A recent article brought back my focus on VPS, which I have been following since past 10 years. In fact, many of us techy people have played with visual positioning systems going back to 2010 – whilst using Vuforia to recognise objects and track them for Augmented Reality and Virtual Reality applications. After a while, this came to the Google and Apple AR systems.
It was around 2015 that there was some discussion on how this could be used to overcome GPS shadow, which is a loss of GPS signal in an urban area due to large buildings blocking the healthy GPS radio waves reaching our devices. Old AR apps like LayAR were an inspiration.
But it wasn’t until 2018 that Google announced the AR VPS at the I/O Developer Conference. It was initially designed to use for walking directions, whereby you could view the world through your camera and it augment arrows to show you the direction. This improved the accuracy by using recognisable objects. Though, from the outset, it was doomed as it was a horrendous battery drain.
Interestingly, around the same time, a company called Niantic had this really great app called Pokémon Go which they wanted to make the characters more realistic by giving them better recognition of the objects around them.
To overcome this, they started to create a usable global visual library called Lightship. In September 2022, Niantic announced the launch of Lightship, the technology which allows the Pokemon in Pokemon Go to hide behind objects and be more accurate in their position.
Although Niantic created an impressive VPS, it is designed for enhancing the AR experience in the real world using a huge database of 3D real-world objects.
What is exciting about the Google VPS is that the new focus appears to be more around location accuracy. Although Google positioning already uses the latest GPS chips which take account of GPS shadow and also use Wi-Fi to triangulate as well as GPRS, the new VPS can provide an even better position, even extending to indoor navigation.
In 2016 Google announced Cartographer, a SLAM (Simultaneous localization and mapping) library which would allow indoor mapping by using movement to calculate location from a start point. Could the integration of VPS capture the start point with an accurate location to then allow SLAM to be used for a reasonable calculation of where we are indoors?
Also Read: Getting lost with GPS? Don’t worry, Google Maps VPS is here
