Reconstructing Transit Vehicle Trajectory Using High-Resolution GPS Data

TL;DR

This paper presents a method to reconstruct smooth, continuous transit vehicle trajectories from high-resolution GPS heartbeat data, enabling detailed operational analysis and delay identification.

Contribution

It introduces a novel approach combining local regression and monotonic cubic spline interpolation to reconstruct complete vehicle trajectories from sparse GPS data.

Findings

Reconstructed trajectories are smooth and continuous.

Method enables delay detection near infrastructure.

Applicable to various transit performance evaluations.

Abstract

High-resolution location ("heartbeat") data of transit fleet vehicles is a relatively new data source for many transit agencies. On its surface, the heartbeat data can provide a wealth of information about all operational details of a recorded transit vehicle trip, from its location trajectory to its speed and acceleration profiles. Previous studies have mainly focused on decomposing the total trip travel time into different components by vehicle state and then extracting measures of delays to draw conclusions on the performance of a transit route. This study delves into the task of reconstructing a complete, continuous and smooth transit vehicle trajectory from the heartbeat data that allows for the extraction of operational information of a bus at any point in time into its trip. Using only the latitude, longitude, and timestamp fields of the heartbeat data, the authors demonstrate that a continuous, smooth, and monotonic vehicle trajectory can be reconstructed using local regression in combination with monotonic cubic spline interpolation. The resultant trajectory can be used to evaluate transit performance and identify locations of bus delay near infrastructure such as traffic signals, pedestrian crossings, and bus stops.