included to better understand the benefits of this hybrid RANS/LES approach. The results show that on the medium and fine grids, the IDDES produces an asymmetrical flow topology (known as flow state I) in the near-wake of the vertical midplane, as reported in previous studies. The recommended parameters for the time step (1x10(-4) s) and residual level (1x10(-4)) provide sufficient accuracy of wake predictions to show good agreement with experiments. For the convective term of the momentum equation in IDDES, the bounded central difference discretization scheme is proposed to be adopted for discretization. Additionally, URANS cannot accurately capture this asymmetrical flow field. IDDES proves to be capable of predicting the wake flow field of this simplified heavy vehicle with high accuracy. All obtained conclusions can provide references for the aerodynamic drag reduction of the GTS.
|
|
4. |
- Wang, Fang, et al.
(author)
-
Effects of the windshield inclination angle on head/brain injuries in car-to-pedestrian collisions using computational biomechanics models
- 2024
-
In: Transportation Safety and Environment. - : Oxford University Press (OUP). - 2631-4428. ; 6:2
-
Journal article (peer-reviewed)abstract
- Car-to-pedestrian collision (CPC) accidents occur frequently, and pedestrians often suffer serious head/brain injuries. One major cause is the primary impact with the windshield. Here, we use a numerical simulation method to study the influence of the windshield inclination angle of a passenger car on pedestrian head/brain injury due to CPC accidents. The range of the windshield inclination angle was set at 24 degrees-50 degrees, with an interval of 2 degrees. The results show that the windshield angle significantly affects the pedestrian kinematics and exerts different effects on the head injury when evaluating with various head injury criteria. Regarding the head peak linear/rotational acceleration and acceleration-based head injury criterion (HIC)/rotational injury criterion (RIC), the predictions at the secondary impact stage have no clear relationship with the windshield angle (R2=0.04, 0.07, 0.03 and 0.26, respectively) and their distributions are scattered. In the primary impact, the peak linear acceleration and HIC show a weak trend of first decreasing and then increasing with the increasing of the windshield angle, and the rotational acceleration and RIC tend to remain relatively constant. Regarding the cumulative strain damage measure (CSDM) criterion, the predictions at the primary impact are slightly lower than those at the secondary impact, and the trend of first decreasing and then increasing with the increase in the windshield angle is observed at both impact stages. When the windshield inclination angle is approximately 32 degrees-40 degrees, the head injury severity in both impact phases is generally lower than that predicted at other windshield angles.
|
|