Improving Loop Counting Algorithms for Public Transportation System Service Level Monitoring Through the Incorporation of a Tolerance Mechanism

With the ever increasing amount of GPS-equipped public utility vehicles, we can naturally expect an increase in transportation-related data as well. This data can be utilized for service level monitoring, such as loop counting, which involves counting the number of times a vehicle correctly takes its set route without taking any shortcuts or detours. In this paper, we propose a grid-based algorithm that can be used for loop counting. However, since GPS data is inherently imprecise, the algorithm will not produce very accurate results. Furthermore, overlaying a grid map could lead to cutting through a trajectory awkwardly, creating ambiguity as to which grid cell the GPS data point belongs to. These issues lead to over counting or under counting the number of loops taken, and can be addressed by incorporating a tolerance mechanism, allowing points to fall in a defined region outside the cell. We test the effectiveness of both the loop counting algorithm with and without tolerance on a real-world dataset that was taken by a GPS tracker placed in an EDSA Carousel bus, a bus rapid transit system in Metro Manila, Philippines.