Increasing Complexity of Software in Automotive Industry – Tata Elxsi – MATLAB – Simulink

Increasing Complexity of Software in Automotive Industry – Tata Elxsi – MATLAB – Simulink


a fantastic way to sort of get into the way I would like to explain to you the challenges that we are seeing in the software complexity which is coming through but two points before I get into the presentation one is that the next 30 minutes I would like to really abstract it a bit like the way he used to write code and software and then we would write code and software we would actually look at the problem definition and build a flow chart and based on those decision tables go to developing an algorithm and then go to coding and so on and so forth second is technology is a great disrupter I don’t know how many of you realize because the average age is possibly much lesser less than 30 or a maybe around 30 here technology is a great disrupter its disrupted businesses it has disrupted companies it has disrupted products it has killed and destroyed industry is the way they work so the question is what is actually happening in this automotive electronics world right now are we really understanding what are the challenges that we have in my very limited two years of interactions with many OMS and rnd heads across the globe and many engineers I have if I may commit the discretion I think that we just don’t still understand the impact of what’s happened I’ll be really talking automotive any longer or is it that technology is entering into the automotive architecture and we don’t still understand the impact we have absolutely not enough in a sense of the word understood the impact that this greeting of technology into our architectures is going to cause to the way we work the way we design and the way we look at things so hopefully in the next 20 25 odd minutes I’d like to share our views on this very interesting topic keeping in mind these two broader themes just to take away from the slide today Lexi is 5,000 strong embedded product design company basically doing design and creative services financially extremely strong continues to remain focused on embedded product design and strong financials extremely important if you have to engage in this disruptive cycle of automotive electronics take away from the slide which I would like to show a mirror establish a bit later consumer electronics automotive broadcast and medical electronics these are the only four places where we work we focus only on these four areas and automotive happens to be a biggest business space around three and a half thousand people and long-term engagement with many many customers and there’s a significance of this slide why has put it here I’m not trying to sell that Alex your plug it but I want to establish a connection between what you see in this slide with with the evolution of the architecture as we go forward so this is CES 2016 I think all of you would have heard the name 2016 cs2 2016 and some of you would have gone through some announcements first of all it is a consumer electronics event so what is what are the automotive guys doing they’re fine they have been there for some years but why is it that they’re setting up huge shops in this event year after year and what are the announcements they are making how many announcements are automotive folks making about engines and fuel efficiencies and and the lightweight material and so on and so forth how many such announcements have you seen in the last one year look at the announcements that they make just go through these infotainment active safety cameras lead our user interface applications connected car control your home while sitting in a car all type of electronics absolutely digital clusters don’t want analog clusters driver behavior either cognition image processing payment gateways content consumption is this what automotive architectures are all about when was the last time you people thought while you were setting up a system that an automotive car architecture would look like something like this with these type of announcements coming in but maybe I could be wrong maybe this is an aberration but let’s look at what people talk off three trends broadly people will talk of five six seven trends but broadly three trends a connected car what does it mean it means connected infotainment it means telematics it means app so connected car so what happens to a connected car neck automation of vehicle functions automation of vehicle controls and that would then lead to semi autonomous car or autonomous car 2025 2030 doesn’t matter it’s insignificant whether we reach autonomous car architecture in 2025 or 2030 does not matter what matters is vehicle functions are getting automated driver assist system emergency braking steer by wire rigid just showed those you know a dash applications where a car stop it’s all electronic and then lastly efficiencies lightweight electric vehicle hybrid electric vehicle fuel alternatives so these are the three broad strengths and you’ll be surprised to see that the major trends are non the first two categories are non classical automotive strengths let’s go further now this is not my data it’s just on my data this is data which has been gathered from various sources from experience from talking to people so I was really now understanding what we are talking we are talking a car is this how a car is supposed to look like 20 million lines of code in a fighter plane 100 million lines of code 100 million lines of code 100 plus issues in a car issue has an SOC and SOC in normal terms is a CPU of which your computer has its octa-core is dual-core it’s quad-core it has a memory footprint that’s what an issue is all about hundred-plus use electronics software is more than 40% of a car architecture today it is dangerous let’s go back and let’s look at what is it that we are looking at they looking at a car but this is not what a car was supposed to be and then you go down and you say okay fine I mean you know these could be lies but then you go to their annual reports and they say 90 percent of R&D spend on electronic systems maybe it’s unbelievable so let’s go back SEMICON industry the fastest growing area for SEMICON industry is automotive somebody says is seven and a half percent growth somebody says it’s nine percent good again does not matter what is the percentage of growth what matters is this is thirty two billion dollar fifty two billion dollar plus amount of investments are being made on renesis Texas Instrument Infineon and Nvidia’s of the world so now let’s go back and look what is the type of complications that we are talking of in car electronics today it’s got tremendous dimensions it’s got abs it’s got a sociis it’s got inclusion of electronic software it has got inclusion of apps it’s got inclusion of cameras radar sensors electronics electronics hardware so the car actually is no longer a domain of the traditional manufacturing automotive engineers it has tremendous size tremendous amount of inclusion from the electronics folks and technology of this nature has been there so therefore in my view limited view this is a cycle of disruption that we are entering and the immense challenges of this disruption are still not realized by the automotive folks because to them it was a very very closed proprietary network within the car now has this happened in the rest of the industries it has have these type of challenges being addressed in the past in other industry successfully yes take a bank you go to an ATM you have an accountant carve a road or coat route somewhere you go to Bangalore Pune wherever in the world or somewhere in the world you go put your credit card you swipe get your money back whether it is in dollars pounds rupees does not matter you get their money in the right in exactly a predictable time so what’s happening you insert it verifies you it authenticates your identity it has a camera is recording your presence it goes back to the data center in the data center it calculates it checks whether you have X Y Z can the money be disbursed in what currency in what notes it has to be in what currency and then it updates your record and then flushes it back on your web on your phone on whatever device and this is happening millions and millions of people are doing this at a single instance now technology is agnostic it does not matter which industry it is trying to address problems it doesn’t matter whether it’s the problem of banking retail or a building automation or automotive it’s false complicated problems and that’s exactly what is happening and we don’t realize it so now is this something which is going to happen in a car which is now taking proportions of digital car and we have not yet seen a car with complete automation we seen parts of cars which are automated so we are entering into a era which is a very very interesting leader and either where there will be huge amount of disruption and that disruption is going to be from technology and if that is going to be the disruption what exactly is it that we are doing one question many people ask easy is connected card for sure it is that’s an argument we should not even get it that’s the philosophically argument or long-lost don’t even bother it’s exists it exists on the roads of India there are car models in India which give 2g and 3G through SIM card it’s connected whether the capability of a connected car is of the highest order or whether it is of a basic order is a different question connected cars in India exist even Euro lower / are connected in a certain way but a connected car fleet management all of these are signs of the fact that it is a reality it has existed and it will continue to exist so no argument on this my second third slide was consumer electronics broadcast medical and automotive and look at what a connected car is ready it’s a car where technologies are converging if you are going to have a car which is connected it’s going to be connected through protocol communication protocol right now who G 3G and people will move to 4G and the good part is that since the network connectivity across the globe is still not standardized for G’s even if the evolution of 4G or LTE is still happening it means that the car the model where the car will be very strongly connected through a very well-established 4G networks or 5g networks is still to happen and we all know when bandwidth is made available to any device what happens it’s huge it has huge disruption and huge opportunities so there is communication perspective to a car there is a broadcast content broadcast perspective to a car content what are people going to consume do you play CDs or cassettes any longer why should I not play why should the passengers behind in the rear seat not play or watch a cricket match while you are driving from Mumbai to Pune and just pay 10 10 rupees or 15 rupees it’s possible so you will see lot of cable operators you will see a lot of broadcasters you will see lot of companies developing apps trying to reach out to the automotive segment to bring content into the car and then you have people like medical you know people who are in the medical area trying to develop app for personal well-being all of this goes back to what I am trying to establish since the first the first thing that a doctor look at the architectural problem that we are being faced with their inclusion so what problem are these trying to solve and what is the nature of the beast that we are trying to solve here it is a confluence and convergence of technologies so let us go forward a very simple example again and these are smaller things when implemented in a car architecture you will realize the type of issues that we end up with when you’re writing code you write a code for a certain stack a certain performance and by the time it comes to productions of performance expectations footprints are totally different it is consumer electronics which is coming into the car the SOP for a car is clearly between two to five years debatable but let’s take it as three to five years and the consumer electronic components like software architectures software tools software resources they change every three months of six months so now here is a situation where we recently were in the process of designing and architecting a architecting a heads-up display solution for a short for a very big company and they are looking at this to be implemented in 2018 the architecture the budget constraints and features and functionalities which are expected are being designed on the associate each other today a huge mismatch and Midway after the architecture was designed we just called up the associate and said give us a roadmap for 2018 and we Reina is that in 2018 roadmap if we use those type of associate would be cut by half features and functionalities are already getting embedded and this is the type of architectural implication some of these things have and it boils down to the way software gets written the complexity of algorithms complexity of state flows complexity of the way we want to model and simulate these conditions huge gap take a connected car for example and this is where you know what I just said how are you going to simulate a condition where you are going to have five or ten or fifteen cars on the road and they’re not going to you’re going to test your algorithm or whether they’re going to crash with each other whether they will to stop identifying after identifying your obstacles are they going to have lane departure not what is happening to driver behavior you can’t keep practicing this for every use case on the car and no longer are we now talking only infotainment and therefore only some dumb system of C C++ or html5 or QT and stuff like that now we are talking active safety which is complicated image processing algorithms which help make decisions in real life for a driver to either take a decision off of the car to take a decision how are you going to emulate these things how can you write these codes how can you create realized scenarios not possible but this is the way we are thinking right now the way we approach these things is a very classical traditional way let’s do this we use X amount of tools for requirement gathering we will simulate in a certain manner we’ll run it through but look at the tools that I was very glad to see the data scientist module that he is that Richard showed these have to get embedded in the flow chart itself how do you simulate a situation which is so complicated which has so many use cases and where you have to simulate real-life situations of cars semi autonomous car hundreds and thousands of them on the road not possible to do it through through very simple architectures I’ll skip this in the interest of time and come to this aspect you’re now going to dry design and architecture for a car a car arguably again these are still initial data a connected car can generate anything between 50 to 80 GB of data per hour running 40 to 80 GB of data per hour of run in India you have three million cars today in China eighteen million cars gets sold in u.s. 19 million passenger vehicles get sold so do you really realize what is it that we are getting into who will manage this data who will handle this data oh you know look I have designed a software architecture it will scale really does it have the capacity to handle this type of data who will do data modelling who will simulate the user use case so it’s a very very interesting situation to be in because this data will be generated 24 by 7 and now look at the type of data which comes in this data will got will be which would have to do with location it would have to have to do with GPS it has to have data which is processing data for driver assist system and giving driver some instruction or giving car some instruction it has data about driver behavior and car behavior we access to go to the OEM it has some data which has to go to some service agents and then there are apps they are connected at so this is not a single type of data it is heterogeneous data how can you ever simulate it this is the problem I still don’t think many of us a large number of us even look at when it comes to designing a car architecture and a car actually architectural drawing board and these are the initial estimate multiplied by millions of cars that you think of x 24 bar 7 x number of countries and remember just remember the cars that are going to be running on the road which are connected also have to have somewhere on the air upgrade and software on the arab red so who’s going to do that if I am an OEM and I have 10 models in 20 countries I have to decide which car which model here has to get firmware on their plate who does that who is trying to grapple the monster these are the challenges that are facing us in a very very big you know in a very very big way so I think these are just you know these sort of slides go back and emphasize what Richard talked of data modelling data analyst analytics the importance of some of these things but more importantly you see what’s emerging is that when you are now going to drive an architecture the architecture cannot be an architecture which is pertaining to only your infotainment or only to your car network or only to your a dash or only to your powertrain or something like that it is multi dimensional it is multi domain you have to have an architecture which looks at the connectivity and the state flows across different type of domains it has to emulate conditions of data flow from X a to Y while crossing and looking at interconnect and then it sees how this data has to go out of the car and with what relevance and with what priority it has to go out and unfortunately because we are the way we are which is automotive it has its own processes it has its own life cycle of decision making we are a lot behind the speed with which technology moves so your architectures of today are not going to be relevant for implementation on new technologies and new software’s two years down the line so instead the reason that we have a car which has 100 million lines of code is there a reason that today you have any one particular om or anybody which has infotainment systems and equal number of architectures for an infotainment system is it sustainable if I have 10 car models I want to have 10 architectures for infotainment and then how do I reduce costs by offshoring it to India and after India where wait now No the answer lies in building an architecture which is scaleable confirming to standard confirming to compliance with open 80 is allowing people to contribute and this architecture should be should have complete simulated performances on various associations design it keeps evolving and gets handed over to a Tier one or service companies like us now we see a large amount of this thing happening actually so let me not say that you know I just have five six minutes left so I’ll just skip to the last slide so this is happening in the last one and a half years almost every o M worth its while has opened up and invested hugely in R&D centers in the Silicon Valley next to Google and next to Apple they do realize that they have to be close to the center of innovation technology innovation and therefore by being closely linked to the technology innovation ecosystem they would understand the changes and trends that are happening and bring it back into car architecture you will see many people talking of common platform earlier they used to talk of vehicle platform and now people are talking of reducing vehicle platforms on let’s say 12 to 8 but people have increasingly started talking of having software platform so there is this huge amount of change which is happening and this change is being driven by technology changes this change is being driven by compliance by obligations that OMS have towards regulatory authorities and also by the user the user music who’s expecting of extremely good user experience after all we are all engineers we don’t design products for ourselves we design products for a user which is a driver what do you want to have a very complicated experience how many of us like to have a very complicated experience of working with the phone no one of us we don’t design forms right so somebody who’s designing phones is making it simple so we have to build that simplicity in the user experience and that simplicity has to come down to the way we handle data and the way we manage our software architectures so of course there are other issues which are very very important issues who’s the owner of this data this is a philosophical debate but this has huge implications on technology the way we are going to architect cool the customer trust who will manage this data earlier when you had telematics the service provider like a Vodafone or any other company would have an obd device and they would manage the data Doheny was not interested but now if the oyon is going to provide you with a car which is connected and they are going to they’re the ones who are going to vouch for that connectivity do they own the data will they manage the data how will they manage this data whatever architecture will they build privacy and so on and so forth so here is the situation a concluding slide that they possibly have to realize that a car today is very digital a car does not have standards the interoperability of various architectures is a huge issue the dying the change and the disruption happening at every level is humongous most as a car network is possibly on its way out last two or three years ethernet is coming in having designed applications for Ethernet do we even know what will be the impact of performance of Ethernet being implemented in a car you have autonomous car coming in semi autonomous cars coming in v2 X coming in v2 v coming in v2 y coming in you have Geneva coming in you have AGL coming in you have cutie coming in you now have Android coming it you have cosplay coming it you have mirrorlink which one does it not even know how many will adopt but you also now have Toyota saying I want to implement HDL HDL is for open standard for applications which is non at the non-google so where is all these inputs coming into one framework I have one common approach towards software architecture it does not exist yet so this is where investments are happening at a larger level this is the type of problem OMS and tier ones are trying oils are trying to solve for us it means we need to relook sooner or later and relearn the way we have coded the use of tools is very very very important these things cannot be simulated on cars running on the roads of Pune Bangalore or Europe or wherever it is no it’s not possible very important to have multi domain simulators very important to have these concepts like the data science very important to have as close to production ready code as possible technology and while we do all do this technology changes in the Geneva forum in the AGL forum in Japan they are going to announce a unified code base which we mean that what we’re trying to do is get Geneva code base get a GL code base and get tides and code base did they ask anyone of us who design infotainment architecture water user what impact it will have in the way we ask you take this infotainment system no that’s my point technology is a great destructor the type of technology inclusions which are happening in the cars are the MEMS this is a moment of opportunity for all of us personally and professionally does not come in a lifetime next two to three years huge disruption and that is the way we work the way we model the way we code the way we have to understand the complexity of software which is going inside the car it is very different than what we are used to absolutely just think of it absorb it for a moment and you’ll realize what is the type of an exciting era we are entering into this industry is here to stay these investments are here to stay and when investments happen when disruption happens all of us and get good jobs we earn good money and I think MATLAB will have a huge market share going forward there’s no one else for MATLAB fantastic tool so this is what you know exactly 30 minutes over I think we believe a total HCV believe we totally are committed to this vision and that possibly is why we have grown qualifies percent last year absolutely this is the way we live and this is the way we are thinking absolutely down to every engineer trying to make our engineer understand we are entering into a era in automotive which is different it is disruptive it has complications which we have not gathered understood it’s time we learn and get on with it and that will be the mantra for success at a company level at individual levels and at a cleaner


Leave a Reply

Your email address will not be published. Required fields are marked *