Zenuity was formed at the beginning of 2017 as a joint venture between Volvo Cars and Autoliv, after both companies identified the huge shift about to happen in the automotive world, especially in the area of safety and self-driving technology. Automotive giants like Volvo have realised that they need to prepare for the future, investing a lot of money into future technology and new business strategies. Speaking to Erik Coelingh, Technology Advisor and Vice-President of Zenuity, I find out how the newly-formed company is working towards the rollout of autonomous technology. "To prepare for the future, you need to invest a lot of money in this technology, have a lot of competence and be able to move quickly,” he says. “Working quickly in the automotive industry is not trivial because there is a certain culture and way of working that is inherited from making metal components rather than developing software. So, by forming a separate company, we can create a new modern culture where we can focus on agile software development." Zenuity obtained all relevant software and IP from Autoliv and Volvo Cars, which have also contributed 150 skilled employees each. Zenuity also has the advantage of selling the software back to its two founders. This means that Zenuity can generate revenue by developing software for Volvo Cars and sell it to other OEMs through Autoliv. "In some senses, it’s a really simple business plan,” says Coelingh. “We have to make sure that Volvo Cars has the latest and greatest technology and it's Volvo's responsibility to implement our technology into its vehicle programme. It is then Autoliv's responsibility to package our technology in a complete system and sell it to other OEMs.” Zenuity is also building other alliances in order to deliver a complete stack of software to its customers, recently partnering with Nvidia on the development of artificial intelligence.
Although we are seeing a new mobility era, most cars are still being sold in a very traditional way where automakers are selling products to consumers. This is because the lion’s share of profits still come from a buy-to-sell model. However Coelingh believes that this will change extremely fast. “We do not know what the new model will be exactly; it depends on how automation and carsharing kicks off, which will significantly affect car ownership. There is still a question when autonomous cars will be on our roads and how the market will form around it, as we don't know who will own them,” he says. “It could be the government, carmakers or the cities - nobody knows yet, which is why the automotive industry is so exciting at the moment. There will be a lot of change in a unique playing field." All that is certain, is that the biggest shift in the automotive industry is, and will continue to be, the amount of software being developed, forcing companies to release software more often than in the past. “The car will take a lot more responsibility of driving itself, creating software with extreme robust requirements which is a huge challenge for many."
There is already quite a lot of software within modern cars, but there is little to no maintenance on it, such as scheduled updates. This will change in the future, with over-the-air updates improving the overall user experience. This is already common in the IT industry, especially with mobile phones, where companies are constantly updating products with new features and applications. Coelingh agrees that we will start to see this in the car industry as well: “In the long term, we believe that software on cars will be deployed the same as apps on mobile phones and we will make sure that we maintain our software so that our vehicles can be updated with new functionality all the time, creating a safer and more convenient car for our customers. This in itself will create new business models, such as how you own a car, through such things as subscriptions to the software updates like you would experience in the technology industry." The problem that this overhaul brings to the automotive industry is cyber threats and glitches - a new wave of issues for automakers that are not used to dealing with them. Over-the-air updates are the most direct way of combating this, however there still needs to be some form of initial security and thought to the protection of the driver and the systems before the software is rolled out. By having the advantage of an automotive background, Coelingh explains that security and functional safety is in Zenuity’s philosophy, now applying it to IT software development. “It is about finding the right balance between being fast and agile, but ensuring safety and robustness; this is where the core tension is in our product development. You cannot play around with this kind of technology,” he warns. When it comes to autonomous technology, the main challenge is to ensure that the function of the technology is safe in all cases. This can be difficult, as developers must figure out a solution for every situation, which is impossible due to the lack of autonomous cars on public roads. “Traffic is so incredibly complex, which creates a sense of the unknown,” adds Coelingh. “The way of figuring this out is by collecting data to test and improve the software. When looking at unsupervised automation, the car has to be able to deal with all situations on the road and, unfortunately, that is a very large space.” One way of approaching this issue is by limiting the functionality in certain areas as developers learn about the environment and move on as the vehicle develops. Things such as weather and road conditions can significantly affect an autonomous car in many different ways, so it is important that software specialists learn and develop appropriately. "Autonomous cars will not be able to function in a snow storm where visibility is low and lane markers are not visible, so you would not start to develop the software here. The best approach would be to start in a limited area with sufficient space so that you can improve the scope and grow,” says Coelingh. “The second way of overcoming these issues is for developers to supervise the self driving vehicles so that data can be collected quickly. After a while, you will have sufficient data to understand the environments in which the car is capable of operating ."
To aid autonomous vehicles, governments and cities could build dedicated traffic environments to help pave the way for self-driving vehicles on public roads. However, it is difficult to predict how this will happen as nothing has been put in place yet, because autonomous cars are still some time away. Coelingh believes that such things as scrappage schemes and dedicated lanes will not help the long-term success of autonomous vehicles “You cannot scrap all of the older cars to make way for autonomous vehicles. It will take too much time. I also think that dedicated lanes will be a problem as they will not be built until there are self-driving cars on the road and autonomous cars won't be rolled out if these lanes are not built,” he says. This leaves the industry with a very difficult problem to solve, but Coelingh advises that, “the only answer is to build self driving cars that can drive autonomously in the existing traffic system. From an infrastructure perspective, it helps to have lane markers, well-built roads and readable traffic signs.” Fundamentally, this requires the same driver aids as what are already in place to make it easier for humans. He also believes that getting access to data through the cloud will be a main driver for progression. “We are testing a system that shares traffic information through the cloud within a vehicle fleet, so it can identify such things as congestion, variable speed limits, bridge openings, traffic lights and more.” This seems vital for autonomous cars, allowing information to be shared over the cloud and make life easier for both human and automated drivers.
Although autonomous cars are far away, newer vehicles will start to play an enormous role in functionality over time. Software as a service is starting to take off as companies see more support, leading towards autonomy. But, Coelingh says, how fast the unsupervised autonomy grows is still uncertain. “I don't think that all cars will be self driving in ten years time, however I expect to see premium vehicles with some form of unsupervised automation, as well as driverless taxis in larger cities.”