Interview: David Mitchell on How weather will influence the brain of autonomous vehicles

David Mitchell is the Vice President of Digital Media, Emerging Platforms at AccuWeather, responsible for the suite of accessible mobile and connected products within the emerging platform’s marketplace. David is also one of the key speakers at City as a Lab Summit in Ljubljana, just 12 days away. We spoke with him about some interesting aspects of the weather influencing the coming era of autonomous vehicles.

One of the important questions about autonomous vehicles is, will they be safe during bad weather and how can you help the passengers stay safe?

With autonomous, one of the biggest things is many sensors around the car. They know what is happening very closely or in the proximity of the car, even things like the weather. However, what is happening around the curve or in the next half an hour is not known. And I think it is very important what is happening, especially if freezing or heavy rain is expected and the road conditions are going to change dramatically – or not change at all. Sometimes, the sky looks bad and gives the wrong information about what is going to happen. In this case, the information “don’t worry, the weather will be fine” can be in a big help for a driver.

We have a habit of going to holidays usually crossing the Alps where the weather changes constantly. It happens that let’s say a thunderstorm is just 100 kilometres away. If I knew that, I would rather go for lunch in the meantime. This information would be very helpful for autonomous vehicles, that need to make a decision on either continuing or changing the route. What role will AccuWeather have in the process?

One of our core products for connected vehicles is the mini cast on the route. It predicts the weather in the next 120 minutes edited on a minute-by-minute basis. Therefore, you actually know what the weather will be when you get to a certain point that is one or two hours in front of you. Technically, it tells you if you are going to experience rain or not on your route.

There is numerous information needed for making a system – an autonomous brain in the car to make a right decision. Which ones?

Where really useful dramatic products happen in the connected car space or even in city space, is combining information. I think for a long time, we really had good single sets of information, weather being one of them. I can find the information and in some cases, I can look at the camera where I can find info about the section where I need a tire change or if this section is closed etc. But I don’t think anybody has done well by combining those data sets, which are kind of limitless. So far, nobody took the time to combine the coolest or the most useful types of data sets.

Here we come to the real answer about the weather products. I am certain that the most compelling product is not a standalone construction or traffic product, but it’s the product as a result of common work of all the organizations and delivering this combination of data in a standalone product.

We are in the early discussions with certain companies about combining information that is going to be in the navigation system. That is the first piece. It’s not going to be an AccuWeather app, but we bent it into the navigation apps which starts to get closer to the idea. However, there will be other pieces missing that could provide even better information. To make that brain useful, we need to gather all the pieces.

If I am going to stop because of the weather or traffic, I will run out of my options. What happens next? Should I eat lunch or go to a cool location, a location I have never been to before, are any of my friends close by? Therefore, there’s a lot more information we could combine.

Is or will AccuWeather be able to provide information about surface – mud, snow and other road challenges to provide information for car’s right decision? With the information about the temperature on the micro-location, will we be able to calculate how firm or soft the snow is?

We currently provide some information about how weather conditions will affect roadways.  With combined data sets that include road surface type, information on if the road has been cleared or treated and weather data the information you list is possible.

How will the system work for a single user? How will the user gain micro weather reports and predictions and how much will that cost?

Right now, there is a lot of visual information, so it does not distract the driver in the middle of a drive. Therefore, we will visually show the weather prediction on the route and the waypoints that are constantly installed in the navigation system. In addition, we will constantly update that information within the next two hours and provide the warning in special colour-coded signs.
Green is going to show beautiful and sunny weather, red will warn about dramatic conditions and purple about heavy wind or even tornados, that could dramatically affect the drive.

We are also thinking about the voice-informing the driver because we think that there are huge opportunities to proactively notify the driver for the upcoming weather changes.

Who will be the right customer for your data – Tech companies that will take care of the autonomous system to work or some other companies, individuals?

There are several organization in the car ecosystem that are customers or potential customers.  Our weather data can certainly be provided to organizations developing software and algorithms to support autonomous and connected vehicles. In some cases, we will build our own products for direct consumption by consumers (drivers).  Even when we build the product, the best products will include data from the vehicle and other content providers.

What technology do you use to gain detailed information and how precise actually is?

There is a ton of data that goes into it. We don’t own any hardware, so we are purchasing data or get them for free from government institutions. Therefore, the income of data for our forecast information is huge. We are also getting data from connected vehicles today, but so far not in a large amount. When I talked about our Minicast product that is based on one one-kilometre grid in most locations, one of the most important inputs to that feature is real high-quality radar. It shows the location of the precipitation and we can take all the other meteorological data and start to talk about what is happening on the ground. In most cases, it’s highly accurate within the two-hour window.
Nowadays, cellphones have things like barometers or temperature sensors, so we can grab information from those, but the problem is that they are in our pockets, so we cannot just take the information from them and assume they are as good as those from a well-maintained sensor.

Do you also use high-performance computing to work on these tons of data?

We do. There are actually some very early projects where we use some quantum computing support for this, and how we can use that automate what is currently done with humans through machine learning and quantum computing.

Who are your most common users of micro-weather predictions?

I was on the phone yesterday with a company that produces a product that replenishes nitrogen for farming soil. I learned that rain depletes nitrogen, which is why the exact weather forecast is very important when they fertilize the soil. There is also a story from India that happened around four years ago: In a village in a countryside, one of the younger inhabitants of the village had a cell phone, opened up an AccuWeather forecast every day and wrote the weather forecast with a chalk in the middle of the street.

Then we have companies and different organizations and tens of millions of users that use this information to make everyday choices. We are deployed in Jaguar Land Rover cars in different scenarios and some other cases as well. The exposure of the AccuWeather brand is three to five hundred million users a day.

Why is it important for you to collaborate with other brands in City as a Lab environment and define the future of mobility?

The collaboration gives us a great opportunity to meet new organizations or new people in existing organizations. The other reason is that Europe is ahead of US in thinking about transportation and mobility issues. There is a great opportunity for us to play a role in the early days of autonomous driving and the more we talk with other companies at the summit, the more we can learn from them, while also spreading our knowledge.