Welcome to Reimagined Mobility podcast here with Meghan Cottam software developer for mobility technologies. Thanks for joining me, Meghan. Yeah, thanks for having me. It's your first podcast I just found out so it is. I have to blast you over the next 20-25 minutes and maybe just start out with you. Started with us in the ADAS space, software development for ADAS space tremendously growing field. Why eight outs and why software? Why did you ever choose software engineering?
Why did you ever choose ADAS as an area that interests you? Maybe give us a perspective there. Yeah. So I actually started in test and development for for ADAS. So I helped with validating the production vehicles or before production and making sure the software and calibration was was good for production and also helped to develop. And like define the operating parameters for upcoming ADAS
features and, and stuff like that. So yeah, you know most people don't that buy new cars don't really know how the ADAS systems work or haven't had experience with them before. So it's really important to design them and calibrate them so that they're intuitive, robust, smooth, you know, things that give the drivers confidence and and want them to leave the features on and disengage, disengage them because they're too maybe too annoying or to too aggressive.
So yeah, I worked doing that and defining the expectations for those features and would work with controllers, engineers and design engineers on how to meet those expectations. And so working in that a couple of years, I wanted to pivot and kind of get on the other side of that development. So to the at the controls, the zones. Yeah. So that's kind of what got me here. I had a lot of experience behind the wheel with ADAS and had an expectation of how it would behave.
So I had that background and came here to start designing controls and working on that. So what you described before about features people are not comfortable, they don't know it. They may not perform perfect, or at least to somebody's personal expectations or they turn it off. I mean, you described my wife. That's exactly what she does. I do like the adaptive cruise. I don't like this. I don't like that. Just turn it off and I. I turn everything on I can.
And your hands off want to see how it works at times and then put the hands back on, you know, cover my eyes. If I have a driver monitoring system to see how fast is it to pick these like, you got to stop this, you got to do this. So maybe you go back to that a little bit for a moment.
How far do you think we have to get still that really more people even even again, somebody like my wife who's driving, usually more modern vehicles that has some of those features in it, but still always seems to turn them off. How much further do we have to go as an industry to really get to a point where it's not a question anymore about turning it off, it's a question about why? Why not turn it on? Right?
You know what I'm saying? Yeah. Where do you see with your background are both again in helping define and make people more comfortable with these things. And now also on the software development side, you have somewhat of a unique perspective, in my opinion, from both ends. What do you think we still have to go or how far do we still have to go to to to get that confidence? And maybe it's just a matter of it's just time. Just people need to get used to it.
It's not a matter of the technology getting better or more reliable or, you know, instead of 99%, it works 99% of the time it has to work. 99.99999%. But share a little bit your perspective, how you see it personally. Yeah, you know, I think one of the biggest things that I that I've seen is people need to know, understand how it works. And so I think the big gap there is it might work well or under the right conditions or it's meeting, you know, the design parameters.
It was it was meant to but people don't know what it's what to expect from it. They don't know like the conditions in which, for example, AEB should be working. You know, I would get I would read customer feedback sometimes and they'd say, I try to run into a wall and it and it didn't stop serious. Yeah. And I like I had to hit the brakes and, you know, that's not what it was. It was designed for. So we're looking for classified objects for vehicles or pedestrians or cyclists.
It's not going to stop for a while. It's not going to stop for a random box, which it's just you need a lot of data to be able to do that. So I think one day we'll get there and maybe that's when people that's where the technology needs to be for people to to trust it or want it. But I think the biggest gap right now is you get in the car and you turn the button on and you don't intuitively know. You don't feel a difference necessarily, right, when you turn a button on, Right.
Yeah. So if you have adaptive cruise control on, you don't know what are the limitations of this feature, at least someone who hasn't been using it. And at least in my experience, most drivers don't actually open the owner's manual to it to get the idea of what the features capable of. So I think they either expect it to do a lot more than then it can do based on, you know, the sensor limitation or they they just don't trust it because they don't know what it's what it's capable of.
So I think there's just a big disconnect and teaching drivers or or newer customers on how to use their vehicles. And I think a lot of that has to do with like the user interface as well. You know, you have a new icon and sometimes we don't know what it means, what that means. Yeah, yeah. Interesting. I mean, you brought up a point about when you when you shared about the feedback of the customer. I try to drive it into a wall. I heard a story once.
I don't know if that's true or not, but I heard a story when Cruise Control first came out that a lady was in a minivan driving, put the cruise control on and thought that allows her now to get back and take care of a kid that was crying in the back. And I guess that happened on a highway. I luckily the nothing happened. But she found out later. I was later told that that's not what this is all about.
This is well before we've talked anything about Lane keep or a B or adaptive cruise or any other stuff that's out now. Right. So it's interesting. So from a software perspective, that's the way we develop software in the industry. Has that changed because of what you just explore and the perception is of people of of ADAS features or how they use it, or if we're really kind of we continue to push forward in the journey to get to level three, level four than ultimately level five.
Or have we made some adjustments as it relates to not necessarily the coding language that we use, but features and how we're maybe implementing? And again, you mentioned to each of my right, are we more conscientious about that than maybe we were two years ago where we say, well, we put an icon there, who cares? It's about the AB function, not about the making of the driver aware of it. Has software changed and how we develop features because of that or not?
Like the core software I would say has been probably the same, but definitely like the the user side, that type of software and design has changed. You see more, more feedback from the vehicle. I think before was kind of just an add on and now, you know, the the cockpit is designed with this idea that the vehicle will give you will give you feedback.
And in terms of the actual controls, I think, you know, making the responses more intuitive, make the driver feel more comfortable with how the vehicle is performing. And I think you'll see a lot more of that with with features that include machine learning or things where they can actually learn how the driver wants to drive the vehicle. And that's where you'll see the software, really the core software change.
And in terms of how aggressive should I accelerate, how how much distance do you like to keep to the vehicle in front of you? And you can learn that through, you know, just daily driving and and then when they they press cruise control, it will be tailored to them. And they don't have to go through a screen with a lot of settings to tailor it to them and then try them out. And so I think over time, yes, we will see it be more tailored and more intuitive for drivers going forward.
Would you say then then with with that answer you just give and I assume then that maybe three or four years ago the AI portion of ADAS was nowhere near as important as it is today, and it was more a function of the focus on more of of accuracy, of object detection, object avoidance. Lane Keep all of the all of those features for level two two plus.
Now going into level three, would you say then AI really has gained a tremendous level of higher appreciation now because of what we've learned over the years as it relates to adaptation? Most definitely. I think you you would have, you know, a lot of you can see it with the change of like the cluster design as well from going from an analog to more digital because you definitely have more freedom to give that feedback to the driver.
I think in newer vehicles I see more and more the ADA screen keeps getting bigger and bigger and giving more feedback. And and so I think that's like the biggest place you see it. And then you have other companies adding lights and the steering wheel or on the side or different interfaces. Yeah, no lights or voice or vibration or whatever it might be, Right? Yeah, exactly. Interesting. So let's maybe pivot a little bit because you also you also do software for electrification,
BMS specifically. Yes. Tell me a little bit, is there a big difference for you as an engineer working on a BMS software versus working on an ADAS? The reason I'm asking is BMS, let's say not going to argue. What's more important to ADAS or the BMS, but the BMS certainly not as user facing right as we just talked about ADAS is does that make it more challenging for you?
Does that make it more I need to be, you know, more careful when I do something that's customer facing versus something that's not. Does it make a difference for you? Does it not make it does it make it more attractive? Share a little bit the differences. When you look at between ADAS development and, let's say, BMS development? Good question. Yeah. So with like you said, with with BMS it's definitely more behind the scenes software and it's important for, for safety and longevity
of the battery and powertrain. But I personally think ADAS software, there's a little bit more pressure there because if you're if you get it a little wrong, it's very obvious since everybody knows and. A little wrong is very wrong. Yeah, you don't have that in BMS I mean, if you calculate a certain SOX algorithm that's it is. It's definitely important. Of course I think you just in terms of oh, I guess it could be more important because it's it's harder to find right away.
But yeah, BMS is definitely challenging as well and in in terms of the actual software that goes in it, it's it's it's a lot more than I think people realize what a BMS. Yeah yeah okay. But it has been I think around or more defined whereas I think in ADAS you kind of have a little bit more liberty and your strategies and everybody does it a little differently and there's like pros and cons to all those type of ways.
Yeah. Do you have you kind of mentioned more pressure, maybe ride Is there more pressure? Because not only is it customer facing, but also it's you can cause a crash, whereas maybe with a BMS let's say for a moment the worst that could happen or let's say what could happen is I run out of juice sooner than what the display told me because of the wrong calculation with it. But it's not necessarily a a matter of of of life here. Potentially with one side you could crash.
On the other side you just stranded on the side of the road. As a software developer, do you think about these things when you write software, is that you kind of have to you think about it, but you have to put this aside, otherwise you get hesitant and you don't trust yourself? You know, what I'm trying to say is there something like that? Because I always wondered, I've never done software for ADAS is not not my area of expertise.
But I always kind of felt like with some people that I work with, they always like, well, this is a safety critical thing and I'm really concerned about this. Are at a work on a on an infotainment system as if a button doesn't work, a software doesn't work. Okay. But if this feature doesn't work, it potentially can cause a crash where somebody could get injured. This is something you guys think about or you think about when he writes offline.
I mean, it's definitely you definitely want to write safe software. Sure. I wouldn't say that it's it's on my mind the whole time. There's so many tests, so much so much tests you have to do and you know where to develop the software way to develop the feature so you can develop it in the safest way possible.
You know, you even before you do the controls, you define the feature, you define the limitations, the accelerations, the max steering torques and you define those in a way that safe or at least can be overridden by the driver. So the driver is always in control of the vehicle. And by doing that, I think the software, there's not as much question as to whether it's safe or not. Of course, you still have to do a lot of testing, do simulation, do it.
Mil, Sil, all the all the types of testing you can to to make sure it's behaving within that design, the design specifications. Current Yeah, I think the pressure that goes on, on designing the feature for so that's you know that's fortunate for, for me now.
Yeah. So you bring up a very good point there's a lot of obviously processes as an entire tool chain from the definition all the way to the definition of a ADAS feature, all the way to the testing and validation of one within that tool chain. What is to you the most valuable tool or methodology Let's see. I don't know. You know, I'm thinking about functional safety.
I'm thinking about a SiL test or a HiL test or maybe a vehicle in the loop test or lots of other things in between that software people are using. What to you are the most valuable tools that you use? Good question. I would say for me personally, I would say it's like model in the loop or SiL testing because you can think your your software is doing really well.
You design and maybe it's passing on your small test, but once you, you, you model the car and run it through a simulation and you see it's actually not stopping for, you know, something that you didn't even think of before. You can really catch a lot of issues there and make sure your your software is passing all the the software requirements before it even goes into a vehicle.
And then once you're in the vehicle, of course, that's that's very important to make sure it's working well with all the other features and dynamically. But you can get, I think, so much done in simulation nowadays that really cuts a lot of time from the development process. And are you doing it on a consistent, almost like agile loop on Sprint? So you're right, software for three weeks and you test it for two weeks or are you doing it concurrently with the development?
You also run a model in the background, so to speak, or in parallel to you developing share a little bit. And how you doing them. Yeah, so we usually do model updates and then test that like on a lower level and make sure the software we were implementing or changing is running the way we we expect it to.
And then using like the Agile process will do integrate everybody's software changes and do a simulation and make and to ensure that everybody is functioning together appropriately and then catch a lot of bugs or see your work actually working in a very short period of time. So it's very satisfying. Okay, so cybersecurity has become a big part of automotive space. I would say again, I think it was big. I would say seven or eight years ago, everybody talked about it.
Then I felt like it kind of got not away, but it went dormant. Now it's really back again and very, very important. Clearly, we know why because everything gets connected. Everything is interconnected. The vehicle gets more automated, right? We go to a higher level of autonomy, level two plus three now pushing into four or not on a widespread deployment at least. So cybersecurity threats or cybersecurity is a bigger issue.
How much do you, when you write software, have changed things or need to change things now from an approach or from a coding perspective with cybersecurity, with functional safety as well? ISO 262016 with functional safety and cyber security now more on the forefront than it may have ever been in history. I think in my experience, I've been doing controls for like a little under two years. So it's, I think, been on the on the forefront for like the whole time
I've been doing controls. And I think the way things are modeled now, you know, you have functional safety. The autos are method and model based systems approach where I think cybersecurity isn't as much of an afterthought anymore. You kind of make it into like the the controls process. So on a model level, it can be implemented there. So it's not like it was before. I would say more of an afterthought that you would go back and try to make it, you know, a safer, safer software.
So I think that is the is where you'll see the biggest change is mainly in the the controls process. It's kind of embedded in and from the beginning. Okay. Okay. So for you, it hasn't been much of a change because you're saying over the last two years that has always been from the get go. Yeah. Came from that. Okay, that's that's fair. Maybe last question for you.
I usually ask people that when we specifically talk about technologies, would you would you go back into software engineering again if you could rethink when you graduated from high school and what am I going to do?
Or would you say no now that what I know now, I would probably go into different areas or you say, Yes, I would, but instead of going to work, I maybe get my, I don't know, PHD or going to AI or share a little bit like that for some of the people are listening or watching and maybe try to decide what career path to take. Yeah, I mean, I would I would do so I would do that especially in mobility.
I think the experience you get in a in a new vehicle or in mobility in general is so tied to the quality of the software in that in that vehicle. So there's, I think, so much power in the software you design. You can have bad software that kind of ruins the whole experience. So I, I really enjoy software and controls and would definitely choose that again because you just even in the future, I think we're going to see that more and more too.
So I think with time, software is going to become even more important and you really do have the power to control that experience for, for people. MM Yeah. I'm not going to ask if software is more important than hardware because I know the answer. I feel very good. Thank you again for your time. Yeah, Thank you so again. Thanks for listening to Reimagine Mobility podcast. If you like some episode, please subscribe and tell a friend.