AUTOMATION LEVELS & AV ARCHITECTURE

In less than 6 years we will have about 8M autonomous or partially autonomous consumer vehicles on the streets.

But, what makes a vehicle totally autonomous?

The Society of Automotive Engineers (SAE) defines 6 Levels of automation. The main difference between them is the implication the driver has in terms of control over the vehicle and in the dynamic driving tasks (DDT) such as event detection and response. In Level 0 the driver has full control of the vehicle while in Level 5 there is an absolute automation, there is no need of any human attention. The latter means that the driver doesn’t have to resume control under any circumstance because the system has situation awareness and decision-making capabilities and therefore will be able to respond in any environmental situation. AV technology needs to develop systems that perceive elements with spatial and time data and more importantly, to understand how human brain works and makes decisions.

Driving automation systems are categorized into these Levels based on three key points: (1) whether the vehicle performs DDT, (2) whether the vehicle performs DDT fallback and (3) whether the system is somehow limited by an Operational Domain Design (ODD). Let’s explain these last two concepts in more detail. DDT fallback covers all actions and manoeuvring required when the DDT fails, and the vehicle needs to reach a non-risky situation. ODD are operating conditions such as vehicles features that are designed to function under certain geographical or environmental restrictions and can limit the system’s autonomy functionality. Vehicles with a higher percentage of automation will have both DDT and DDT fall back assumed by the system and not be limited by any ODD.

Now that we know this, what are the pillars of AV architecture?

There are three basic components: perception, decision and action. Perception involves data recognition and collection using a variety of sources such as cameras, radars, GPS systems, etc.. This leads to journey planning, positioning and detecting any changes in relevant objects or in the environmental context that may affect the drive. Decision includes mission planning and behaviour planning: plan the route to get to the destination considering traffic or moving objects. Finally, actions are executed, the system actuates controlling the behaviour of the vehicle.

At the moment, there are no Level 5 vehicles, there are fully autonomous cars undergoing testing in several tech companies, but the truth is they are not present on the street. Level 4 cars are available but only in reduced areas due to legislation and infrastructure issues. Actually, Volvo and Baidu have teamed up to develop Level 4 vehicles mainly geared towards the taxi sector in China. The future of AV is quite promising, but customers need to feel safe and confident not only physically but also on the digital security scope.

Did you find any interesting points in this post? Are you missing anything? Let us know below in the comments!