General Motors continues to pour money and time into perfecting vehicle autonomy.

One of the integral themes of an autonomous driving future is communication. Vehicles will be required to rapidly and constantly talk to one another, relaying position, speed, and intent, with the same demands applying to our roadways and intersections which will be expected to relay weather, traffic, and safety information in real time as vehicles approach.

Pedestrians, however, pose a unique problem. In busy cities, pedestrians and drivers routinely communicate non-verbally by making eye contact or gesticulating, but an autonomous vehicle doesn't have that privilege. Its machine code is dependant on a series of binary questions it must ask itself in order to determine if the person is a threat, which, if affirmative, currently results in the car coming to a halt, which on occasion has caused a rear-end collision, or two.

Well, GM is working on a system in which autonomous vehicles will be capable of better understanding the motions of pedestrians by communicating with their Internet-connected devices, be it smart-phones, wearables, or perhaps even future advancements made towards transhumanism.

According to a patent application published on January 30, 2018, by the US Patent and Trademark Office, GM is working on a Vehicle-to-Pedestrian communication system which will be able to determine a pedestrian's presence and proximity based on the interface between the car and the device.

According to the document, the system aims to establish digital contact between the vehicle and pedestrian, focusing in particular on blind situations, where neither party is "aware" of the other's presence.

For example, a pedestrian moving quickly toward the road from behind a visual impediment like a bus-stop, building, or even other pedestrians, may not see or hear the approaching vehicle, especially if they're focused on their phone or using headphones. However, there's also a rarer, albeit far more dangerous situation this system is hoping to avoid: the unsighted pedestrian on a rural road or the hard to see person standing on the side of a freeway. 

After quickly running the location data through a path-prediction algorithm, GM's proposed interface will then either alter the autonomous vehicle's behavior, warn the pedestrian of the vehicle's impending arrival, or in emergency circumstances, like an impending autonomous vehicle crash, the pedestrian(s) could be warned to GTFO.

In some cases, the algorithm may simply determine that neither party can "see" each other, and in turn alert both of each other's presence, without the need to take further action; or it could ask the pedestrian to wait before entering the intersection because the road is icy and the AV approaching might not be able to stop. 

While much of the patent's examples describe autonomous vehicles, the technology is not limited to use when an AV is being operated autonomously, pointing to the potential for the technology to debut before we reach full-on private autonomy.