Lessons for the IoT from 8 World Championships | IoT Forum | McLaren Technologies

Caroline Hargrove is Technical Director at McLaren Technologies where she is responsible for the commercial application of many advanced technologies used in automotive, elite sport and medical devices. She spoke at IoT Forum 2015 about some of the ways that McLaren has applied the lessons from their Formula One Racing programme – true ‘connected cars’ – to problems in the real world.

Slides, Video, Transcript Below

Slides from Caroline Hargrove, McLaren Technologies talk at IoT Forum Here



Transcript Below

The BLN is now Business of Software

Business of Software runs conferences for people that build great software businesses and products. To access talks online, hear about new events, contact speakers and stay in touch with great ideas, share your email address with us.

We'll NEVER sell your email. Unsubscribe anytime.

Caroline Hargrove, McLaren: Because I’m the last speaker this morning, I’ve gotten permission to show you a video. It’s a corporate video, but there are a lot of cars in there. I thought I would start with that to wake you up before lunch… [Applause]

Mark Littlewood: [from audience] The first corporate video that I really enjoyed! [Laughter]

Caroline Hargrove, McLaren: I felt really bad when I said I wanted to show a corporate video, but I was like, “No, it’s going to be really different.” The reason I wanted to show that video is because people hear McLaren, they think Formula One.

Now, there’s definitely we’ve got something in common, all the McLaren company. We are data rich. My background is Formula One, but then I moved to starting McLaren Applied Technologies seven years ago. We were three people with the job of exploiting that technology to other markets.

I wanted to show that we also do nice automotive cars as well. We are three different companies, even though we are in the same building and we share the same DNA.

What I wanted to show with a little bit of the F1 story

We’re one edge of the spectrum for the Internet of Things. I often think that’s not what we do. We just process a lot of data. We’re in a data rich environment. We have data cycles.

The data cycle in Formula One is very fast. Of course, there’s the cycle of producing the car in the first place. Then there’s every race. We go through this. We simulate. We have a simulator. We run pre-race analysis. We monitor what’s happening. We debrief after a race. We design. We change components.

Just as a quick fact, for example, we will collect up to 500 sensors on the car that is related to kilohertz back to the factory. If we change gears at one point, we’ll bringing it up to 10,000 hertz. We beam that all the way to working where we make some decisions during a race.

We have facilities like this, where we put the driver in the loop. That’s something I worked on for years, simulator. Again we are using the data a lot to try and improve the car. When the car’s start not too good, we can’t do miracles in the simulator, but we do try to tune it as well as we can for each race, per driver. We put a driver there. Each time the driver will try as we control the different variables that you can’t control when you are at a track. That’s really useful.

We also will do loads and loads of simulation in the background. Old slide, I’m not revealing anything that we can’t. This is obviously stuff that happens in the background. Then that’s consumed – and, again that’s an old slide, because I’m not allowed to show we are using the latest ones. Just to show an example of consuming a lot of data very fast. What we’re…The message I want to convey today is the end of the spectrum that we’re at is to try to get insight from a lot of data. Insight is actionable and is also forward-looking.

The dynamic statement that we were saying earlier, that’s what we are about. We are trying not just to look at the now, but what should we be doing in the future? That’s just an example in Formula One. In fact, it’s changed quite a bit. The idea is each track is very different. How do you know how far you are with a competitor? We look at it as a circle.

When you are looking at a circle, each track is very similar. At any one time you can see if I pit, I’m going to be coming out at that point. So if there is a safety car happening now, and I’m under stress, I can just look up my screen and decide whether or not I should pit or I should call a driver to pit.

We are trying to give these decision support tools to our race engineers at the track. We also run lots and lots of simulation in the background. As data come in through the race, we rerun all this simulation. We’ll rerun 100,000 of these Monte Carlo simulation each time a new data point comes in.

We can do that. We’ve been doing this for years.

What we’ve been interested in doing is can we apply this in other sectors?

The first one I was putting up there, is the data centres because that is very relevant. Somebody asked earlier, where is all the data going? Well it will be going via data process and data centres. They currently consume 2 percent of the world’s electricity and it’s growing. They are often located in deserts, like in Arizona for example.

One of the work we’ve done in this area is to see whether we can with them, analyse a pattern of use. Things that it can’t change is server to server and they consume electricity. What it can change is if they can look at the pattern of use of those servers, as well as targeting the cooling – and we also look at the HVAC design of cooling for individual units. So you’re not cooling an entire place, but you are cooling the bits that need cooling, hence reducing your costs.

That’s the kind of area we are involved in. We work with other businesses, so we don’t have this idea of privacy and access to data issue because we work with their own data normally. What we do a lot is algorithms, analytics design and getting insight and giving that back to the users.

We’ve seen earlier, ironically a car platform showing something else. I could have shown my car platform, but…We work on production line, for example. This is another example where it’s a data rich environment, as you heard before. There’s lots of machines that actually will contribute some data back into the RMP system or manufacturing system.

Ironically for us coming into this market, we realized that these are not dynamic systems. If something goes wrong – and believe me, they go wrong in these kind of environments much more than I ever thought. You do your planning. You have a two-week planning timeline. Then something goes wrong. So the operators, they are doing what they can to decide which pill should be done at which time, because they might be going for different markets. They all have different variations of that pill, etc. Why not just rerun your planning tool, whilst you’re getting any information of what went wrong and plan maintenance, when best to do your changeover, etc?

So the information is often there and not enough is done with it. Exactly from what a lot of speakers have said, if you don’t get a huge value out of your data, you’re not really exercising it enough. In fact, we’ve been in plants where we’ve said, “Have you got the data?” and they said, “Oh yeah, but we turned it off. There’s no point collecting it. We’re not doing anything about it.”

We said well turn it back on and then we’ll come back. Without the data, there’s not much we can do. [Laughter]

Another example – drilling.

I’m just giving examples of the kinds of things that have been…We’ve found exciting and we’ve been completely outside. We don’t have knowledge, detailed knowledge of course, as you’d imagine on drilling. But we worked with experts that have that knowledge. What we try to do is get them to have more use of their data.

Here the big issue was in drilling, the driller. It’s like our Formula One driver and they are really cool and they’re really good at what they are doing, but they are in short supply. They are getting older and training them is quite expensive, etc. What we wanted to try to do is to give them this – if you think about it as a second screen – is more help in decision making on the spot that is forward-looking.

So they have geologists that looks at data and produce models. What they don’t do again is adhere to a plan for the drilling. They start the plan and things go wrong because they do. Geology is not known exactly, etc. How do you do something about it that is slightly more dynamic, real time so that the driller knows all the different factors that he needs to know to do the optimization on, has some help in making that decision.

For example, drilling fast is good because drilling there in that North Sea environment is about £1 million a day. If you go too fast, the well integrity can be compromised and safety measures can be compromised as well. Any one time, when you have to make a decision, why not support that with the data we are collecting?

There’s tons of data being collected there and they’ve got tons of historical data as well that you can lean on to see whether the models make sense in that particular area. You combine the two and you do some type of predictive modelling. What are the odds that the same thing will happen again once I have this particular issue?

It’s currently in use in two wells, so we are doing OK.

We have done a similar system for air traffic control.

That’s not in use right now. We’d like it to be, but it’s not quite in use yet. We’ve worked with Heathrow on helping air traffic controllers. I’m like a broken record, saying the same thing, but they do their planning at the beginning of the day and then they run through the day with the air traffic controllers managing any issues.

Now, of course there will be issues. Heathrow’s full as you might know. I know I’m under the flight path. I don’t want another runway. I’d rather they run more efficiently. The idea is if there is something that’s going to go wrong, especially when you have an inkling that it might go wrong, because the weather, there’s a weather warning on something. You know it’s coming. Why not plan for that?

What we’ve done is a dashboard, which I haven’t included in time. But it’s a dashboard that lets the air traffic controller say, “What if I do this? What will it do to my KPIs, which are on-time departure or on-time arrivals or number of missed connections and all of that? It can be a number of them, too many to do in your head.

But computers can do that really easily. We’ve got the data, why not do that? Allowing them to look at ahead of time the impact of making a decision. We’re not making the decision for them. We are providing information to help them. So it’s decision support, not decision automation.

We’ve done also on the design side.

We are doing this work with GSK. The work with GSK is mainly on clinical trials, but these ideas, as we heard this morning, are to give some intelligence to the devices so we know how our patients are using them. Depending on how they are using it is the drug effective? If they are not using it correctly and you are expecting the patient to do something and they are doing something else, there’s a huge implication.

If you haven’t got a clue how they are using it, how do you know how effective your drug is or will be for a particular type of population? So we’ve done some of that work with them.

We’re also looking…

The last few slides are looking more into the future.

What do we want to do? Where do we think this Internet of Things can really help us? One of the things we started doing with surgeons is monitor them as they are moving in their training. One of the questions that came to use is lots of bright medics want to be surgeons. Not every medic would be a good surgeon. [Laughter]

You might only want to be operated by the people who have dexterity. Dexterity is never really looked at when…grades and how well they have done in their exams is what they are normally judged on. The idea is we got lots of commissions of tasks they are doing when they are learning and using machine learning. Not very difficult at all. We’re classifying how people are in overtime and looking at a longitudinal. That kind of problem can be done more widely.

We’re very interested in the more personal care. We’ve heard more people mention before. We’re interested more in mainly the data analytics side. So we’re conscious that a lot of data is being collected. I heard that at the privacy discussions this morning. That’s what’s really interesting is putting a patient with patient-like you type data behind. Getting you through your care journey more efficiently and possibly involving you more.

A lot of people are talking about this. It’s not really done in practice. We would like to get involved in that so we are starting baby steps in this area.

Finally, my last slide was the other thing we are interested in is the environment.

It’s all connected to the Internet of Things. We don’t make sensors. We use them. We make sensors only for racing cars. That’s not we use here.

The idea in this context is can we learn more…Can we have more intelligent environments that will help both the patient and the medics in the operating theatres be more efficient or have tools at their fingertips. Things run better, information collected better, all of that we are talking about.

We are almost putting an envelope around that. Can we add value by adding more and more of that functionality from the data back to somebody making decisions about it?

So that’s it for me today. Thank you. [Applause]

The BLN is now Business of Software

Business of Software runs conferences for people that build great software businesses and products. To access talks online, hear about new events, contact speakers and stay in touch with great ideas, share your email address with us.

We'll NEVER sell your email. Unsubscribe anytime.

Mark Littlewood: Do you have a couple of quick questions?

Caroline Hargrove, McLaren: Yes, please.

Audience Question: You are saying you are applying it to various specialist domains. You’re the generalist, but you are working with people who are the specialists in those domains. My question is are you working with those in the customer – those specialists – or are they partners or are you bringing them into McLaren? Are you building these up these practices in certain domains?

Caroline Hargrove, McLaren: That’s right. Health is one that we really want to go into. We are still trying to find where our best value add would be in this area, whereas in other areas we’ve done some more work . We’ve done some work in say, telemonitoring in terms of clinical trials. We haven’t done enough in closing the loop with users in that area.

Otherwise, in energy market, we are going in with partners, getting some traction. Eventually what we want to do is build our IP, what Alex was talking about earlier. Not necessarily with the same partner, but we are rarely going to go out to market directly. That’s not what we do. We go out via people who are established in that area or are willing to be breaking ground in that area.

Mark Littlewood: Final question.

Audience Question: One of the big justification or motive for spending the massive, massive amount of money on motorsport is the trickle down technology into helping sort of everyday lives of the populace. Do you have an example of such technology that McLaren’s worked on that have trickled down to affecting the everyday lives?

Caroline Hargrove, McLaren: Yeah. It’s a good question and I think we did it more before. The rules and regulations have been so constraining, that we haven’t done that much of this trickle down. Certainly the carbon fibre revolution Formula One have been very big on that.

Even our road car has been the first carbon fibre chassis as a single piece chassis for any of the production cars. I’d say something recently is the hybrids. Our Formula One cars are the most proven, they are hybrids essentially. They have a very efficient energy recovery system in a battery. So we’ve gone from 150 kilos of fuel to 100 in one year. Cars go the same speed, so that kind of technology goes down to more efficient use of fuel, that type of thing.

Less directly what we are doing there in terms of design, but from a data perspective what we’ve learned to analyse it very, very quickly and be forward-looking and doing lots of data simulation and running those in a cloud for example, that’s things that Formula One has been doing for a while and we are applying elsewhere. So less in your face, but more in the back.

Thank you.

The BLN is now Business of Software

Business of Software runs conferences for people that build great software businesses and products. To access talks online, hear about new events, contact speakers and stay in touch with great ideas, share your email address with us.

We'll NEVER sell your email. Unsubscribe anytime.