In the EU AdaptIVe (www.adaptive-ip.eu) and L3Pilot (www.l3pilot.eu) projects, we made a distinction between "driving scenario", "driving situation", and "traffic scenario" as follows:

A driving scenario is, "The abstraction and the general description of a driving situation without any specification of the parameters of the driving situation. Thus, it summarises a cluster of homogenous driving situations. Driving scenarios are typically short in time (t < 30 s) and only a few vehicles are involved. An example is lane change to the left lane."

A driving situation is, "A specific driving manoeuvre (e.g. a specific lane change with defined parameters). Thus the driving situation describes in detail a driving scenario e.g. a lane change at 60.8 distance of 10 m behind the host vehicle in the adjacent lane and with a velocity of 65 km/h."

On the other hand, "A traffic scenario describes a larger traffic context, which includes different (not pre-defined) driving scenarios. Typically in a traffic scenario a large number of vehicles is analysed over a longer time period. An example of a traffic scenarios could be on a three-lane motorway section with ten highway entrances and exits, a speed limit of 130kph for a period of one hour."

I have attached the source project deliverable, which can be referenced as:

• Hibberd D, Louw T, Aittoniemi E, Brouwer R, Dotzauer M, Fahrenkrog F, Innamaa S, et al. (2018). From Research Questions to Logging Requirements. L3Pilot Deliverable D3.1.

Dear Zhigang Xu,

I suggest you to see links and attached file on subject.

-Abstract Keywords

www.aomartin.co.uk/sds-repository/downloads/beattie1998self.pdf

-Fine-Tuning VoBroadband Test Capabilities | EDN

https://www.edn.com › Design Center

Best regards

Thanks so much!

Zhigang Xu I think this may be an English language issue. The word "scenario" (note the "c" in "scenario" is slightly different than what you have written but I think is what you meant) is sometimes defined as "a postulated sequence or development of events" (dictionary definition).

When a scientist or researcher presents "scenarios" in a research project, it refers to the series of events or types of events that occur in the study. This is why the word is used so frequently, and, why it is never defined.

So, as an example, if I conducted a study on users operating a car, part of the scenario is what the people are asked to do and when they are asked to do it. (that is the 'scenario' for the users). The other aspects might be the conditions of experiment itself (let's say, automated or non-automated vehicle). Both together are still the 'scenario'.

I hope this helps!

I think it depends of the context, whether you are talking about driver or vehicle context.

Thank you!

I guess it implicates a prospective, maybe a speculative view. It represents more than an observed situation onto roadways. Therefore the definition of future boundary conditions seems to me to be essential. That could be technological assumptions concerning the vehicle based functionalities as well as behavioural acts of the other traffic participants which are not automated equipped. Additionally it depends on different national or regional regulations in respect of traffic organisation. But, this is my definition based on our work on this topic.

I think you started a interesting discours!

Best regards! Heinz

I am sorry, I misspell the word, it is "scenarios"

I found an interesting explanation of scenarion as below.

"A scenario combines the actions of the ego vehi- cle, the static environment (e.g. infrastructure and weather), and the ongoing activity of the dynamic environment (in- cluding the other traffic participants) for a certain period of time. Typically the duration of a scenario is of the order of seconds . The end of a scenario is the first situation that is irrelevant to the scenario. It is assumed that a scenario can be completely constructed from the sensory output of the ego vehicle, possibly extended with external information like roadmaps and weather data. No information of the drivers actions or intentions is considered, since the scenarios are automatically generated from cars driving on the public road"

[1] E. De Gelder and J. P. Paardekooper, “Assessment of Automated Driving Systems using real-life scenarios,” IEEE Intell. Veh. Symp. Proc., no. Iv, pp. 589–594, 2017.

A scenario describes the temporal development between several scenes in a sequence of scenes. Every scenario starts with an initial scene. Ac- tions&events as well as goals&values may be specified to characterize this temporal development in a scenario. Other than a scene, a scenario spans a certain amount of time.

Scenes in a scenario are linked by actions and events. According to figure 7, a scenario is a single path of a temporal sequence of actions&events (edges) and scenes (nodes) out of the tree representing the entirety of all possible future scenarios for a given initial scene. Other than scenes, a scenario spans a certain amount of time. A scenario needs to include at least one (initial) scene and actions&events to fully specify a path in figure 7. However, a scenario may also be specified by a complete set of scenes, while the actions and events just covering the elapse of a specified time.

[1] S. Ulbrich, T. Menzel, A. Reschka, F. Schuldt, and M. Maurer, “Defining and Substantiating the Terms Scene, Situation, and Scenario for Automated Driving,” IEEE Conf. Intell. Transp. Syst. Proceedings, ITSC, vol. 2015–October, no. September, pp. 982–988, 2015.

I agree with the definition you cited. If we have a consensus about that. The next step would be to discuss the use cases of such scenarios.

• Are they good for a better understanding of daily traffic?
• Could they serve for prevention of accidents?
• Are they an appropriate tool proving functionalities of automated vehicles?
• Are they a methodical approach to cope with the complexity of triviality ("silly events" between cars) occuring on our road-network?
• Could they be helpful for roadway design planning? ...and so on!

• Are they good for a better understanding of daily traffic
• ----Re: Yes, when we decompose daily traffic into simple scenarios, it is easier to understand.
• Could they serve for prevention of accidents?
• -----Re: we can in-depth analyse the natural scenarios and get the spatio-temporal distribution and statistical paramters of accidents, so in the road design or safety improvement term, we can make some changes.
• Are they an appropriate tool proving functionalities of automated vehicles?
• -----Re: Yes, most professinals achieve consensus on this issue and build their own scenario database.
• Are they a methodical approach to cope with the complexity of triviality ("silly events" between cars) occuring on our road-network?
• -----Re: Maybe, it is one of useful tools, but should not be the only one.
• Could they be helpful for roadway design planning? ...and so on!
• -----Re: Certainly.

On the internal conbustion engines test beds, the scenario test is the manner wich the engine is excited by engine command parameters: regime-load-acceleration in the test procedure.

A scenario not only refers to the actions&events, but also the entities that links with the actions&events.

Scenario-based tests, such as DARPA Grand Challenge and DARPA Urban Challenge, just require an autonomous vehicle to pass a special region safely within a limited time (DARPA Grand Challenge and DARPA Urban Challenge 2004–2007; Buehler et al. 2009; Campbell et al. 2010).

I used the term "scenarios" for a lot of times in my papers about intelligent/autonomous/automated vehicle testing. Thanks for your questions. I will make a more clear definition in my next paper on this issue.

Li Li, Yi-Lun Lin, Nan-Ning Zheng, Fei-Yue Wang, Yuehu Liu, Dongpu Cao, Kunfeng Wang, Wu-Ling Huang, "Artificial intelligence test: A case study of intelligent vehicles," Artificial Intelligence Review, vol. 50, no. 3, pp. 441-465, 2018.

Li Li, Wu-Ling Huang, Yuehu Liu, Nan-Ning Zheng, Fei-Yue Wang, "Intelligence testing for autonomous vehicles: A new approach," IEEE Transactions on Intelligent Vehicles, vol. 1, no. 2, pp. 158-166, 2016.

Thanks！

In the EU AdaptIVe (www.adaptive-ip.eu) and L3Pilot (www.l3pilot.eu) projects, we made a distinction between "driving scenario", "driving situation", and "traffic scenario" as follows:

A driving scenario is, "The abstraction and the general description of a driving situation without any specification of the parameters of the driving situation. Thus, it summarises a cluster of homogenous driving situations. Driving scenarios are typically short in time (t < 30 s) and only a few vehicles are involved. An example is lane change to the left lane."

A driving situation is, "A specific driving manoeuvre (e.g. a specific lane change with defined parameters). Thus the driving situation describes in detail a driving scenario e.g. a lane change at 60.8 distance of 10 m behind the host vehicle in the adjacent lane and with a velocity of 65 km/h."

On the other hand, "A traffic scenario describes a larger traffic context, which includes different (not pre-defined) driving scenarios. Typically in a traffic scenario a large number of vehicles is analysed over a longer time period. An example of a traffic scenarios could be on a three-lane motorway section with ten highway entrances and exits, a speed limit of 130kph for a period of one hour."

I have attached the source project deliverable, which can be referenced as:

• Hibberd D, Louw T, Aittoniemi E, Brouwer R, Dotzauer M, Fahrenkrog F, Innamaa S, et al. (2018). From Research Questions to Logging Requirements. L3Pilot Deliverable D3.1.

@ Tyron Louw Thank you so much! Now the definition of scenario is clearer and clearer. Thanks to researgate giving us this great platform.

I think the given definitions by Hibberd et al. and by Ulbrich et al. seem very reasonable to find a common language to speak about use cases in automation etc.

However, I think one should be careful in stating values on how long a scenario is, without being more specific. This might limit the scope of the definition of a “scenario”.

In traffic (not just automation), when we speak of “situations” or “scenarios”, we use terms like parameters, aims, intention, action, development, change,… These terms refer to states and processes (change of states). All states underlie temporal variability and depending on the amount of change, we may consider a specific time period as a different situation. Personally, I would promote not thinking of situation as “bins of time” but as a “multi-scale slice of time”. For example, as stated by Tyron Louw, a lane change situation may be defined by travel speed, distances to other vehicles, relative speed etc. How many of such parameters are needed to fully describe the “situation” and predict the behavior of vehicles? May we also include the number of lanes, time of day, weather, country? I think there is a smooth transition between “driving situation” and “traffic scenario” and all parameter expressions exist at once. For example, a driver may choose actions depending on a multitude of factors on different timescales like the momentary predicted sideways motion of the car (use steering wheel), the headway to the leading vehicle (prepare lane change), traffic density/weather (choose travel speed), available time (route choices). Additionally, there is loads of human factors on different timescales (driving experience, age, temperament, mood, distraction,…).

The point I am trying to make is, that we don’t only need “a situation is YX seconds long”. We need “factors XYZ have XYZ amount of influence on the course of action for XYZ period of time. (All other factors may be treated as noise, as they don’t give us enough information/predictive power.)”

@Matthias Powelleit It is a good idea that a driving situation should be divided into multi-scale slice of time.

Tyron Louw Nice and helpful description. Thank you!

I think it's about what are you looking for, I mean if you are talking about outside environment of the vehicle or inside environment of the vehicle. It also depends on what kind of experiment and/or analysis you are going to do or willing to do.

Hello, I agree that we have a multiple need for scenarios. A underestimated topic in the discourse is that the vehicle needs for steering in the autonomous driving mode an internal database of scenarios comparable with the experience and empirical routine of a human driver. Detection instruments will not be efficient enough. If the automated functions fail the take-over back to driver is a kind of helpless solution.