Date of Award

12-11-2025

Degree Type

Thesis

Degree Name

Master of Science (M.S.)

Department

Electrical & Computer Engineering

First Advisor

Dr. Kamrul Hasan

Abstract

LSTM retains information from long-term data and is highly effective in generating time-based forecasts. Recently, BiLSTM has enhanced the precision of traffic predictions by assimilating information from both antecedent and subsequent directions via bidirectional data flow. BiLSTM demonstrates exceptional efficacy in the analysis of long-term traffic data. Our study presents a custom forecasting model that leverages advanced LiDAR sensor technology in combination with state-of-the-art deep learning techniques to improve road conditions. Eight LiDAR sensors were strategically deployed along the 26th Clarksville Pike corridor to collect high-resolution traffic data—including vehicle speed and count/flow—over the period from September 1, 2024 to November 30, 2024. The rich dataset was used to design and implement a Bidirectional Long Short-Term Memory (BiLSTM) network capable of capturing both forward and backward temporal dependencies in traffic flow patterns. This dual-directional approach enables a more accurate understanding of traffic dynamics, thereby facilitating timely and informed decision-making for traffic management. Extensive experimental evaluations reveal that the proposed model achieves an impressive overall detection accuracy of approximately 93\%, underscoring its potential as a reliable tool for short-term traffic forecasting. Our findings indicate that predictive models such as this not only enhance traffic safety but also contribute significantly to the efficient operation of intelligent transportation systems for 15, 30, 45, and 60-minute time horizons compared to other traffic forcasting model, offering a scalable and robust solution for real-time traffic monitoring and management.

Download

Included in

Engineering Commons

Share

COinS