Multimodal Fusion Object Detection Technology in Intersection Traffic

The complex nature of the traffic scenarios in autonomous driving and the concept of the intelligent transportation system place high demands on perception system reliability. The sensors used by individuals are normally affected to a great extent in perceiving information when they are subjected to harsh environments such as harsh weather, lower light sources, and complex environments. The need to make use of the multimodal sensor technology has become urgent in creating overall visual and all inclusive perceptions. This paper takes a systematic analysis of the latest developments in this discipline, placing more emphasis on the analysis of three main issues, namely, the problem of feature complementarity by modal diversity, poor adaptivity to complex environments, and the failure to recognize long-tail targets and open environments. To overcome these hurdles, this paper goes further to expound on the common coping styles, propose commonly used datasets in this technology field and finally discusses the areas of concern, potential research opportunities and emerging trends with regard to multimodal fusion object detection technology.