---
title: "Scenario Complexity Score"
slug: "scenario-complexity-score"
updated: 2026-03-31T22:26:47Z
published: 2026-04-01T12:30:17Z
canonical: "docs.safetypooldb.ai/scenario-complexity-score"
---

> ## Documentation Index
> Fetch the complete documentation index at: https://docs.safetypooldb.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Scenario Complexity Score

## **Overview**

The system gives each driving scenario a complexity score between 0 and 10, showing how simple or complex the situation is. It works with different types of data, so it can analyse scenarios from various sources. The complexity score ultimately measures: "*How challenging is this scenario for an autonomous driving system (ADS) to handle correctly without failing?*". The complexity score is essentially a proxy for "how much does this scenario stress-test the ADS?" ![](https://cdn.document360.io/d9bf0084-0d5b-468a-971b-537dce7153b6/Images/Documentation/image-1774870124555.png)

- For scenarios with fixed values (Concrete scenarios), it provides a single score.
- For scenarios with multiple possible variations(Logical scenarios), it provides a range of scores (minimum to maximum) to reflect different possibilities.

## How the score is calculated

The total complexity score is the average of three sub-scores, each measuring a different dimension of difficulty. Every sub-score runs from 0 to 10.

$Total = \frac{Quant + Dev + Variance}{3}$

### 

### Quantitative score (Quant)

Measures **how much is happening** in the scenario — the quantity and variety of elements present.

A scenario scores higher when it contains more actors (vehicles, pedestrians, cyclists), more types of road infrastructure, more hazardous surface or weather conditions, or unusual lane configurations. A busy urban intersection in heavy rain with multiple vehicle types will score much higher than a single car on a straight dry motorway.

> Think of this as: *"How crowded and varied is the environment?"*

### Deviation score (Dev)

Measures **how unusual or hazardous** the scenario is compared to everyday driving.

A scenario scores higher when it includes rare or dangerous elements — such as an accident cause (mechanical failure, animal intervention), an extreme hazard level (near-miss or fatal), adverse weather, poor visibility, or non-standard road surfaces. A scenario tagged with a fatal hazard and icy roads will score much higher than a routine sunny-day motorway drive.

> Think of this as: *"How far does this scenario deviate from normal, safe conditions?"*

### Variance score (Variance)

Measures **how much the actors move and change** throughout the scenario.

A scenario scores higher when vehicles travel at varying speeds, change position significantly, or perform many different manoeuvres across the scenario's timeline. A scenario where a vehicle accelerates, brakes sharply, and changes lanes repeatedly scores higher than one where everything moves at a constant speed in a straight line.

> Think of this as: *"How dynamic and unpredictable is the behaviour in this scenario?"*

### Score reference

| Total score | Complexity level | What it typically means |
| --- | --- | --- |
| 0 – 2 | Low | Simple, routine conditions. Minimal stress on the ADS. |
| 3 – 5 | Moderate | Mix of standard and mildly challenging elements. |
| 6 – 8 | High | Difficult conditions — adverse weather, dense traffic, hazardous behaviour. |
| 9 – 10 | Very high | Extreme scenario. Multiple compounding hazards or rare, critical events. |