James Thomson: CAV detective

James Thomson leads a team of engineers to dissect connected and automated vehicles, and driver assist technology, find out how it works, and how well it works, with a view to assessing risk.

Where are you working now James, and what do you do?

I’m working for Insurance Australian Group, a general insurer based in Sydney. I’m the Engineering Manager at IAG Research Centre, a physical test laboratory where we perform physical testing, insurance-related risk assessments of products as well as vehicle parts and repair research.

It makes me wonder, insurance, if nothing else, depends on a huge amount of data … how is it you’re balancing research into new technology with, in terms of the insurance business, quite a small amount of data right now?

I think that’s why IAG’s research is perfectly poised to do that, because it has been around for about 25 years. And the backbone of the centre has been formed by engineers, mechanics, and other mechanical and scientific-based staff. We’ve always developed data on an empirical basis, going out and finding things out from first principles.

We’ve either pulled it apart, stuck it together, smashed it, crashed it or weighed it. And with autonomous vehicles and no-one being experts other than the people who build them and their contractors. It’s all very locked away behind closed doors. We have an understanding of how to start unlocking that, how to unpack that, put a name on it, know how to catalogue it, and how to assess its risk.

It’s what we’re used to doing, and is why we’re in a good position. We literally get hold of the technology and we apply our expertise in the space to work out what’s under the hood.

Would it be a top secret for you to tell me the sort of stuff you’re doing on connected cars at the moment?

No, nothing top secret. I’d say the most prestigious and publicly-available project we’re doing is through iMOVE. We’re collaborating with the University of New South Wales to build our own autonomous vehicle – a full-size passenger vehicle. We’re purchasing off-the-shelf equipment, sensors, radar, lidar and camera-based systems which you can purchase around about $20,000 – $30,000 worth of equipment, and we’re going to be installing it in the vehicle using an also commercially-available piece of software.

We’re building this car not only to learn exactly how autonomous vehicles operate, but also to understand the kind of algorithms that would fit inside the vehicle, and when they choose what operations to perform. That’s where the university comes in with its expertise of writing code for autonomous vehicles.

It’s a project in which IAG is thrilled to be involved, working collaboratively with government, industry and education. That’s where iMOVE, and UNSW comes in, but it also, I suppose in the long run, will show the public that we’re putting our money where our mouth is. We’re going to have to insure these things one day so the way we learn about things is by empirical methods. We pull things apart and we build them, and we learn by first principle so we know exactly what we’re talking about.

That’s our biggest project right now. IAG does have other exciting projects underway. I’m not working on them and I believe they are confidential, but there are several other projects that I can think of. All of these are very important to the company. We’re not crazy, but we certainly are pioneers in the Australian insurance field.

I look forward to hearing about those other projects. And prior to IAG what was your path there in terms of study and maybe other work?

I recently turned 38. Of those 38 years I’ve spent 16 of them at IAG. I have been at the research centre since before I even graduated university. I studied aerospace engineering at the University of New South Wales, and in my thesis year I approached IAG, having known about its research centre for some time.

I found out about it in the NRMA’s Open Road magazine, cold called and asked for some work experience. I was successful, stayed for a year pro bono, and I ended up staying on. In that time I’ve worked my way up from the mail room so to speak as a junior engineer. I’m now the manager of research and development at the research centre.

Well there you’ve already answered my next question which is how did you gravitate towards working there. Was there anything that pre-dated study?

I’d say it was not only my background but also the opportunity from the research centre, taking on a junior engineer. It may have been what I studied at university but honestly I’m interested in all forms of engineering, anything that ticks or moves interest me. If I hadn’t have done that, I would have done something similar I would say. Either that or something to do with the human brain, as I find psychology and psychiatry very interesting as well.

I learnt everything from the ground up again when I started here. If you’ve ever been to university what you learn can be very different from how you apply it. How you work with people and developing a method for lifelong learning is more important. Which is why I think the CRC is such a good representation of what working life in general is. You might be good at something individually, let’s say IAG, but you really have to work with other people who have different skills in order to get the very big job done, such as connected and autonomous vehicles.

Veering away from you for a minute, I’d like to throw a hypothetical at you. You have been provided a large bucket of money and been told that you can fix a problem in your area with a reasonable time frame but like I said, on a very large budget, what would you like to attack?

For the last 7 years or so, IAG has been assessing AEB (Autonomous Emergency Braking) systems, a precursor for ADAS systems (Advanced Driverless Assistant Systems). These are the technologies that will stop a car automatically without driver intervention when it detects a potential front-end collision. We’ve always been of the mind that manufacturers will sell or push a technology out there, for the main reason of course being sales. It’s also the trend of technology and the desires of the public that tend to drive these things at times, and some of their abilities tend to get lost in the maelstrom of vehicle ratings and public perception.

We always like to know to what exact degree do these systems work, so we develop our own targets, and we have a slightly different set of criteria to see how well they work under different conditions. If I had enough money, I wouldn’t mind testing every single vehicle under a very different set of criteria, because we know that the tests are calibrated to a certain set of rules, they are calibrated to a certain shape of vehicle. Those scenarios are very rarely seen on the road outside of perfect test conditions.

We’ve also driven the cars. A lot. The vehicles that we test, we take them out on the road and do subjective-style testings, in order to be familiar with the vehicles and technology. They invariably behave very differently under very different scenarios, they’re implemented differently, they brake differently or they swerve differently, and what have you. It’d be fantastic to get a really in-depth matrix of how all of these vehicles behave just so we know how safe they are, and how they operate in different conditions.

Down the line, it would be good to continue that, I suppose in the internals of the vehicle to start looking at how these vehicles make decisions in autonomous situations. Which path to choose, who to stop for, what to swerve for, etc. That interests us enormously. It also interests me from a professional and an personal perspective, because they’re trying to replicate how a human thinks, which borders on artificial intelligence.

All of this, and the fact that they’re being driven by very large, multi-national companies to sell products, but not necessarily with the intention of saving lives first and foremost. They’re just the technologies that need to be in a car in order for someone to consider them as good as the competitors, to be considered as an alternative product to purchase.

And all of a sudden, people are putting a lot of trust in these systems, without knowing how well they really work. We’re seeing most findings coming out of the international researchers like IIHA and it’s mirroring our own experience that under duress sometimes these vehicles don’t work at all. Or they behave erratically, and yet they’re coming at us so fast, and we’re supposed to buy them and trust them.

Really put them through their paces differently from say ANCAP or Euro NCAP, under a different set of conditions. That’s what I would like to do!

So it’s both a more realistic testing environment plus a way of earning trust and educating the public about what they’re looking to buy?

Yes. I think it’s an intrinsic part of educating the public on what these things are for, what they’re capable of, and what you really shouldn’t expect from them. They’re been touted as being the next best thing but there are … there’s going to be pitfalls, absolutely.

Indeed. Alright now part two of the hypothetical is this time, shorter time frame, and a small budget. What smaller problem would you like to attack?

If there was some way of setting up the things I’ve just spoken of virtually. Maybe running some simulations of the same kind of tests. Or at least getting a working example of how this kind of testing can be done in a different manner. Not necessarily in a physical manner but in a virtual testing environment, or something of that nature.

I imagine, yes, that would be cheaper. Do you think that would be as complete, as trustworthy as doing it in a real-world environment?

For some parts of the testing, no, because you wouldn’t be able to properly replicate a driver, and of course vulnerable road users, other drivers, and how they might all react.

Artificial testing, virtual testing, will get you to a certain point. Then there’d be a great gulf between that and the actual truth, which is I suppose why we’re seeing unheard of, exceptional circumstances, like the case of that poor woman in the USA, killed by an autonomous vehicle while crossing the road with a bicycle. That was literally something that couldn’t be accounted for without a human driver.

Right so perhaps your smaller budget idea might be a … just a doorway into something more.

Yes. I think so. So maybe just a few of the aspects to build a little bit more trust but with the rider that … we haven’t completely tested this. We don’t know what it’s really like in the real world.

I’m not a pessimist by any sense of the word. I’m simply an engineer and I look at the numbers. You’ve probably heard that kind of stuff before, but I can’t trust a system if I’ve seen it behave erratically at least one time. And if we’re not telling the public about it, there’s got to be education. If we’re going to try and have everyone on the roads using these vehicles, or having them integrate with people who are not using the vehicles, I’m just expecting a mess. It’s something that we need to research and elucidate for the public, or our customers.

That’s something we’ve always done at IAG, we make media releases, we release our findings, we make YouTube videos, what have you. We believe education is an important part of it. Not just as a shared value, touchy-feely good kind of thing … it is a philosophical thing, it is altruism, but I think the only way this kind of technology will be fully realised and adopted is by looking at all angles. The technology, the training, the education.

Back to you. Of everything you’ve done so far, and this is at IAG I guess, given you’ve spent your entire career there, what is it so far that you’re most proud of?

It’s actually the position of leading a team. Working on the autonomous vehicle is a phenomenal opportunity for me, and it is the next stage in my career given that it’s collaboration as well, and that I’m heading it up, I’m project lead. But there’s something about taking people on the journey with you and teaching them how to then go on their own journey down the track.

I’ve got a younger team than myself, at least a decade younger, and we’ve got some more people coming on board, and I get a lot of satisfaction right up there on the Maslow’s hierarchy of needs that really strikes me deep inside, Scott, that it’s the team aspects, it’s the learning and achieving something as a unit rather than just one person. I’ve been telling people the last six months or so that the advent of this technology, autonomous and connected vehicles, is going to have a bigger impact on the world than I think locomotives and the railway system did several hundred years ago. It’s going to change every aspect of life, it’s got to be done together.

Coming back to it, what am I most proud of? The fact that I have an effective team who are full of intelligent, hardworking, and independent driven people.

Good answer. If you weren’t doing exactly what you’re doing now but you were still working in the mobility space, what have you not done that you might like to do?

This actually brings me back to iMOVE, and the way in which it operates as a cooperative research centre. It’s brought together a lot of people from different businesses, different parts of the community and industry, which I’ve never considered because I’ve never needed to, that’s not part of my job.

Speaking to some of the people from councils or awareness groups or even charities in a peripheral sense … like them I have a heart, and I do care a lot about people.

So that comes through my work and to be able to work in a field that focuses on that primarily, I think would really be something I’d like to explore, working for advocacy, for awareness, education.

I realise I like managing people and teaching people as well, I get a lot from that. It would be good to be able to get this message out, to educate people, run a school for people on how to use the vehicles, something along those lines. Not selling the technology, repairing it, designing it, or insuring it. Rather the people side of these technologies.

As you said earlier, this will be a massive upheaval in everything we do. It’s going to affect and touch so many areas of our lives.

Yes, maybe there’s going to be different business opportunities down the line. Businesses that will develop solely due to the advent of autonomous vehicles, because they didn’t exist before. I don’t know what they might be. For me it could be specific to training or education or … teaching at uni or something about the application of ergonomics and biodynamics and people, the way they interface with vehicles, I’d find that fascinating. Probably move away from the hard engineering and more into the people side and the interface between them.

That, or advocacy for people. I think they have been forgotten about but I think it’s very much on the minds of more than a few people right now. That has been brought to my mind recently, and it really struck a chord with me. So that has informed my answers on this.

And last question. What, technology-wise, are you most looking forward to in the next say, three to five years?

That’s a tricky one. I was driving a brand-new Volvo today, which has some very advanced autonomous technologies inside it. And the way that they’ve been implemented, the particular switch gear and whatnot, really helps integrate well with the driver. The way the vehicle gives the information to the driver there was a very effective HUD unit (heads-up display).

The way the vehicle spoke to the driver, the vibrations coming through the steering wheel, which are called haptics, that’s been quite well done. So the technology used to improve the human machine interface in autonomous vehicles, because we know the handover takeover process is a tricky one. People’s attention span and ability to maintain control of the vehicle after the autonomous portions have turned off using is a sticking point, so I’m looking forward to seeing how technology evolves to prevent that from occurring, or at least shortening the amount of time someone’s not aware completely of their control and surroundings.