COMMENT: AI applications knock down barriers for EVs
When a driver of a gasoline-powered car runs out of fuel, they know what to do. They can call a friend to bring a full gas can, or call a towing company and pay a hefty service fee for a few gallons of gas. It may be embarrassing or costly, and it is always inconvenient.
How does this situation change when that same driver is in an electric vehicle (EV)? We know that his or her friend can’t show up with a five-gallon can of electrons. Does the towing service have the equipment to charge that car by the side of the road, and if so, how long does that take? Or, do they tow the car to the nearest charging station, which could be miles out of the way?
Research from Guidehouse Insights suggests that 72% of survey respondents wonder these same things, all of which fall under the banner of range anxiety.
An ongoing concern
Range anxiety remains a significant barrier to greater adoption of EVs for some car buyers. For many, it is easier to buy another internal combustion engine (ICE) vehicle. In an ICE vehicle, drivers know that even when the fuel gauge is on E (for empty), the low fuel light is on, or the miles to empty is on zero, they still have a few miles left in the tank. They can push their luck for a while before pulling over to the shoulder. And, it turns out, running out of gas doesn’t happen all that often. According to the AAA National Office, fuel delivery accounted for a relatively modest 2.1% of service calls in 2019. More common problems included flat tyres and lockouts.
In the near future, the new car salesman will be able to show a prospective buyer that their EV does not guess its range: it knows its capabilities precisely
The reality is that EVs come with a new set of rules that ICE drivers need to learn. Some are positive. For example, many shopping centres have charging stations conveniently located near an entrance. Those lucky EV drivers can shop while leaving their car to power up, all while unattended. Still, the surveys show that range anxiety is intimidating to many prospective buyers.
Keeping current in your EV
The industry is not waiting for non-EV drivers to come to the realisation that keeping a vehicle charged is not too hard to manage. It is taking a number of proactive steps to limit the likelihood that an EV driver will find themselves in the unpleasant situation of going from range anxiety to range distress.
These steps include adding more charging stations in different locations, developing batteries that have more energy storage capacity, and helping road service companies develop the tools to serve EVs. Additionally, there has been the development of navigation software that shows all available charging station options and lets drivers know if they might not make it to a destination without charging their batteries.
In each case, as new research from Guidehouse Inisghts shows, the industry has applied artificial intelligence (AI) to enhance outcomes. AI ensures that the siting of new charging stations best serves the local EVs. Battery manufacturers use advanced analytics to develop batteries that store more energy and have a longer lifespan. EV manufacturers and tow truck companies use AI to dispatch a tow truck with the right tools that will get there the fastest. However, some of the more interesting applications of AI are within the navigation application.
The current approach for both EV and ICE vehicles involves estimating how much fuel or energy remains in the vehicle, taking the most recent information on consumption, and doing the maths. To avoid the nightmare scenario where the vehicle runs out of fuel while the indicator says there are a few miles to go, the miles to empty app takes off a few miles just to be sure. Drivers are okay if they can continue driving on empty, but trust is broken if they are stuck on the side of the road when the instrument cluster says they have some miles left.
Navigation software has been developed that shows all available charging station options and lets drivers know if they may not make it to a destination without charging their batteries
EV manufacturers are aware of how social media would pick up on even a few incidents where the instrument panel shows there are remaining miles in a dead EV. Such a video would be damaging. The simple solution is to cushion the number, but this reduces the useful range of an EV.
The solution: more precise estimations of empty
The superior solution is to develop a more precise miles-to-empty range estimate. This involves a few generations of enhancements to further refine the estimates, all of which apply AI technology. A rule of thumb for EVs is that range is affected by the three Ts: terrain, temperature and technique.
The first significant leap to improving the range estimate in an EV involves understanding the impact of terrain by collecting and analysing the actual energy usage of all EVs of a specific model along each stretch of roadway. For example, if a given model of EV uses 3kWh when driving the 12 miles (19km) between mile markers eight and 20 on I-55 near Chicago, Illinois, it would be unreasonable to assume that it would also take 3kWh to drive the 12 miles starting at mile marker seven on the Pikes Peak Highway up to the top. The route in Springfield has a two-foot (60cm) gain in altitude over 12 miles while the route at Pike Peak gains 4,720 feet over the same distance.
A better estimate can be determined by tracking the actual energy use of other similar EVs on the same path by collecting the telemetry data from drivers that were already there. This is a lot of data, but all in a day’s work for the cloud. The good news is that EVs already track all this information and many send it to the manufacturer.
This is the idea behind products such as Bosch’s Battery in the Cloud. The Tier 1 works with EV manufacturers to collect the data from EVs, analyses the data with AI, and provides information to the in-vehicle navigation system. When provided with this information, the navigation system knows what it can anticipate rather than simply hoping that the terrain in the recent past is the same as what is coming.
Improved accuracy of information means the navigation system can tell the driver if it is best to pull over at a charging station sooner rather than later. Other companies have similar offerings to collect this information on location-specific energy usage.
Put the AI tiger in your battery pack
The predictions for energy usage for each stretch of roadway are good, and will continue to get better, which brings us to the second T, temperature, or more generally weather. Weather conditions are a major factor in the energy demand, particularly wind speed and direction.
If a given model of EV uses 3kWh when driving the 12 miles (19km) between mile markers eight and 20 on I-55 near Chicago, it would be unreasonable to assume that it would also take 3kWh to drive the 12 miles starting at mile marker seven on the Pikes Peak Highway up to the top
The US Environmental Protection Agency reports that aerodynamic drag accounts for 39% of the power needed to drive an EV. That means that power needs differ depending on whether the wind in Chicago is causing a nice tailwind for the EV, or if there is gale-force wind coming head-on. Weather information at each location is readily available for AI analysis. The navigation system can also account for an unexpected windstorm and tell the driver that the situation has changed, and that getting a recharge is now necessary. This type of notification will become as routine as getting a real-time alert of a traffic accident.
An EV’s battery management system is what calculates how much power is still available in the battery pack. This component is already very smart, and the application of AI will make it even smarter. The battery management system plus AI combo will help get more energy out of what is there and make better estimates of what remains.
In the near future, the new car salesman will be able to show a prospective buyer that their EV does not guess its range: it knows its capabilities precisely, and the navigation systems can get you out of trouble. One more sales objection fades away, and the sale is made. AI has made the current generation of EVs a dream to drive, and future applications make things even better.
Bill Hughes is Principal Research Analyst, AI and Advanced Analytics, at Guidehouse Insights