A Discussion on Shape Models in Extended Object Tracking for Radar-based Intelligent Transportation Systems

This study evaluates the shape models in Extended Object Tracking (EOT) for Radar (radio detection and ranging)-based Intelligent Transportation Systems (ITS). In the perception systems of ITS, road-side ensors are required to cover large areas and deliver efficient descriptions of the scene. Radar sensors with their long sensing distance, make a good candidate for the task. Their high resolution, combined with the extended object tracking, which estimates the position, orientation and shape of the object, could provide a detailed description of the scene. The discussion will address the shapes models employed in such tracking applications, along with the requirements for their implementation. Furthermore, the coexistence of sensors within a network has the potential to enhance the accuracy of multi-object tracking results. The study also considers scenarios, particularly those involving a decentralized fusion scheme, in which the tracking results are combined. The study culminates in the proposal of a concept for a shape model for radar-based ITS.