According to a recent patent filing, Apple’s engineers are working on a self-driving navigation system that uses sensors to determine which route to take, with no reliance on stored or downloaded maps.
In other words, Apple is attempting to develop a 21st century equivalent of the cliched man-who-refuses-to-ask-for-directions.
According to the patent, the self-driving navigation system would use its suite of sensors and cameras to develop its own map of its surroundings. If Apple puts it into production, it would make for a more robust system that is unhindered by the need for pre-mapped data, as is the case with most other self-driving cars currently in the works.
While minimizing the need for maps is a key feature, the underlying differentiator to other systems would be Apple’s reliance on an artificial intelligence designed specifically to work in tandem with chips designed in-house. Theoretically, that constitutes the framework for a standalone navigation system that could be integrated into multiple automakers’ self-driving cars. Think Apple CarNav.
In its most basic form, Apple’s patent isn’t for a navigation system, so much as an AI decision-making process that happens to specialize in navigating. One of the more interesting facets of Apple’s system is its use of vehicle-to-vehicle communications as a key element in its decision-making tree.
Essentially, as a vehicle approaches an area where it must make a decision, it assigns each option (i.e. turn left, right, or continue on the current road) a “confidence factor.” If that confidence meets a predetermined threshold, the choice is made. If the system is not confident enough in a decision, however, it enters a decision-making process that includes gathering data from other vehicles that have taken the same path. Once the decision is made and the car begins along its chosen path, it then relays relevant information to other vehicles, so they can use it for the same purpose.