Estimating Passenger Queue for Bus Resource Optimization Using LoRaWAN-Enabled Ultrasonic Sensors

Abstract

Anecdotal evidence has shown that bus stops around our University campus, specifically those serving express buses to city center, can get very crowded during certain times. This not only causes immense frustration to students who experience large variations in wait-time, but also creates challenges for the university and transport authority in knowing when to schedule extra buses. This article outlines our efforts to instrument a main bus stop on our university campus with Internet of Things (IoT) sensors to monitor passengers queue length. Our specific contributions are as follows: 1) We begin by developing a LoRaWAN ultrasonic sensor for detecting people in the queue. The sensor emits an ultrasonic tone pulse every few seconds, and determines whether someone is in front of it based on the reflections received, if any. Ten sensor units are built, tested, and tuned in a lab environment to achieve optimum detection accuracy and data transmission rate; 2) Next, we install these sensors at a 6-m interval along the campus fence bordering the bus-stop. We develop an algorithm to infer the number of passengers in the queue from sensor data and demonstrate that it achieves reasonable accuracy with a mean absolute error of 10.7 people (for a queue size of up to 100 people); and 3) we develop an optimization model to reschedule bus dispatching time, aiming to minimize total wait time of passengers. We show that a reduction of up to 42.93% in passengers’ wait time can be achieved by adopting demand-driven bus scheduling.