The evolutionary path to the autonomous car

Evolution vs. disruption

There are generally different approaches to developing innovations. The classic form of innovation is a step-by-step development. We also know this as natural evolution. In contrast to this is the radical transformation in the form of a disruption, which could also be described as a revolution.

We also find these two approaches in the automotive sector. While traditional carmakers such as Volkswagen, Mercedes or BMW, with their decades of experience, are relying on a gradual evolution, newcomers such as Tesla are trying to change the market disruptively. This applies to both electrification and autonomisation.

Both approaches have their pros and cons. The disruptive approach is ahead in terms of pace of development, while the step-by-step evolution is ahead in terms of safety. The experience helps the traditional manufacturers, while the newcomers have no legacy issues that could slow them down. Ultimately, there is no wrong or right approach on the way to the same goal. New manufacturers cannot make up the experience gap, while traditional manufacturers cannot simply start from scratch.

However, the experienced manufacturers do have a certain advantage: the most important requirement for autonomous driving is safety. And that speaks more in favour of the evolutionary approach.

Outlook: What does the future hold?

The two highest levels of autonomous vehicles, which are designed to manage complex situations without any human guidance or control, have not yet been fully developed. Although the first vehicles that can operate completely autonomously already exist, they can only do so in strictly defined scenarios. In reality, however, there are so many different and sometimes highly complex parameters to consider that the technical systems are still overwhelmed.

Here are some of the parameters that an autonomous vehicle must take into account:

• Structural traffic situation: road surface, traffic signs, junctions, traffic lights
• Dynamic traffic situation: other vehicles, cyclists, pedestrians
• Action parameters: accelerating, braking, steering

These ten parameters alone result in a multitude of highly complex scenarios for autonomous driving, all of which have to be tested and resolved with the utmost safety in everyday practical situations. Particular attention must be paid to those parameters that are subject to human influence, such as the behaviour of other vehicles.

A look at the accident statistics in German traffic provides some insight. The most common cause of accidents is ‘errors when turning, turning, reversing, starting and moving off’, followed by ‘failing to observe the right of way’. Autonomous vehicles could greatly reduce this type of error, but they would be powerless against other vehicles taking the right of way.

When it comes to structural traffic parameters, carmakers also depend on the federal, state and local governments, which can support them by making infrastructure adjustments. One important project here is C-Roads, in which authorities and road operators are promoting the introduction, testing and harmonisation of cooperative intelligent transport systems and services (C-ITS) on European roads.

This kind of collaboration could provide a blueprint for autonomous vehicle manufacturers. With common standards, (partially) autonomous vehicles could communicate with each other optimally and merge into an overall system. This would further reduce complexity and drastically increase safety with each additional participant.

Text: Falk Hedemann