Signal Processing and Data Analysis for Real-time Intermodal Freight Classification Through a Multimodal Sensor System.

Identifying freight patterns in transit is a common need among commercial and municipal entities. For example, the allocation of resources among Departments of Transportation is often predicated on an understanding of freight patterns along major highways. There exist multiple sensor systems to detect and count vehicles at areas of interest. Many of these sensors are limited in their ability to detect more specific features of vehicles in traffic or are unable to perform well in adverse weather conditions. Despite this limitation, to date there is little comparative analysis among Laser Imaging and Detection and Ranging (LIDAR) sensors for freight detection and classification. To address this research gap, this work presents an end-to-end prototype that uses a multimodal sensor bundle to classify freight in transit into four categories (single unit, semi-trailer unit, multi trailer, and single trailer). In addition, this work quantifies the potential of six commodity sensors used, four lidar sensors and two video cameras. The results show that using a LIDAR sensor, a machine learning model can classify the four categories with a 94% accuracy. Due to current limitations in data, a computer vision model based on best practice in deep learning is capable of obtaining a precision of 70%. The work concludes by providing an analysis of the state of the current prototype system, and recommends pathways to improve to meet the final objective of real-time classification in a constrained-resource system