ADASDevPro Logo
Product was added to your cart

Cart

A Brief Intro

(Dilbert, BMW Cinema Screens, Google, Disney, Dream Sellers, Tea)

Hello dear reader and welcome aboard.

In this series of blogs entitled, as I hope you noticed, “How to Design an Autonomous Car”, the implication being that you didn’t just follow the pretty pictures, I am going to go on a journey, a potentially introspective but hopefully not too self-indulgent journey, but one that I invite you to join me on regardless.  My aim is going to be to document all that I can in regard to how one should design, at a ‘System’ level, an Advanced Driver Assistance feature robustly, the foundation of today’s autonomous vision; I will define what I mean by ‘System’ later. I’ll be touching upon concept and requirements capturing, through to failure modes and functional safety, or in other words, all the stuff at the top of the V.

Autonomous Car

I am however going to skip over previous engineering process failures as, although interesting and worthy of a blog in their own right, there are far too many to mention here, and so I will stick simply to what works.  I am also not planning to get too deep into ‘Functional’ design.  Again I will define ‘Functional’ later on, but for now and for the foreseeable future, or at least until the avoidance of the ‘Functional’ begins to interrupt the flow of my ‘System’ level monolog, I’m going to steer clear.

I am also going to assume that you are reading this blog, which will essentially be a brain dump of all that I can remember of the last ten years, it’s fading fast, of designing, implementing and producing ADAS features, because you’re looking for inspiration for a personal autonomous car project, a guiding hand to get you to where your boss and perhaps your customer wants you to be, or you just got lucky and found some free online training.  Well hopefully you’ve come to the right place whatever your vice, but alas only time, and the number of publicly posted comments to the contrary, will tell.

But before I start, although as a society we seem to be inclined to accept the first thing that Google presents us with as the truth and thus the validation or rejection of our own ideas and thoughts, as crazy as that sounds, I suppose I should identify myself in case you are one of those individuals that require a slightly more robust method of authentication and acceptance.

My name is Dirk Parker, and I am an Engineer; that sounds like a Dilbert sketch I read once! Over the last twenty few years I have been an aerospace software engineer and software manager, an automotive software manager, operations manager, consultant department manager, lead engineer, chief engineer and more recently the director of the company ADASDevPro, on whose website you’re currently sitting.

Over the last 10 years I have also developed a number of ADAS features for a number of large European OEMs, many I have seen through the whole life cycle from requirements to full scale production.

The list includes:

You will notice that I have not mentioned any feature above what the SAE J3016 considers level 2, and with good reason.  Although I am a big fan of the idea of the autonomous car, (it is a necessary goal given the exponential rate at which car ownership and subsequently traffic congestion is increasing), despite the media hype, I do not expect to be hailing a driverless road vehicle in my lifetime, and I hope not to die any time soon.  Willingly putting my offspring into a car, waving as they disappear off down the road, hearing them cheer as they select the latest movie from the Disney channel on their BMW designed wrap around smart cinema screen, while the ‘car’ attempts to navigate the route, something that millions of people with a life time of driving experience, are still incapable of doing without hitting something, whilst also being child proof, won’t be happening any time soon.

However, if you are aiming to conquer the world of the autonomous car, I wish you the best of luck and these blogs, and thus I hope your subsequent understanding of Systems Engineering and ADAS development, will most definitely put you on the right track, but you are undoubtedly going to find that in today’s world, it is simply not possible to achieve your autonomous goal safely, en masse, at least not to the extent that the number of people killed, or the number of people suing the OEMs, is outweighed by the benefit of the autonomous car, and it won’t be for a very long time.  For the record, it is worth saying here that, as great and as groundbreaking as it may sound, recording millions of miles of synchronised video and CAN data is only great and groundbreaking, if the company recording all of the data has enough time and of course enough people to trawl through the petabytes of data looking for all of the instances of when it went wrong and then, working out why it went wrong, fixing it, testing it and formally releasing the update.  Even if they can install the bug fix (if they’re lucky) or redesign (if they’re not), Over The Air, it requires a vast amount of potentially prohibitively expensive energy to resolve every issue quickly, no matter the size of the OEM.  What about the benefits of Automated Testing and Shadow Driving I hear you cry?  Maybe I will cover my thoughts on that one in the future too.

And let us not forget that the jury is still out on the question of whether we should ever remove the human element from the control loop of such a massive collective machine that is the global road network; there are still too many missing pieces to the puzzle, the law being one of them.  As of today, it is still not clear how a Level 3 feature, functional enough and unobtrusive enough, to be worth paying for, can operate without a significant risk of litigation.  Asking a driver to take control of a vehicle after a situation has arisen, while the car was in control, in which a crash is imminent, forcing the driver to have to think fast, which will be difficult as the car has allowed the driver to physically and therefore mentally disengage by the very act of allowing them to take their hands off the wheel, (I for one normally last about 10 minutes as a passenger before falling asleep and snoring my face off), and act even faster, assuming they choose to, potentially requiring an amount of skill far higher than any legal driving test requires, is in short, a big ask. Therefore I have so far left Level 3 to the people with the deep pockets, Level 4 to the enclosed car parking experts and Level 5 to the fund raisers and dream sellers.

So why am I writing a blog on designing autonomous cars? Well that’s an easy one, I’m writing this because I want it to happen. I want a fully autonomous car, I want Johnny Cab from Total Recall, I want to be able to nod off peacefully in my car and never again suffer the terror of waking from a micro sleep at 3am as my inside wheel hits a rumble strip, or having to stick my head out the window in an attempt to force my face to stay awake long enough to get me to the next service station. I’m tired of traffic jams and single carriage ways where I find myself driving at 60mph, in the rain, in a car that weighs several tonnes, in the opposite direction to other cars of a similar weight, travelling in a similar fashion, driven by relatively untrained people, with the combined momentum of a small strategic nuke and separated by what? A mere white line. But more than anything, I’m fed up of leg pains, an aching back and a numb backside. Perhaps its my age showing, or maybe I drive long distance far too often, but all of the above is starting to outweigh the joy that I once found in driving. Those moments of bliss where I get to feel what the car under me can do, no longer happen anywhere near as often as they once did.

However although I may have just uncovered the subject of some future blogs, I have seriously digressed and I need to get back on track.  Although in its various forms ADAS has been around for many years, it is, on the whole, still far from perfect. ADAS is vital, now more than ever but it still suffers from the infinite number of issues resulting from the equally infinite number of factors that make up any particular scenario or Use Case (I’ll be talking a lot about these), that affect the Level 3 and above features.  After all these years the detection of painted white lines, grass verges, curbs and even other vehicles is still problematic.  ACC, a feature that has been around for many years, still relies mainly on radar, which is still not great at robustly “seeing” stationary objects, and those that use cameras to classify what the radar can see, can still fail in bad light.  Combined with how difficult it is to robustly test these features, the list of system limitations is endless, and may well be the subject of a future blog, let’s see.

But system limitations is not what we are about, at least not yet.  First, as developers we need to understand what it is that we want to create and why; How to Design an Autonomous Car Part 2: Levels of Driving Automation will go some way to explaining some of that.

Finally you may have noticed that so far I have not included a breakdown of the acronym ADAS.  That’s because I have assumed that you know very well what it stands for and that’s why you’re still reading.  If you don’t know what it stands for, then welcome anyway; stay as long as you like, help yourself to tea and dont forget to subscribe.

Subscribe here
Up next:
How to Design an Autonomous Car – Part 2: Levels of Driving Automation
Read these too:
How to Design an Autonomous Car – Part 3:  Getting started
How to Design an Autonomous Car – Part 4:  Systems Engineering
How to Design an Autonomous Car – Part 5:  Atomic Requirements
The rise of Autonomous Car and the Death of the OEM
Can an Autonomous Car be safer than a Cargo Ship?
Regulation 79.02 and what it means to you
ADAS Market projected to reach $56bn by 2023
World first launch with off-the-shelf ADAS solutions
Menu