Visteon Corporation, that designs, engineers and manufactures innovative electronic products for vehicle manufacturers, is building the first platform that takes a Greenfield approach in developing an Artificial Intelligence-based autonomous system using machine learning, with the feedback to improve its capability. Its goal is to bring in the autonomous driving facility very quickly and be ready when the market and industry is ready for these vehicles.
“A lot more investment and testing have to go into it before it can replace the human being. We are squarely targeting level four and five and not level three. This approach will be superior to what everybody else is doing”, Sachin Lawande, President and CEO of Visteon, told T Murrali of AutoParts Asia in an exclusive interaction. The excerpts:
Q: What are the opportunities for Visteon from the new trend of driving shifting to commuting?
Lawande: This trend is going to take a few years to really unfold. If you look at the volume projections for the next five years, the number of cars going to be built worldwide is not going down; overall it’s going up. If you look at cars per thousand there is still a long way to go in markets like India and China; 15-18 per thousand which is virtually nothing. So for the next five to seven years it will be business as it is today. Even in countries like the US and Japan, having high vehicle density, we do not see commuting having a major impact on ownership. The average ownership of a car on the road today is 11 years.
So we have in general a very aging fleet of cars in the US but the number of miles driven is going up, not down, chiefly because of the lower fuel costs. Fuel costs and interest rates are the two biggest determinants of the automotive market in the US. As long as these two have been low, the market has been pretty good. This represents mainly the East and West coasts but when you come to the South or Midwest, driving is the only way of commuting.
Q: Are you referring to mobility as a service?
Lawande: What we are talking about is mobility as a service and not mere car ownership. Today, the model is to own a car but its use is underutilised; 80 percent of the time it is idle. So in the long-term it is good for the economy to unlock this asset that is lying idle. Trillions of dollars can be unlocked and put to better use. Of course there are other concerns like employment taking a hit (10 million jobs will be lost in the US alone) if we go in for autonomous cars. The multimillion dollar question is: When will it happen? Is technology ready for this?
Q: When would things move to autonomous vehicles?
Lawande: People in general oversimplify the challenge of putting a self-driving car on the road. This transition will not take place till you have a well accepted car driving technology that performs at least as well as humans in terms of driving and safety. The question is how long will it take? My bet is that we will not see a full self driving car till 2030. We know what is going to happen till 2021. From there it’s just a decade giving two to three cycles; there is too much technology to overcome in that period.
Mapping technology like GPS took the industry 15 years to get it from the labs to where it was usable. That technology was simpler than what we are contemplating now. Here the car makes the driving decisions and does all the monitoring at all times and under all conditions; it is called level-5. Level-3 is what we have today, which is that the driving function is done by technology but monitoring has to be done by the human. In level-4, monitoring is done by technology but you still need the human to oversee it.
If you look at the traditional duration that industry has taken to launch new technology, you find that it is 15 years for major transitions to occur in our space. If you go by this, level-4 will be in place by 2025 or 2030, with level-5 coming in 10 to 15 years from there. That is the horizon we are looking at.
Q: Why do you think of a longer timeframe?
Lawande: Vehicle ownership in that time frame will be of people who have been born well after the internet and mapping. Today, it’s not the case. Most of the drivers on the road are not tech. savvy. They do not trust technology instinctively like our kids do. So not just technology transition but a generational transition also has to take place. 2008 is the year when mobility really came in with the iPhone bringing in connectivity with the internet. By 2020, the generation that is driving would expect to have the iPhone level capability in the car as internet and connectivity would have gone much further ahead. That seems to be the time when the transition is most likely.
Q: Can you explain why?
Lawande: Today, all products and technologies are designed for the driving experience. That will still be the business for the next five to 10 years. We are developing what I believe is the most advanced autonomous driving capability in the industry. We are known for clusters and displays of today, not necessarily of tomorrow. The wave that preceded self-driving had a whole range of active safety products like camera-based forward collision warning, rear camera and side line spot detection. Drive- by- wire was happening though the human was in the car making the decisions. It became a system integration challenge to bring in situational awareness by combining different technologies. The first activity in this area happened in 2005 when Stanford University put out a challenge as to who could design a car that could drive 200 miles on its own. That really created the fancy towards driverless cars; the technology kept evolving from 2006.
Active safety devices that were essentially-sensor based and conventional programming techniques ran into the challenge of the level of accuracy they could obtain in performing object detection and classification. Conventional technologies that are relying on cameras and radars are doing just image pattern recognition; at best they can achieve a 90 percentaccuracy hit rate. These are the limits of our current way of programming.
Q: Won’t the current technologies support autonomous driving?
Lawande: The current technologies went up to a point where they could implement forward collision warning and autonomous emergency braking but could not take it to autonomous driving. So there is a real divide between active safety and autonomous driving. The breakdown is in the inability of current systems to learn and implement situational awareness. Earlier, the technology was machine learning which was doing entirely different things like determining what type of shirt you would like the next time you went online. So for autonomous driving we realised we had to develop new technologies. If you have a learning system in place it will get better and better at navigating over the things it has seen. That is the promise our technology is offering; we are building the first platform that takes a Greenfield approach in developing an AI (Artificial Intelligence) based autonomous system using machine learning, with the feedback to improve its capability.
Our goal is to bring in the self-driving facility very quickly. We need to be ready when the market and industry is ready for these vehicles. Of course a lot more investment and testing have to go into it before it can replace the human being. We are squarely targeting level four and five and not level three. This approach will be superior to what everybody else is doing.
Q: What are the driving forces that push Visteon to create new technologies?
Lawande: We have recast ourselves as a technology leader. The automotive industry operates in the industrial era mindset where technology transitions take a long time. But if you look at yourself as a technology company, the chief characteristic will be one that anticipates trends in the space that you want to operate in. So we have built some processes internally and are building up on those to evolve technologies that matter to us.
Technology companies must be comfortable with cannibalisation; that is a very core capability. If you are not willing to cannibalise yourself then somebody else will cannibalise you in a fast-moving world. You will not survive in this field if you do not look like a technology company. All technologies have a lifespan and we have to prepare for the future.
Q: Information management is a core issue for the technology companies. How do you manage the multiplicity of information flow from all directions? What will be the supporting infrastructure?
Lawande: The more data you have the more you have to use technology to manage the data. Today, a car generates data in gigabytes per hour. But they are, we don’t use them; we use them only when there is a problem. But if this data were captured and analysed in detail, imagine what they would tell us on the improvements to be made. We are gearing up for that and I call it the cloud strategy. You can process the data locally, not inside the car. You cannot process them in the internal servers here; you have to do that in the cloud.
The companies which do this will be the leaders of the future. We are building AI or machine learning competence not just for the purpose of driving cars – it is also for managing the data put out by all these cars. It should cover manufacturing also; my goal is to make a single digit PPM (parts per million) built on machine learning and big data processing so that you can put in a check before the event occurs. This applies to all our systems — finance, manufacturing, HR etc. We hope to get incrementally better and better in doing this, to reduce the arbitrary line between IT and engineering. You got to have a vision, and drive on relentlessly to achieve it.
Q: What are the highpoints in your operations and where do you think more work has to be done?
Lawande: We are very good in manufacturing capabilities at our 22 plants. We have just started a new Operational Excellence initiative for manufacturing, with two elements to it – automation and cloud. I believe we are best-in-class today but we need to keep evolving. We have had outstanding success in winning new business, $4.3 billion last year and to date $4.1 billion this year (in 3-quarters). We need to improve our talent and human capital development, marketing operations and cost as a percent of sales. We must have a culture of cost focus, to look for opportunities to reduce cost all the time. You have only two choices, either remain competitive or go out of business.
Q: How are you going to address these three: Talent development, marketing and cost?
Lawande: We have already started working on cost. Dialogue and discussion on it will become a part of our culture here; it was not so earlier. Regarding human development, we have hired a professional HR leader who has begun an internal survey of our capabilities, mapping them into six or seven different areas where we will try to develop the available matrix. The points we have defined are: To be seen as a technology leader, to be customer-oriented, to be a cost leader, to be a truly global company (we are a global company but the mindset is very local) and to act with speed.
I want everyone in the company to continuously evaluate themselves against these five things. Are we aligned with them or not? What can we do to proceed further? You cannot be a technology leader today unless you are a global company; you must have a global mindset.
Q: Have you set targets to achieve this?
Lawande: My target was three years, of which one-and-a-half is over. It is a moving goal post where we keep progressing.
Q: You had talked about the SmartCore technology; how is its progress?
Lawande: We had announced our first SmartCore business win; this quarter we have won the second one. This is definitely starting to roll and we are also engaged with an Indian OEM to launch it in India. India may well be the third or fourth one to be announced. Technology is really going in all directions.
Q: How does the SmartCore technology help the OEMs and the end-user?
Lawande: There are a few problems on how this industry has evolved in the recent past. The evolution has been great for suppliers like us because we can sell more products like clusters, infotainment, displays
etc.; literally the car is hanging on these boxes cabled together inside the vehicle. So we have to look at cost, weight, power consumption and space itself; this industry is proud when the next generation vehicle weighs 50 lbs. less than the previous one. Anything you do to reduce these factors is welcome.
Now a fourth element has been introduced because this is a network connected to the cloud. The more things on the network, the greater will be the probability of a security leak. So when you collapse it all into one box you can effectively reduce the attack surfaces.
From the end-user view let us take the clusters and infotainment systems. Today, for Ford we make the clusters and someone else the infotainment system but they have to show content on a cluster. The content is richer now including left and right directions, colour graphics information, telephonic names etc. that you can get on the cluster fed by the infotainment system. So there are two systems and at some point they have to exchange data – invariably bringing in a pain point in the development of both these systems.
Everything runs off the same chip so the flexibility and user experience is more aligned to what the consumers expect it to be and not limited by how the automotive industry builds products. More and more the battles in the market place are fought over the user experience – the bigger the displays, the more the flexibility. That’s where this is headed.
Q: Which are the key focus markets for you in the next five years?
Lawande: Asia is the biggest market; it gives us the lion’s share of business – more than 45 percent. We expect it to cross 50 percent in the next five years. So we will largely be an Asian company. However, Europe is extremely important because it is the leader in technology. All paths lead to Germany as they have won the world with business; nobody can survive if you don’t first succeed in Germany. These are the two important markets for us.
Q: As Visteon transforms technology, is there rise in patent filing?
Lawande: Quality of patents is more important than the quantity. I won’t get into the number game as it would mean more patents of low quality. What we see is that cultures do have a role to play here. The culture in the US is still one of the most innovative cultures in the world. When you go further east the innovation level drops. India has an interesting potential, not fulfilled as yet. We are not at the innovation levels yet, still at the application levels. So one foot has to always be in Silicon Valley. Germany is not doing much of innovation in our space; it is more in the hardware. India is behind – the number of PhDs is still very low as we have not created the right environment.
Q: Finally, can you give us an update on the data security lab that you are going to set up in India?
Lawande: We did establish a 3061 conferment process inside our engineering that we call VPDS (Visteon Product Development System), a gate-based process having nine gates. This has been modified to bring our cyber security into it through the J-3061 framework. In part of the framework you must have the ability to validate, for which we have set up an initial team of 10 people in Bangalore. That team is ramping up to build the tools for White Hat hacking into the system. All new systems that we build from here onwards will have security architecture in addition to the system software architecture. The first customer product that will see the benefit of that is the SmartCore bin, where it was a mandatory requirement. This team will be the one delivering that capability.
Q: When will the team be fully operational?
Lawande: By end of this year. Of course a team of 10 is not sufficient; it will continue to grow from here. The first product launch is by 2020 so we have to be ready much before that, fully impacting the programme. I expect next year to be a big year for that side of the capability portion. We did do what we set out to do, and we have done it in India.