We are hitting an inflection point in the automotive industry towards autonomous vehicles. This trend is fueled by three current sub-trends:

Connected cars
Automotive electrification
Improved safety

Autonomous cars will increase the complexity of automotive design.

The innovation created by these trends leads to more complexity, which puts considerable stress on engineers. The best way to ensure that all these complexities are accounted for, designed, verified and validated is systems engineering.

The 3 Trends Connected to the Development of Autonomous Vehicles

“In the near future and the coming future, you will see a significant change on the roads,” said Nand Kochhar, vice president of Automotive and Transportation Industry at Siemens Digital Industries Software. “The key differentiators are that all these vehicles are going to be connected… with different levels of autonomy (in terms of SAE’s Level 1 to Level 5). An even bigger change is going to be around electrification. A lot of the ICE (internal combustion engines) will be replaced by hybrids and fully electric vehicles.”

Connected cars, electrification and safety are paving the way towards autonomous cars.

Kochhar explains that these trends are borne out of a combination of consumer and societal pressures. For instance, consumers want to be constantly connected—able to continue whatever they are doing online as they move from their home to the car, to public transit.

“Everyone’s got a cellphone in their hand. If they are watching Netflix at home, they want to move on to the same thing in their car and continue to watch it,” he said.

The connected trend feeds automotive electrification—after all, a connected car is a more electrified car—but perhaps major pressure towards electrification comes from the movement toward a greener lifestyle at the corporate and individual level.

“For environmental safety, that gets into CO₂ emissions. That goes into switching to an electric power trains to address those emissions,” said Kochhar.

Electrification and connected cars are just the first step towards autonomy. These features will need safety and user experience elements which increase the intelligence of the car.

“Autonomy is driven by safety and user experience,” Kochhar explained. “If [consumers] are having a conversation [while] driving, in today’s environment, you want to be hands free and safe. Extremely busy people, when they get to their fully autonomous vehicle, they want to do their shopping and entertainment.”

“When you have an autonomous car, it needs to be connected anyways, because if you’re going to have full autonomy then the car needs to connect to city infrastructure so it can see the lighting system in the city—red, yellow, green—and make its decision after that,” he added.

Autonomous Cars: Where Are We Now?

Kochhar chaired the SAE Motor Vehicle Council when SAE released its definitions of the levels of autonomy. As a result, he is the perfect individual to give us an idea of where we currently are with the development of autonomous cars.

Some companies say they are at Level 3 autonomy; others say they are on the right path. A few are not talking publicly about their progress.

“Safely, we can say SAE Level 2 autonomy,” Kochar said. “I’ve personally driven and seen it. Most large OEMs are producing vehicles at Level 2 autonomy. That means you have a braking function and features like autonomous emergency braking or adaptive cruise control. And you will have several different versions, depending on the OEM, from a steering perspective. Whether it’s a lane departure warning, or lane assist—they all have different brand names. Those have all been maturing very well.”

He continued by noting that Level 3 autonomy is rare. At this point, it comes down to which brands are willing to claim it. Some say they are already at Level 3 autonomy; others say they are on the right path. A few are not talking publicly about their progress.

When pushed for an answer, Kochhar said, “Some are close to Level 3, some will claim they are at Level 3.” This may not sound like an endorsement; however, keep in mind that this is more about the fuzzy borders between the autonomy levels than anything else.

“Level 2 is becoming very standard in my mind because there is another thing driving. Some aspects of autonomy are also becoming a part of the certification programs. A lot of OEMs have committed to have these features standardized by 2022. Others will be taking a little longer. Once it’s standard, from a certification perspective, things mature rapidly,” Kochar explained.

He continued to describe how as Level 3 has more control over the vehicle, the question becomes more about policy, responsibility and accountability. As a result, not every company is willing to announce where they are on the journey to autonomy. Once these features become standardized, certified and codified into law and insurance policies, it will be safe to say we are at Level 3.

A similar process will happen as we move towards Level 4 and Level 5, Kochhar explained. “Who’s responsible? Who’s accountable? It’s where the other aspects come into play to get to Level 5—where you don’t have a steering wheel.”

How Systems Engineering Enables the Design of Level 4 and Level 5 Autonomous Cars

Systems engineering breaks down the silos in development by looking at each team and discipline that is working on a project through the lens of collaboration and engineering management. This defines the scope of the system, and all the problems within that scope that need solving. As a result, the more complex a problem is, the more vital it is to use systems engineering.

Model-based systems engineering (MBSE) makes autonomous cars possible. (Image courtesy of Siemens Digital Industry Software.)

As Kochhar puts it, “Systems engineering is a disciplined and structured way for approaching a problem. The reason it becomes important is because the automotive industry of today is already complex. But now when you add on things like autonomy, the complexity is going through the roof because you have a computer making decisions that normally a human would make.”

Each scenario to which an autonomous car’s AI system is trained to react makes the development more complex to design, verify and validate. At this complexity, siloed tools such as paperwork and spreadsheets break down.

“What you need is MBSE (model-based systems engineering) which can work from end-to-end, from concept to certification,” said Kochhar. “This starts to become the only way to handle these kinds of autonomous vehicle development.”

So, what would systems engineering development projects at various autonomy levels look like? The best way to get an idea is to see how the definition of the project’s scope changes between the varying SAE levels.

“For Level 2, you might say my system is the car, what’s on the road and the traffic scenario,” explained Kochhar. “So, adaptive cruise-control is a good example at that point.”

He added, “For Level 5, now you have a bigger system. You need to understand the city infrastructure, the road conditions and the weather conditions at a much deeper level. You might be connected to the city infrastructure beyond the traffic systems. The degree of complexity increases.”

Systems Engineering Thinking for Level-5 Autonomous Cars

As previously mentioned, a fully autonomous car must be a connected car. “[It’s] entire environment needs to change. For autonomous vehicles to drive around everything else around them, they need to be intelligent and connected. Otherwise, things won’t work the way we know it should work,” Kochar said.

An engineer working on the design of an autonomous car. (Image courtesy of Siemens Digital Industry Software.)

Not only will the design of the cars have its own internal logic, but it will also interface with the logic of all the connected systems around it. Traffic lights kilometers away might signal their status to the car. City weather stations might contact the car to ensure it is driving to the conditions of a developing storm. Even cellphones could notify the car of the locations of pedestrians around the road.

“There is a migration from traditional automotive technologies to the telecommunication, electronics and semiconductor market,” said Kochhar.

To ensure automotive companies can handle this orchestra of communications, they will need systems engineering. This will also help companies plan for scenarios where something in the orchestra is out of tune. In other words, systems engineering not only helps engineers realize areas that need to be designed for the system to work, it also defines the potential errors and redundancies to ensure the system is robust.

“Those are the kind of deep dive things, thinking and solutions needed, in my mind, to get to Level 5,” said Kochhar. “With your failsafe systems, from a mechanical and software standpoint, you have to make sure they work. That’s where a lot of the standards come into play.”

And as Kochhar already mentioned, once these systems end up in the realm of standards and certification, companies will start to announce the existence of fully autonomous cars. It will be at this point that companies will be comfortable taking on the full responsibility of every decision their cars make—and the only way they will reach that level of confidence and certification is with systems engineering.

To find out more about the relationship between systems engineering and autonomous cars, visit Siemens.