With the sparse electric vehicle chargers network and longer charging times, range anxiety is still relevant in 2023, and drivers fear longer journeys and whether the remaining battery capacity would suffice for the next charger or destination. From the consumer’s perspective, there are still some barriers to overcome
Preventing range anxiety can be summarised as follows:
Estimating driving range (equivalently energy consumption) with much higher accuracy could dramatically improve the driver's confidence and thus alter the whole car industry towards electrification.
However, an experienced EV driver will tell you that the in-car range estimation is sometimes off by a stunning 20-30% across the board, just enough that drivers stop trusting the cars.
Since all the in-car range predictions periodically update the remaining driving range along the route anyway, note that the range prediction at the start of the trip determines the prediction quality
Physical energy consumption model
As it happens, most carmakers describe how car battery discharges during the trip using physical formulas. In other words, physical formulas try to describe to which extent the car speed, traffic, terrain, temperature, cargo and passengers weight, tire size and pressure, battery type and age, and myriads of other factors affect the battery energy consumption and thus limit the range.
Despite including more and more parameters, the physical battery discharge models fit only to a certain extent and fail miserably in the real world.
Physical energy consumption models can't handle dynamic factors
For example, air resistance can quite precisely be described as a function of altitude, but no mathematical functions can describe whether or not the driver would decide to drive moderately or recklessly fast.
Next, EV drivers will notice how the car preconditions the battery to the optimal charging temperature so the car chargers faster and safer. Although preconditioning speeds up the charging time and preserves the car battery, it also drains a significant amount. When navigating to a supercharger, Tesla automatically starts preheating the battery before the arrival to achieve optimal charging conditions.
Physical energy consumption models are unaware of many dynamic factors and thus unable to include these dynamics in energy consumption prediction. This is why the physical models produce significant energy consumption errors when the dynamics of driving conditions increases. None of the physical-based range predictions can consistently provide accurate results. This inconsistency which occasionally goes up to 20-30%, is the main reason for the range anxiety.
On the other side, the consensus is that Tesla‘s battery consumption and range estimation are the best on the market, and Tesla‘s drivers feel the most confident and trust the reported range.
Tesla’s range prediction does not solely relies on physical formulas, and over the years, Tesla has rolled out a series of improvements in range prediction. The current model evaluates multitudes of car sensors‘ data and combines them with machine learning algorithms to provide the most consistent and trustworthy estimations. Tesla‘s range prediction learns how an actual car and driver consume energy in different circumstances.
To name a few, the data taken into account for navigation incorporates tire pressure, energy loss due to the use of 12V accessories, battery heating and cooling, crosswind, headwind, humidity, and ambient temperature.
On top of the excellent range prediction, the in-car navigation includes supercharger wait times and occupancy to automatically reroute to a less busy charger, stepping up a notch driving experience and boosting the drivers’ confidence in the car and the brand.
If we can create an intelligent vehicle system that learns one's driving pattern based on real-time sensors, traffic, and environmental conditions, we can predict battery consumption more accurately. The machine learning (ML) model will keep learning as one's habits or driving style evolve and vehicle and battery characteristics change.
To prove the feasibility of the machine learning (ML) approach, as part of our EU-funded EV range research, Mireo has developed EV-specific GPS navigation with advanced, AI-based range estimation. To accomplish this, we’ve leveraged Mireo SpaceTime, a car data acquisition and processing platform, and produced impressive results. We tested the built-in battery range prediction on several EV cars, including Tesla Model 3, and achieved highly consistent and precise results. Predictions given at the beginning of the trip proved to have no more than a 4% error.
The upcoming blog post explains how having the right tools at our disposal enabled us to achieve such a level of precision and consistency in our EV range prediction model.
AI based EV GPS car navigation made for eliminating range and charger anxiety
The in-car EV range estimation is every now and then off by a stunning 20-30% across the board.
To mitigate EV range anxiety, machine learning algorithms yield far more consistent results.