Search

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. [ABSTRACT FROM AUTHOR]

The engagement of the lap belt with the pelvis is critical for occupant safety during vehicle frontal crashes to prevent occupant submarining. This study aims to develop a predictive model for submarining risk based on anthropometric parameters and lap belt positioning using finite element (FE) analyses. FE analyses were conducted using human body models representing various body shapes (a 50th percentile male, low and high BMI males, and a 5th percentile female) in three seated postures (standard, reclined, and slouched). The lap belt-ASIS overlap and the belt-pelvis angle were used as key parameters for predicting submarining risk. A logistic regression analysis was utilized to correlate submarining occurrence with the initial values of these two parameters at the beginning of impact. Subsequently, this submarining prediction model was applied to computer tomography (CT) measurements of human subjects in different seated postures (upright, reclined, and slouched), and submarining risks were calculated based on the developed model. FE simulations indicated that submarining was more likely to occur as the initial belt-pelvis angle approached zero and there was a smaller initial belt-ASIS overlap. The logistic regression analysis demonstrated that the initial belt-pelvis angle and belt-ASIS overlap were statistically significant for predicting submarining risk. The derived model effectively distinguished submarining occurrence based on the initial values of these two parameters. The application of the submarining model to CT measurements of human subjects showed that submarining risk was lower in the order of upright, slouched, and reclined postures. In the reclined posture, the high submarining risk was attributed to a small belt-ASIS overlap and a rearward-tilted pelvis angle; whereas in the slouched posture, the risk was mostly associated with a rearward-tilted pelvis angle. The submarining prediction model was developed based on the belt-pelvis angle and the belt-ASIS overlap. This predictive model may help to design restraint systems for various body types and seated postures of occupants. [ABSTRACT FROM AUTHOR]

To investigate the global kinematic differences between normal and emergency avoidance postures of cyclists at different crash speeds and the risk of head and lower extremity injuries during vehicle crashes. Seven postures (three normal and four emergency avoidance postures) were selected based on video information from 200 bicycle accidents in the VRU-Traffic Accident Video (VRU-TRAVi) database. Fourteen simulations were performed using the Total Human Safety Model (THUMS) 4.0 occupant model considering the seven postures and two-vehicle impact speeds (20 and 40 km/h). The global kinematics of the cyclists, as well as head injury parameters such as peak linear acceleration, head injury criterion (HIC), peak angular acceleration, brain injury criterion (BRIC), maximum principal strain (MPS), cumulative strain damage measure (CSDM), and lower extremity injury parameters such as von Mises stress and bending moment were calculated to assess the risk of head and lower extremity injury. The results showed that the moment of head contact with the vehicle was earlier and the slip distance of the pelvis was shorter in the emergency avoidance postures compared to the normal posture at vehicle speeds of 20 km/h and 40 km/h, which reduced the head wrap-around distance (WAD) and increased the head impact velocity. In addition, the emergency avoidance postures did not significantly reduce the risk of head injury compared to the normal posture; instead, they increased the risk of injury to the lower extremities. [ABSTRACT FROM AUTHOR]

In momentum-based impact models used in accident reconstructions, the restitution coefficient is a key role to determine impulses and delta-V accurately in car-to-car collisions. The restitution coefficients can be defined as the normal and the tangential direction in the contact plane, and the same value of the normal and tangential coefficients is employed in a two-dimensional PC-Crash model. Ishikawa (1993) proposed an impact model with the normal and tangential restitution coefficient, however, to the best of the author's knowledge, few studies have further investigated tangential restitution coefficients in detail. In this study, the normal and tangential restitution coefficients of the impact model were investigated based on the finite element (FE) analysis. FE simulations of angled barrier crashes with varying barrier angles as well as 16 car-to-car collisions were conducted. As the contact plane slides, the ratio of the tangential impulse to the normal impulse becomes the same value with the maximum friction coefficient. The tangential restitution coefficient is related to whether sliding occurs on the contact plane, and can be expressed as a function of the impulse ratio. The tangential restitution coefficient decreased with increasing the impulse ratio and is positive in non-sliding collisions and negative in sliding collisions. An approximate expression for the tangential restitution coefficient was formulated as a function of the generalized impulse ratio. The calculated impulses of the impact model with normal and tangential restitution coefficients were closer to the FE analysis than the PC-Crash model which uses a constant restitution coefficient of 0.1. [ABSTRACT FROM AUTHOR]