International Journal of Vehicle Systems Modelling and Testing (4 papers in press)

Regular Issues

• Semantic context-induced fast fusion network based driver attention prediction in complex scenarios  
  by Jingllei Ren, Hailong Zhang, Yongjuan Zhao, Cong Lan 
  Abstract: Clarifying driving intention through the utilization of the visual selective attention mechanism remains a pivotal research question in a domain of advanced driver assistance systems and human-machine collaborative autonomous driving technology. This paper proposes a semantic context-induced fast fusion network (SCFF-Net) segmenting the RGB (Red Green Blue) video frames into images with different semantic regions and develops attention strategy to fuse the semantic context features of semantic images with the features of RGB frames to explore the complementarity among different features. A mixed model of self-attention and convolution integrated with the self-attention mechanism is further introduced by combining the global perception capability and the local feature extraction capability. Experimental results on the driver attention in driving accident scenarios dataset show that the proposed SCFF-Net can effectively improve the prediction accuracy of driver attention and the computing efficiency. It can also reduce redundant calculations.
  Keywords: driver attention prediction; AC-mix; complex driving scenarios; computer vision; deep learning.
  DOI: 10.1504/IJVSMT.2025.10069760
   
  
• On dynamic path planning based on the DBSCAN-AGA algorithm  
  by Yongjuan Zhao, Jiangyong Mi, Hailong Zhang, Pengfei Zhang, Wenzheng Cheng, Haidi Wang, Chaozhe Guo 
  Abstract: With the advancement of intelligent vehicles, unmanned driving technology has achieved significant progress, particularly in low-speed park settings. However, challenges arise in the park connections due to the dynamic variations of passengers and the complexities of road conditions, making it difficult to implement dynamic path planning for traffic demand distributions. This paper introduces a path planning algorithm based on the adaptive genetic algorithm (AGA) for connecting vehicles on variable routes. This approach involves constructing an origin-destination (OD) matrix based on passenger’s origin and destination points, and incorporates the density-based spatial clustering of applications with noise (DBSCAN) to reassign traffic demand by adjusting routing of connecting vehicles according to traffic demand and road network traffic conditions. The obtained results validate the effectiveness of the proposed method, demonstrating that the DBSCAN-AGA algorithm exhibits strong robustness and reliability in dynamic environment path planning.
  Keywords: intelligent vehicles; dynamic path planning; complex road conditions; DBSCAN-AGA.
  DOI: 10.1504/IJVSMT.2025.10070326
   
  
• TWICE dataset: digital twin of test scenarios in a controlled environment  
  by Leonardo Novicki Neto, Fabio Reway, Yuri Poledna, Maikol F. Drechsler, Christian Icking, Werner Huber, Eduardo Parente Ribeiro 
  Abstract: Addressing the challenge of ensuring the safety and reliability of autonomous vehicles in adverse weather conditions remains a pressing issue. In response, we have developed a comprehensive validation dataset. This dataset comprises data collected from environment perception sensors such as cameras, radar, and LiDAR, gathered both on a real test track and replicated in simulation for consistent test scenarios. Our dataset serves as an important asset for assessing and validating the performance of object detection algorithms specifically designed for autonomous vehicles, particularly in challenging weather conditions such as rain, nighttime, and snow. We have documented various test scenarios involving different entities of interest, including cars, cyclists, trucks, and pedestrians, drawing inspiration from the European New Car Assessment Programme (Euro NCAP). The sensor data captured in our laboratory is the result of simulationbased test drives conducted within a novel hardware-in-the-loop testing framework. This setup not only utilizes the same perception system employed during real-world test drives but also incorporates a digital twin of the actual proving ground. Spanning over two hours of recordings, the dataset encompasses more than 280GB of data.
  Keywords: autonomous driving; environment sensors; camera; radar; LiDAR; hardware-in-the-loop.
  DOI: 10.1504/IJVSMT.2025.10070327
   
  
• A PSO-based method for road rut measurement with line-structured light  
  by Yuanbo Mu, Qingzhou Mao, Guangqi Wang, Chaowen Tu, Dehui Lai 
  Abstract: Road rut depth is a vital metric for evaluating pavement quality, traditionally measured manually but now assessed using automatically laser-based device for the efficient data collection. This technique faces challenges due to the complex and variable nature of rut profiles and road conditions, leading to inconsistencies and interference from lane edges and debris. This study presents a novel method for rapid and precise rut depth measurement by employing line-structured light technology integrated with the inertial measurement unit and GPS sensors. The device undergoes meticulous calibration for accurate 3D road surface data acquisition, involving both line-structured light and positioning sensor calibrations. The collected high-resolution data is then refined using particle swarm optimization algorithms to enhance the accuracy of rut depth estimates. Experimental results demonstrate that this method does not only improve the measurement accuracy and efficiency but also shows strong adaptability, making it a reliable tool for the road quality assessment.
  Keywords: rut depth; line-structured light; 3D measurement; PSO; particle swarm optimisation.
  DOI: 10.1504/IJVSMT.2025.10070437