Operational Scenarios using the PEGASUS format - Part 1
- Sahil Subodh Bane
- Jul 13
- 2 min read
Why Scenarios Need a Format
You cannot test an automated driving function against "traffic." The real world offers an effectively infinite set of situations: every combination of road layout, weather, other vehicles, and split-second timing is its own test case. Scenario-based safety is the industry's answer to that problem, but it only works if scenarios are described in a structured, repeatable way. Without a common format, one engineer's "car cuts in ahead" is another's "aggressive merge in heavy rain," and coverage of the operating space becomes a guess rather than a claim.
The PEGASUS project gave the industry that structure. Two of its ideas in particular, a layered model of scenario content and a set of abstraction levels, have become the backbone of how operating scenarios are described today.
What is PEGASUS
PEGASUS was a German research project, funded to answer a few hard questions about Autonomous driving
How to standardize a procedure for testing Automated vehicle systems?
How to develop a tool chain to safeguard automated driving?
How to create cross-manufacturer methods for safeguarding highly automated functions?
Out of that work came a structured way to describe the driving environment and a vocabulary for scenarios at different levels of precision. They have since spread into standards such as ASAM OpenSCENARIO and ISO 34501. This post explores the format proposed by the Pegasus project.
The Six-Layer Model: What's in a Scene ?
The PEGASUS project describes a structured way of representing every operating scenario in the form of a six-layer model, where each layer describing a certain aspect. Take an example of a scenario of a vehicle driving on a highway when there is snow. This is also shown in the figure below
Layer 1, Road. The geometry and topology of the road itself: number of lanes, curvature, gradient, lane markings. For our example, a straight 2-lane highway.
Layer 2, Traffic infrastructure. The fixed furniture of the road: signs, barriers, traffic signals, guardrails etc. In our example here, this includes speed limits, a lane divider and buildings on either side of the road (although building don't matter so much to vehicle dynamics)
Layer 3, Temporary modifications. Changes laid over layers 1 and 2: construction sites, cones, temporary bollards. Worth noting that although this layer was defined in the PEGASUS concept, it was not actually implemented within the original project scope.
Layer 4, Movable objects. All traffic participants and their maneuvers: other vehicles, motorcyclists, vulnerable road users. In our example, there is a car in front at a speed of 45m/s and a relative distance of 40 m from the ego vehicle.
Layer 5, Environment. Conditions that influence the other layers: weather, lighting, road friction.
Layer 6, Digital information. V2X messages, HD-map availability, and other digital data the vehicle can draw on. In the original model this layer was added to the earlier five specifically to cover connected and map-based information.
The power of the model is that it forces completeness. A scenario missing its environment layer or its digital layer is not a simpler scenario, it is an underspecified one.