Understanding vehicle movement from A to Z
Breaking down the trips algorithm
The trip construction algorithm consists of three stages:
- Stage 1 – Filtering for relevant GPS points.
- Stage 2 – Labeling GPS events based on vehicle movement and engine status.
- Stage 3 – Defining trips based on GPS events, and calculating statistics for each trip based on GPS points.
The diagram below shows a detailed example of how a trip is constructed from raw GPS points, where the x-axis represents a particular point in time.
Transforming GPS points -> GPS events -> Trip events. Only some points are filtered through the next stage. In particular, though there may be several start and stop GPS events (see MidTripIdle in Trip 2), only a few qualify as start and stop trip events.
There are three stages to building a trip:
Stage 1 – Filtering GPS points
At this stage, the raw GPS data is filtered down to only those that are relevant for building trips. This filtered data also will include information such as speed, ignition status and geolocation information.
This crucial step filters through the massive amounts of GPS data points to only keep meaningful data points. The resulting data includes information about the vehicle’s previous and next data points. Interpreting each data point requires the context of earlier and later GPS points.
Stage 2 – Labeling GPS events
At this stage, GPS events are classified based on a specific point in time.
A GPS event is an intelligent labeling of a GPS point based on the context of earlier and later GPS points. There are only four possible GPS events, each identifying a state of the vehicle’s movement or engine:
- On: The engine is currently running and was not running in the immediately preceding time point.
- Start: The vehicle is currently moving and was not moving in the immediately preceding time point.
- Stop: The vehicle is no longer moving and was moving in the immediately preceding time point.
- Off: The engine is no longer running and was running in the immediately preceding time point.
This stage also includes a comprehensive yet efficient handling of a wide variety of edge cases, including GPS signal loss and unreliable ignition signals.
On (in red) and Off (in white) Location
Start GPS events. Note that there are many GPS points that do not qualify as GPS events.
On/Off and Stop GPS events – there are overlaps between start GPS and stop GPS points.
Stage 3 – Defining trips
At this stage, trips are defined by identifying which pair of Start and Stop “GPS events” should qualify as the Start and Stop “trip events”. Each pair of Start and Stop trip events then define a distinct trip. In this stage, important data is gathered such as start and stop location, start and stop times, trip duration, distances, as well as pre- and post- idling information.
Complete trip – overlaying the path of the trip
A Start GPS event qualifies as a true Start trip event if any one of the following criteria are met:
- The previous event was an On event
- The previous event was a Stop event that occurred more than 200 seconds ago
A Stop GPS event qualifies as a true Stop trip event if any one of the following criteria are met:
- The next event is an Off event
- The next event is a Start event that occurs more than 200 seconds later
Each pair of qualifying Start and Stop events are combined to define a trip. It is straightforward to then define basic metrics such as distance traveled in the trip as well as the duration of that trip. With a well defined start and stop to the trip, other important statistics can then be calculated from the raw GPS points, such as the time the vehicle spends at different speeds throughout that trip, as well as idling durations.
Trips in action
Transportation engineers and planners often are not only interested in the trips a vehicle takes, but also on a vehicle’s ultimate origin and destination. Consider the case of a long-haul truck’s journey from a supply depot to its ultimate destination. This journey could consist of several trips related to the truck’s operation, such as weighing stations, refueling or rest stops. Due to the careful construction of trips, we are able to flexibly “chain” trips together to determine a vehicle’s ultimate origin and destination using post-trip stop information. This is useful information when looking at border crossings, as well as when vehicles move to and from ports.
For many commercial logistical efficiencies, understanding where a vehicle is coming from and what their final destination is helps to keep freight moving at the right pace. Considering the consequences witnessed with multiple disruptions in the supply chain, using trip data along with origin and destination information can aid in investigating delays and their impacts on truck congestion at ports.
Combining trip information with vehicle industry data can also help planners gain a better understanding of which industries are contributing the most to freight traffic overall. Taking a year-over-year approach, planners can then see where the biggest increases were identified in traffic volume.
Getting granular into a vehicle’s purpose, armed also with trip and journey data can also help start to fill in gaps for transportation planners when it comes to understanding vehicle behavior. For example, identifying vehicles used in door-to-door delivery services and then being able to hone in on their trip behavior can highlight infrastructure adjustments that might be necessary to accommodate, like curbside parking.
Origin and destination and “trip chaining” information provides a new level of data-driven insights to planners for those critical infrastructure planning and transportation funding decisions.