AI may defeat EV Range Anxiety - Part 1

What is it like to drive an EV?

On average, when it comes to non-truck, both private and commercial vehicles, everyday drives with electric vehicles (EV) and internal combustion engine ones (ICE) don't differ much. The data shows that such vehicles in North America travel less than 70 km daily (43 mi), which is way below the advertised range even for entry-level electric cars, making them perfectly suitable for day-to-day activities.

However, owning an EV for longer journeys is a proper chore.

The sparse network of charging stations for electric cars and the long charging times are the biggest differences between driving EVs and ICEs. Driving EVs for longer distances can be summarized as follows:

1. Charging takes time, and the charger network is sparse.

• Typical charging on an in-house charger takes about 8 hours, 3h on a public charger, and 20 minutes on a fast charger (supercharger).
• For longer trips, the goal is to head for superchargers and hope there are enough of them along the way.
• Not every charger fits, charger availability information is not always available, and every charger company has a different membership app to use and pay for charging.

1. 20 - 80% range is the optimal operating battery range.

• Running the battery in the 20 - 80% range extends its lifespan and also yields the fastest charging times (charging from 20 - 80% takes roughly the same time as charging 80 - 100%).
• For safety reasons, drivers always prefer a 10-20% battery backup.

3. Highway driving halves the car range.

• Longer journeys mainly occur on highways, but EVs tend to double the consumption on highways.
• The range significantly varies on the road type, speed and driving style, AC/heating usage, weather, route conditions, to name a few. This is usually the root of all evil.

The halved advertising range on highways combined with a safety battery buffer and sparse fast chargers network is what makes the longer journeys nerve-wracking.

Let’s demo what it’s like to drive a 3 year old Tesla Model 3

The advertised range of our Tesla Model 3 Standard Range Plus is 350 km, which translates into a 250 km - 330 km highway range in cold and mild weather. With a safety buffer in mind, this 250 km range translates into charging every 150 km, assuming the fast chargers are available on the route.

#1 Zagreb - Pula, 270km drive is impossible to do in the wintertime.

The only super-fast charger (supercharger) on this 270 km trip is in Karlovac, just 50 km from the starting point. However, driving 220 km with a 250 km battery range in wintertime and on mountainous highways is not for the faint-hearted. But the other options are far away from practical:

• visit a slow public charger in Rijeka
• visit a 2-stall fast public charger in Rijeka, hope it won't be occupied
• 70 km detour via Senj supercharger
• cross-country trip via Ljubljana, Slovenia

#2 Zagreb - Perugia: a 770 km drive requires homework

It's like driving with a toddler - everything depends on when and where the car wants to eat and sleep.

The safest option is to recharge every ~150 km on superchargers or make 4-5 chargings. A driver must expect that something might go wrong and carefully plan the route.

Drivers use either in-car GPS navigations or various route planners to find the optimal routes, but nobody does this tedious chore by hand or relies on seeking the suitable, operating, and available charger at 20% battery left.

But battery consumption varies on dozens of parameters, from car and battery type, route and road type, temperature, AC/heating usage, number of passengers; the sky is the limit. And you have to ask yourself the question:

How on Earth can a route planner know how the car discharges the battery to suggest when it should recharge?

It is crucial to note how accurate battery discharge prediction needs to be; otherwise, recharging locations will be way off. However, since it's difficult to accurately predict battery consumption when it depends on so many parameters, many in-car GPS navigations underestimate or overestimate the remaining battery range.

An experienced EV driver will tell you that the in-car range estimation is sometimes off by a stunning 20-30% across the board. And sometimes, after the nerve-wracking drive to the nearest charger, the suggested charger on the route might not even work at all or might be too busy.

Most of the time, the range estimation is accurate enough, drivers don’t completely drain out the car battery seeking a charger, and most of the time, chargers work just fine. But when the range estimation makes a mistake, it does it colossally - enough for drivers to stop trusting the in-car GPS navigation and refer to it as a guess-o-meter.

This is precisely what the world calls range anxiety - the fear of not reaching the destination with an EV car with enough battery remaining.

Tesla does it better

However, the consensus is that Tesla does it better than others. Tesla drivers simply have more confidence in in-car navigation for longer journeys.

Although Tesla's range estimation fluctuates wildly in the first few minutes of a trip, once it stabilizes it comes very close to actual energy consumption. This is the number one reason why Tesla's drivers can rely on the car's estimation.

Next what makes the difference is Tesla's network of fast chargers - superchargers where a 20-minute charge can bring a 150 km range.

Tesla's in-car navigation also reroutes to less busy chargers, saving charging time. Note that real-time chargers' availability information is almost unprecedented nowadays.

Can Tesla do it better?

AI-based GPS navigation for EVs

With such control over the car's usage and overall high-tech company's offerings, one would expect higher accuracy and more stability in EV range prediction, but Tesla still doesn't do it perfectly.

The answers to the question of how to improve Tesla navigation are summarized in the next blog post, which is the direct result of Mireo's EU project research of efficient AI-based solutions for combating range anxiety.

As part of this project, Mireo has developed an AI-based GPS navigation for EVs, which offers a maximum of 4% range prediction error in range prediction, alongside routing and intelligent charging.