Apple may use Machine Learning to improve the GPS data in Maps
Although the GPS is a widely used technology for geolocation, which is particularly useful for navigation while driving, it is not necessarily as precise as it might seem. The application Google Maps or Apple Maps, show, occasionally, the wrong location of a user for a number of reasons.
These problems can include interference in the GPS signal caused by trees, mountains, tunnels, signals that are reflected on the buildings of a city, solar storms, and even rare cases of radio interference or noise.
These problems are not limited to GPS, other global navigation satellite systems (GNSS) such as Glonass, Galileo, Beidou and others may suffer the same problems.
The patent application published by the patent and trademark Office of the United States with regard to the “Positioning satellite-based automatic learning“. In short, it is a way to analyze the GPS data and comparing it with data from a model of machine learning.
The idea is that the device receives the estimated position based on a signal of the GNSS, acquiring a set of parameters associated to the aforementioned position. Is then provided in a reference position near the estimated position of the device, to facilitate the correction.
In the following, is generated of a model based on the estimated position of the device from the machine learning, the reference position and a series of parameters. This model is then used to estimate the position of the device for future GPS readings, up to a certain period of time or until a user does not move in an area in which the parameters and the model are not accurate.
In fact, the device generates the model using the two sets of positioning data to determine the distance from the GPS coordinates received with respect to its real position. For example, in a city with tall buildings, the model could be informed that the signal may be reflected and, in consideration of this, together with the position readings, previous and to the directorate general of transit, develop a more accurate location based on erroneous data.
Every week, Apple has numerous patent applications and, although the questions indicate areas of interest for the research and development team of the company, does not guarantee the possibility for Apple to add it to a product or service in the future.