The problem of large calculation models in bus–two-wheeled vehicle traffic accidents (TA) leads to the difficulty of balancing the calculation efficiency and accuracy, as well as difficulties in accident reconstruction. Herein, two typical accidents were reconstructed, based on the rigid–flexible coupled human model (HM) and the Facet vehicle model, and the vehicle damage conditions and the human head biomechanical injury were analyzed. The simulation results showed that the physical process of the human–vehicle collision was basically consistent with the accident video, the windshield fracture was consistent with the actual vehicle report, and the human biomechanical injury characteristics were also consistent with the autopsy report, which verified the feasibility of the simulation model, and provides a basis and reference for forensic identification and for traffic police to deal with accident disputes.

Due to the characteristics of multibody (MB) and finite element (FE) digital human body models (HBMs), the reconstruction of running pedestrians (RPs) remains a major challenge in traffic accidents (TAs) and new innovative methods are needed. This study presents a novel approach for reconstructing moving pedestrian TAs based on a humanoid robot method to improve the accuracy of analyzing dynamic vehicle–pedestrian collision accidents. Firstly, we applied the theory of humanoid robots to the corresponding joints and centroids of the TNO HBM and implemented the pedestrian running process. Secondly, we used rigid–flexible coupling HBMs to build pedestrians, which can not only simulate running but also analyze human injuries. Then, we validated the feasibility of the RP reconstruction method by comparing the simulated dynamics with the pedestrian in the accident. Next, we extracted the velocity and posture of the pedestrian at the moment of collision and further validated the modeling method through a comparison of human injuries and forensic autopsy results. Finally, by comparing two other cases, we can conclude that there are relative errors in both the pedestrian injury results and the rest position. This comparative analysis is helpful for understanding the differences in injury characteristics between the running pedestrian and the other two cases in TAs.

The field of modern autonomous driving and unmanned systems places significant importance on planning and semantic navigation tasks for autonomous vehicles. The successful execution of these tasks relies heavily on mapping and localization technologies. We propose a method to address the difficulty of using trajectory maps of autonomous vehicle cameras in complex environments for planning. The proposed method constructs a topological map that is convenient for subsequent planning and semantic navigation. To enhance the integration of trajectory maps with topological maps, we employ the ORB-SLAM2 algorithm to create the camera's trajectory map, and we extract the keyframe information to obtain frames from the data. These extracted keyframes are then fed sequentially into a junction recognition network that utilizes ResNet18 transfer learning. Then we connect the resulting nodes that are recognized as junctions to generate the camera's topological map. The integration of trajectory and topological maps has been found to be advantageous in improving the planning and navigation abilities of ground robots.