Industry Voices: AI Can find the Hidden Catch with Batteries in EVs

Opinion piece by Jim O’Brien, general manager, Americas, Ravin AI

Used cars often come with a hidden catch: an accident not properly disclosed, or a transmission that is near the end of its life.

Electric vehicles are no different but, for them, the hidden catch often comes in the form of the health of a vehicle’s battery. While batteries may be created equal, just how well they do their job after a few years on the road depends on how they were used. Even though battery health is a key factor in the pricing and value of used EVs, unfortunately for consumers, there’s no easy way to tell just how healthy a battery is because that would require knowing how owners have treated vehicles and how much stress was put on the batteries. However now, artificial intelligence and advanced data analysis could help purchasers get a better idea of what they are buying and help sellers, whether private or dealers, fairly and competitively price their vehicles.

Batteries usually come with an extended warranty that can last up to ten years, with replacements available if the battery fails to recharge to at least 70% capacity. So, if an owner replaces their EV after four years, the buyer can rest assured that the battery will work or be replaced but exactly how well will it work? After four years of use, some batteries charge quickly while others will charge very slowly; some will charge to 75% capacity, while others will charge to 95% capacity.

Those differences should have a significant impact on used EV pricing; after all, the battery is perhaps the most important component in an electric vehicle. There are a lot fewer moving parts in an all-electric vehicle; fewer things are likely to go wrong, making battery health an even more prominent vehicle attribute.

However, like with internal combustion-powered vehicles, buyers don’t always have the opportunity to extensively test all the functions of the EV they are buying, such as charging time and capacity, which means they have to take the seller at their word on how well the battery is working.

Those buyers who want to dig deeper into the battery health can do some basic investigation into how the vehicle was used. Did the owner use it to drive mostly on highways, or was it used mainly in urban, stop and go traffic – conditions that are more likely to cause a deterioration in battery capacity and power? Was the vehicle mostly used in an area with a harsh environment, with great differentials in heat and cold (for example, Arizona) that would require constant use of the air conditioner or heater, or in a more moderate climate (Vermont)? An EV’s battery powers not just the vehicle’s powertrain, it powers everything and if capacity is used for an always-on AC, it will wear the battery down even further.

Today guesswork shouldn’t be necessary. EVs are almost, by definition, connected vehicles and data on driving habits and vehicle usage is, in theory, readily available. All that’s required to develop a system to determine actual battery health is analysis of the data streaming in from an EV. AI systems could analyze that data, comparing vehicle usage data with environmental conditions, mileage, location, and other factors, and come up with a score that will definitively indicate just how healthy a battery is.

In addition, such data can be used to encourage drivers to use their vehicles in a manner that will extend the life of their batteries as much as possible. Data analyzed by AI can be translated into concrete actions a driver can take to ensure long-term battery health. These steps can also be gamified to create an encouraging user experience.

For example, AI systems would determine whether a driver is stepping down on the accelerator too aggressively after a red light – an action that could result in excessive energy use. The dashboard could display a gauge that shows expected long-term battery efficiency based on specific actions like accelerating with the driver given a reward (like a free service check) for accelerating properly now and a bigger reward for reaching an efficiency goal for battery performance over time, such as ensuring the battery maintains at least 90% capacity throughout the life of the warranty. Thus, the feedback from AI analysis would be used to prompt drivers into enhancing their vehicles’ battery health. Some automakers have implemented early stages of this feedback with limited success and not really for general consumers. More can and must be done with AI partners.

Accomplishing this will also require effort from the service shops that automakers work with. Service shops could contribute data on vehicle condition and usage, an effort that will be ongoing even as they continue to service gasoline-powered engines. In order to evaluate data on EVs, shops will have to install special equipment, an expense that will be offset by the fact that there will be less for them to do in maintaining an EV which has far fewer moving parts than a standard vehicle. Thus, the resources, time and money, they would have used on changing oil filters or fixing gaskets will now be dedicated to data collection, with some of that data contributing to extending the life of an EV’s battery.

The ultimate reward for drivers, of course, is the enhanced value of their used EVs and the clear feedback drivers receive from data analysis systems on the health of their vehicle’s battery will allow them to set the proper price when they sell or trade in. The minutes saved in the recharging of a vehicle with an efficient battery (for example, halving the full charging time for their vehicle, compared to vehicles whose owners didn’t treat their batteries properly) can, and should, be worth thousands of dollars in resale value or trade-in value That’s a fair way to gauge the health of a vehicle’s battery and to ensure that drivers who do their utmost to maintain battery health are rewarded appropriately.