An average new car is managed by between 70 and 100 Electronic Control Units, and constantly monitoring itself and its surroundings with hundreds of sensors.
To emphasise the challenge, some call it “The Mother of All Tech Battles”: In our effort to digitalise, connect and automate every aspect of mobility, we need to handle steeply increasing system complexity, cyber threats, new business models, and lawmaking issues, just to name a few of the many obstacles ahead.
As experts in embedded and IoT solutions, and a trusted and experienced technology partner to the transport and automotive industry, at Data Respons we face these challenges on a daily basis. And based on that expertise we feel we have a fairly clear view of the things to come in the 2020s, a decade that doubtlessly will bring enormous changes in the mobility sector.
A computer on wheels
The vehicle industry has come a long way since the introduction of the Cruise Control, the first vehicle feature that integrated mechanical and electrical systems. That was in the 1950s. Now, cars are computers-on-wheels running millions of lines of code. An average new car is managed by between 70 and 100 Electronic Control Units, and constantly monitoring itself and its surroundings with hundreds of sensors.
Partly, this revolution has been triggered by other technology areas, for instance telecom. There has been exponential growth in memory and processor power, while components have become cheaper, smaller and more robust. On top of that, a set of new components like radar, infrared cameras and ordinary cameras are being added to the system, further increasing complexity on all levels.
Facing this complexity, vehicle architecture is evolving as well. Currently, there are two main approaches in vehicle architecture: Either one large computer serving the whole vehicle or a distributed set of computers with a network between them. Both paradigms have their pros and cons, and it remains to be seen which approach will prevail.
Increasing amounts of data
However, as autonomous driving is slowly developing, component and system complexity is increasing, and with it the amount of data to be processed. To handle that complexity it is tempting to look for inspiration in aerospace and aviation, or similar domains operating complex, mission critical systems. That makes good sense, e.g. when it comes to data analysis, as these areas produce a lot of data to be processed, analysed and combined in the most efficient and correct ways.
But there is a difference. In aerospace and aviation you operate in controlled areas, and traffic is heavily regulated. Though obviously not without risk, it happens in a fairly controlled environment. That is not the case with a self-driving vehicle navigating in an urban area with its unpredictable mix of conventional cars, pedestrians, children, pet animals etc. So, is the object detected by the car’s radar a rock, a plastic bag or a child? Or is it somebody walking across the street with a bicycle, like in Temple, Arizona, in March 2018, when a woman was killed by a self-driving Uber car?
A thing of the future
A truly autonomous vehicle is still a thing of the future, although Tesla is leading the way with self-learning algorithms. But there have been a number of accidents in the US which clearly indicate that autonomous vehicles cannot be trusted 100 per cent. They still depend on driver intervention, although some car manufacturers seem to be over-selling their partially automated vehicle, e.g. by using the term “Autopilot” to describe its Driver Assist system.
Following an investigation of a crash in 2018 in California in which the driver of a Tesla died, Robert L. Sumwalt, chairman of the US National Highway Traffic Safety Administration, summed up the situation in this way:
“It’s time to stop enabling drivers in any partially automated vehicle to pretend that they have driver-less cars.” (New York Times 26.2.2020)
To be on the safe side, it is sensible to restrict fully autonomous vehicles to controlled environments like industrial sites or harbours.
But all that may change quickly. Industry roadmaps show, that by 2025 almost every car manufacturer will have a fully autonomous car in its product portfolio. From that point on the number of autonomous vehicles will rise quickly. When approx. 50 per cent of all vehicles have become autonomous it would make sense to gradually allow autonomous vehicles in non-restricted areas. We could see autonomous driving in semi-controlled environments like for example on motorways. Regulators may decide, that some parts of a motorway only are to be used by autonomous vehicles, with drivers switching back to manual when leaving the motorway and heading for urban areas.
The powertrain is simple
As mentioned, with improvements in vehicle autonomy the complexity of the vehicle will increase significantly. With one exception: the car’s powertrain.
Compared to an electrical engine a conventional combustion engine has more mechanical parts, and it is much more difficult to control injection times, combustion in the cylinders etc. In this regard the vehicle of the future will be simpler. However, when it comes to the powertrain manufacturers face an altogether different challenge: What will be the fuel of the future?
Although it is widely agreed that the conventional fuel combustion engine will be a parenthesis in human history, the battle about what will come next is still raging. Currently electricity seems to be gaining the upper hand, but although batteries are a much more efficient way of using energy than gasoline, they have an environmental impact, requiring rare minerals and recycling when worn out. For instance, to extract 1 ton of lithium requires 2 million litres of water. And cobalt, another rare metal required to manufacture batteries, comes primarily from thousands of small, private mines in the highly unstable Democratic Republic of Congo, often involving child labour.
For these reasons, a probable future scenario could be a combination of a battery pack on-board the vehicle, combined with electrification of roads through induction via the road surface or other energy transmission technologies.
But all that is extremely hard to predict. By the end of this decade things may have changed and other superior technologies may have emerged.
Regardless what powertrain technology will prevail, multi-layer connectedness will be the dominant feature of any future car. The vehicle will connect to its immediate surroundings, to local infrastructure, to other vehicles, and to the cloud, all at the same time.
The vehicle will monitor its surroundings through an array of different sensors, and it will receive data from surrounding infrastructure like traffic lights or an approaching emergency vehicle. Also, vehicles will communicate with each other. This short-rage vehicle-to-vehicle communication could come into play, when a vehicle is part of a train of vehicles. If the car in front detects an obstacle and hits the brakes it will instantly signal to the cars behind it to brake as well. Thus a vehicle can extend its on-board sensor capacity to thousands of additional sensors in its vicinity.
In addition to vehicle-to-vehicle and vehicle-to-near-infrastructure communication there will be long-range connectivity enabling other features, like user-based insurance, condition monitoring or various car-as-a-service solutions.
Securing data lakes
The data produced by the vehicle is stored in large data lakes, to be utilized for instant analysis, for development of new services and much more. Manufacturers, scientists, authorities and others can dig into these data lakes e.g. to design more efficient logistics and mobility systems. Accessing and utilizing these vast amounts of data ought to be beneficial to all involved, provided that integrity and privacy is guaranteed.
Obviously, these new possibilities create many risks as well, and in this it is crucial to stress the importance of cyber security and data integrity. Imagine if criminals could get access to data showing that a car and its owners are out of town, leaving their house unguarded, making it an easy target for break-in. Or imagine a trucking company trying to hurt a competitor by breaking into its system, downloading faulty roadmaps to the competitor’s fleet management system, deleting freight orders etc.
When it comes to cyber security and data integrity, the mobility industry can look to sectors in which these issues are mission-critical, banking and finance for instance, where transactions, access, and confidentiality are guarded by state-of-the-art technology.
Rewriting laws and regulations
Also, the digitalisation of mobility will give lawmakers some hard nuts to crack. Laws will need to be rewritten, nationally and internationally, and there will be tough debates on the freedom of the individual and the right to privacy versus what’s best for society and for the environment.
As an example, everybody would probably agree that it would be in everybody’s best interest to allow an approaching emergency vehicle to take over control of vehicles in front of it and force them to pull over, for it to reach an accident as quickly as possible.
But how about taking that scenario one step further: In everyday traffic, should local authorities be allowed to take control of a number of cars and reroute them to avoid congestion? Or maybe even prevent a number of vehicle owners from using their vehicle for a period of time, for the sake of the environment? Imagine walking out to your car in the morning to drive to your office, just for it to tell you “No, not today, please use public transportation instead”.
Digitalising and automating the mobility sector will not only pose big challenges to lawmakers. Businesses are also taking huge risks and making significant investments in technology, knowing that some of it may not make it to mass production.
As it is widely known, Tesla, the technology leader in autonomous electric vehicles is burning billions of dollars. Still, in 2019 Tesla produced only 367.500 vehicles, next to nothing compared to the world’s large-scale vehicle manufacturers. They churn out between 7 and 10m units a year.
With so much happening in the mobility sector, manufacturers have a lot on their plate. Simultaneously, they integrate new components into vehicle systems, they develop algorithms for autonomous driving, and they work with electrification of the powertrain.
Adding to that, the challenges of electrification of the powertrain and autonomous driving is attracting significant investment from companies outside the vehicle sector. New companies focusing on either developing algorithms for autonomy or technology for electrification are emerging. And as profitability in the vehicle industry is slowly shifting from metal and mechanics to software and services, there may very well be a new Ford or a new Toyota among them. The future may see completely new business cases in the vehicle industry, shifting from a car brand as we know it to a service provided by a nondescript shell on wheels.
Truly, these are exciting times in the mobility sector.