A Deep Learning Approach for Drayage Truck Identification Using Inductive Loops

Drayage trucks play a crucial role in the logistics and supply chain, facilitating the movement of goods over short distances, typically between ports, rail yards, distribution centers, and other cargo-handling facilities. However, they are mostly powered by internal combustion engines and impact nearby communities with elevated noise and air pollution. This study proposes a drayage-focused truck data acquisition method that leverages existing infrastructures to provide a better understanding of drayage activities with the potential to support efficient policy-making. A single loop-based truck body type classification was designed with a specific focus on the identification of drayage trucks. The proposed model is capable of accurately distinguishing forty truck body type trucks with an average correct classification rate (CCR) of 81 percent. The model performance on drayage trucks (trucks hauling 20ft, 40ft, and 53ft intermodal containers) was improved over the weighted cross-entropy loss function without significantly compromising the model performance on other classes. The CCR value of drayage trucks is 91 percent which significantly outperforms the state-of-the-art models and has the potential to yield accurate real-world drayage truck activity metrics.