Estimating Passenger Loading on Train Cars Using Accelerometer

TL;DR

This paper proposes a low-cost method using accelerometer sensors to estimate passenger loading on train cars by analyzing vibration patterns, offering a cheaper alternative to existing advanced passenger counting systems.

Contribution

It introduces a novel accelerometer-based approach for estimating train passenger load, combining theoretical modeling and experimental validation.

Findings

Accelerometer data can reliably reflect passenger loading levels.

The proposed method is cost-effective compared to EO/IR sensor solutions.

Feasibility demonstrated through theoretical and experimental analysis.

Abstract

Crowding on train cars is a common problem plaguing the major public transit agencies around the world. On one hand a crowded train car presents a negative experience for the passengers, while on the other hand it indicated inefficiencies in the train system. The Federal Transit Agency is interested in reducing the crowding level on public transit train cars. Automatic passenger counters (APC) are commonly used to count the passengers boarding and alighting the train cars. Advanced APC solutions are available based on EO/IR sensors and visual object detection technology, but are considerably expensive for large scale deployment. This report discusses a low-cost approach to APC by using accelerometer measurements from train car to estimate approximate passenger loading. Accelerometer sensor can measure train car vibration as the train moves along the rail tracks. The train car vibration changes with the passenger loading on the car. Detecting this change in vibration pattern with changing passenger loading level is key to the accelerometer based APC solution. Moreover, accelerometer sensors present a low-cost APC solution compared to existing EO/IR based APCs. This work presents a (i) theoretical model analysis (ii) experimental data driven approach to demonstrate the feasibility of using accelerometer for passenger loading estimation.