GM May Have Found A Way For Self-Driving Cars To Predict The Future

But there's another approach to driver assistance systems that's proactive, or sometimes predictive, where data like weather and previous navigation mapping can inform new vehicle systems about what the road ahead looks like, so the vehicle - like a Chevrolet Silverado EV - can set itself up to best prepare for the incident, obstacle, bump, turn, or storm ahead of the car.

GM's New Predictive Tech Patent
One of GM's latest tech patents with the USPTO refers to a new advanced crosswind-assist system, mostly for larger vehicles like SUVs, pickup trucks, and commercial vehicles like vans and bigger trucks. The core of the system is a new sophisticated control unit that can manage both real-world live inputs and predictive data inputs and merge them into a seamless driving experience.

There are three key technologies that are working in harmony in GM's new system: a so-called disturbance quantifier, a so-called dynamic observer, and a PIDD computational model (proportional, integral, double derivative).

The disturbance quantifier measures actual live vehicle motion along with predictive data inputs to see just how much the vehicle is being disturbed by a force, in this case, the blowing wind. The dynamic observer is designed to predict stuff like wind speed and direction, and gust intensity. Based on the severity of the wind, the PIDD module then calculates the control signals for the various connected vehicle systems in response to the unwanted motion in an attempt to mitigate crosswind swaying and vehicle movement.

There's Potential For Broader Applications
The patent does not specify whether this is for a self-driving system, but these technologies could easily be applied to self-driving vehicles to make them smarter and safer. A computer driving can use the predictive data far quicker than a human could likely respond, and make minute changes to the vehicle systems for optimized handling and comfort.

Other predictive applications could include a system that raises the suspension for predicted speed bumps or reported potholes that it's been informed of from other data sets. You could also imagine this sort of predictive tech would be handy for towing, and could help self-driving vehicles avoid or handle better in problematic weather that might mess with their sensor suite. If the car has been trained on the road ahead, knows the severity of the storm it's driving into, and has the computational power to make split-second changes in vehicle dynamics, then it will naturally be a safer and more comfortable self-driving experience.

Future vehicles will likely be even more closely connected to one another with Vehicle-to-Infrastructure and Vehicle-to-Vehicle tech systems currently in development. Once the network of vehicles on the road can communicate in real-time, in live settings, the prospect of a self-driving future seems far more plausible. This crosswind assist may not seem like it, but it is one small safety step in a larger effort to make driving safer and further push the adoption of self-driving systems that can handle themselves, and even sort of predict the future.