Apple thinks it can improve location accuracy by applying machine learning to Kalman estimation filters, a just-published patent application reveals. Kalman filters are popularly used in GPS and robotic motion tracking applications. And, now Apple wants to use machine learning along with Kalman filters to bring the accuracy of positioning down to centimetre-level.
How Machine Learning Can Help GPS
According to the patent application, Apple proposes:
- A device that implements a system for estimating device location based on a positioning system comprising a Global Navigation Satellite System (GNSS) satellite, and receives a set of parameters associated with the estimated position.
- The processor is further configured to apply the set of parameters and the estimated position to a machine learning model that has been trained on a position relative to the GNSS satellite.
- The estimated position and output of the machine learning model is then provided to a Kalman filter for more accurate location.
The device, say an iPhone would generate a machine learning model, for example, by comparing GNSS position estimates (or estimated measurement errors) with corresponding reference position estimates (where the reference positions correspond to ground truth data).
In one or more implementations, the ground truth data may be better (e.g., significantly better) than what a mobile device alone can perform in most non-aided mode(s) of operation. For example, a mobile phone in a car may be significantly better aided than a pedestrian device, because the motion model for a vehicle is more constrained, and has aiding data in the form of maps and sensor inputs.
Tall buildings and tree cover can confuse the positioning systems to accurately locate the user. So, Apple wants to generate machine learning models on the device that would predict the user’s location based on its training as well as a reference position.
Today even Apple’s rivals praise Apple for it has done to the electronics industry. In an in-depth CNBC interview with Huawei’s founder and CEO Ren Zhengfei earlier this year he spoke about how Apple has revolutionised the era of the Internet. In his ascent, however, Apple has put many traditional companies to dust.
According to a 2018 CNBC’s report, there has been a dramatic decline in worldwide shipments of cameras. The chart above from Statista illustrates this fall, which also coincides with the peaking of Apple iPhone and its ever-improving camera.
So, the companies those who outsource their GPS improving services will be watching the new ML-based GPS patent closely or even might be rushing to build something of their own. However, this might not be the case in this modern era of mega collaborations.
Last month we saw one of the biggest corporate crossovers of the 21st century, when the tech giants, Amazon, Apple and Google, along with others announced their plans to develop compatible smart home products together.
Gone are the days where companies build something up from scratch (with the exception of Tesla). If your rival company is good at something you are not, then you either buy a startup that works solely on that technology or join hands with the rival. So, Apple’s patent to improve GPS in the upcoming 5G era might receive a warm welcome.
Of course, there always will be a debate about whether one should patent widely used technology, which can hand over infinite leverage to a single entity.
That said, the last two years has since increased attention in seeking patents over ML-based techniques. Last year, it was Google, which has been in the news for patenting machine learning techniques such as batch normalisation. Companies like Google and Apple have been leading the AI race for quite some time. It can also be possible if it is a routine to apply for a patent for their innovations and this new-found obsession over ML patent news is due to the rising popularity of AI globally.
At the end of the day, it comes down to whether you should risk years worth of intellectual property to a potential patent troll or safeguard it through patenting and then democratise the technology to the masses. It has been the latter, for many years and we have to wait and watch if machine learning-based patents find an exception as we go forward.