AI is getting a lot of attention these days, mainly as it is applied to many applications. Continuing stories about Microsoft’s ChatGPT and ImageGPT create human-like creations that are hard to believe and pose new issues for us to grapple with.
One key area AI can be beneficial is in cars (both electric and gasoline). As a learning system, AI in vehicles can improve safety and performance, reduce emissions, extend battery life, and help reduce driver distractions. However, as with the emergence of any new technology, it depends on how it is applied and used.
Even non-learning AI systems can do speech recognition reasonably well, but a learning machine can do better. It can evolve and improve what would be autocorrect issues. As it gets familiar with a specific voice, for example, it can compensate for accents, speech impediments, and improper grammar and more accurately decipher speaker intent. This is key in eliminating the large and complex touch screens in modern vehicles that are more distracting sometimes than using cellular devices.
AI can also do a better job of preventing accidents and collisions. Once integrated into the external sensors, it can more accurately predict when another driver performs lane changes erratically, when something is approaching a collision zone, and perform self-driving more accurately. It is unknown when the crossover point will be reached when self-driving is demonstrably better than human drivers, but it is coming.
This is especially true for detecting objects on the road that a human doesn’t see. With video and LIDAR, the ability to perform edge enhancement will highlight anything from road kill to potholes and debris in the driving path. Heads-up displays on the windshield can inform the driver before the driver can detect an object. This is especially true in snowy, foggy, or other reduced visibility conditions.
Even navigation systems can benefit from learned behavior and preferences. Often, a GPS system will route the driver through roadways off the beaten path. During a snowstorm, for example, the entire street may become obscured with snow covering. A driver may not know where the road ends on secluded back roads and a lake begins. The navigation system can avoid dangerous routes when visibility is obscured by tying navigation to weather conditions.
Optimizing battery life is also something an AI can help with. We may not even be aware that the rate of acceleration dramatically influences how long a battery will last before needing another recharge. The same holds true with optimizing gas mileage and reducing emissions with legacy fuel-based vehicles. Accelerating just a little slower can improve driving range. The AI can also learn how to help an electric vehicle charge faster.
Experience lets the AI know if the battery is starting to fail before a motorist becomes stranded. It can also warn or prevent catastrophic battery pack failures by preventing too rapid a discharge. This is important in cold climates where, in addition to driving, heat is needed. A stranded motorist will need every minute of heat. This has already been an issue with electric vehicles where people have frozen to death in severe climates.
How an AI can handle a vehicle in snow or ice may be most beneficial. It may detect black ice before the driver and avoid a spinout. It may also react faster when fishtailing or sliding and do a better job of steering with anti-lock brakes engaged.
It will be necessary for the vehicle to store knowledge and information locally because communications networks can fail. Knowledge upgrades at a dealer may allow a car to go to school and expand its knowledge base due to learned behavior from a larger population sample. This is an evolving technology in many ways, and learned behavior must be scrutinized to be safe for everyone. Pretty soon, we will all be backseat drivers.