One of the basic problems in track and trace for local deliveries is getting live and accurate location data of persons or items. Traditional logistics companies have used RFID or standalone GPS chips for this purpose, but the rise of smartphones means such standalone devices are not required. In the on-demand economy the ubiquitous smartphone is the source of location data. Let’s see what location data from smartphones looks like and how it can be used to build a smooth tracking experience.
Smartphones determine outdoor location through a combination of methods—cellular towers, WiFi, and GPS. Among these methods, GPS is known to be more accurate with a potential precision of mere centimetres. However this potential is difficult to achieve with ordinary equipment and raw data has measurement errors. In the figure below compare the raw GPS data points in red against the actual path which is shown in blue.
We compared measurement errors in raw location data from an Apple iPhone 5 and a Motorola Moto E. Measurement error for a data point was taken as the shortest distance of the data point from a nearby road. A greater distance implies higher measurement error.
The iPhone showed higher accuracy with 80% of data points within 5 metres of a road. In contrast roughly 20% of the points from the Moto E were more than 20 metres away from a road. Note that these data points were collected at the finest precision setting possible and a coarser setting could give worse results. Raw GPS data is not only noisy the noise varies a lot from phone to phone.
With such variable measurement errors tracking can be a rough experience. Imagine a hungry customer getting distressed seeing his food stalling and skidding on a map. Or a patient getting anxious when a nurse in-transit appears to be on a different road altogether. Cleaning raw GPS data is crucial to give a smooth and useful tracking experience.
What makes this problem challenging is that cleaning needs to happen in realtime as the customer is waiting for the delivery. As measurement errors vary from phone to phone the solution needs to be agnostic to the phone type. At HyperTrack we are building filters to clean all raw location data in realtime. Compare the animations below to see how the tracking experience feels like before and after our filters are used.
Tell us what you think!
Subscribe to HyperTrack Blog: Imagine. Build. Repeat.
Get the latest posts delivered right to your inbox