We must prepare for all real-life autonomous driving scenarios


When I started to write my previous column on the relationship between autonomous and connected automation, I asked myself how much knowledge an autonomous vehicle would have to have and understand to complete an end-to end journey. The more I considered it, the more of a challenge it became. The situation is complicated by there being multiple levels of automation and varying road environments.

By definition, SAE has six levels (0-5) of automation. We can also identify five major road environments: multilane highways, single carriageways, rural roads, urban main roads and residential streets. The outcome is a 6×5 matrix of scenarios that need to be considered. However, that simplification may be illusory because the five main categories of road that I have identified can be further divided. In England (but not in the other countries in the UK) there are three varieties of motorway, depending on whether there is an emergency lane – and if there is, whether it can be used by traffic under certain conditions. Similar motorway arrangements exist in Germany and the Netherlands.

Rural road types can be segregated into dual carriageways; single carriageways; two-lanes; and two-lanes with an additional lane in the middle for overtaking – and this latter road type may be restricted to one direction or it may be used by both directions of traffic.

In urban areas there are further permutations, as well as additional facilities for buses, trams, cyclists and pedestrians. It is clear that there is a lot to know and work out for a single journey.

Into all this we are now bringing automated vehicles and it is clear that there will be many types. SAE levels categorize these types at a basic level, but each vendor will want to have its own differentiating factors. The behavior of vehicles operating at levels 3-5 is determined by their programming, but who determines what that programming will be?

Will a vehicle be dominant if it has to arbitrate with other vehicles when merging onto a motorway or emerging from a priority intersection? Will the vehicle strictly obey traffic regulations or will it exercise some degree of judgment depending on the situation?

It is not clear whether the attributes of ‘character’ have yet been explicitly defined and who will be responsible for determining them. It could be the role of governments to determine the limits of character, or they could be set by the manufacturer. Or it could be the vehicle user who selects the character of the automated driver they want. Further, with connected, automated vehicles there is also the potential for the road manager  to step in and externally influence the character of a group of vehicles on a certain stretch of road to improve the traffic flow situation for all vehicles.

The potential of automated vehicles and connected highways raises questions that have hitherto been seen as a matter of public education rather than actual traffic management. As we embark on what may prove to be a major shift in the way roads are used, we need to think about such matters. Once the scale of autonomous vehicle use has increased, it will be impossible to go back, and the result could be a flawed system that does not deliver fully on its promise.

Share this story:

About Author