Causality-based Transfer of Driving Scenarios to Unseen Intersections

TL;DR

The paper addresses generating realistic driving scenarios for unseen intersections by transferring movement patterns through a causal Bayesian-network framework. It combines knowledge-driven parameterization (via a six-layer model and ontologies) with data-driven learning (Greedy Hill Climbing and do-calculus) to identify and apply causal relationships between scenario parameters. Evaluation on the inD dataset shows that sampled parameter configurations can yield plausible trajectories at unseen intersections and reveal infrastructural influences such as construction sites. The approach offers a path toward scalable scenario generation with reduced data needs, though it relies on adequate data diversity and careful discretization to ensure generalization across varied urban layouts.

Abstract

Scenario-based testing of automated driving functions has become a promising method to reduce time and cost compared to real-world testing. In scenario-based testing automated functions are evaluated in a set of pre-defined scenarios. These scenarios provide information about vehicle behaviors, environmental conditions, or road characteristics using parameters. To create realistic scenarios, parameters and parameter dependencies have to be fitted utilizing real-world data. However, due to the large variety of intersections and movement constellations found in reality, data may not be available for certain scenarios. This paper proposes a methodology to systematically analyze relations between parameters of scenarios. Bayesian networks are utilized to analyze causal dependencies in order to decrease the amount of required data and to transfer causal patterns creating unseen scenarios. Thereby, infrastructural influences on movement patterns are investigated to generate realistic scenarios on unobserved intersections. For evaluation, scenarios and underlying parameters are extracted from the inD dataset. Movement patterns are estimated, transferred and checked against recorded data from those initially unseen intersections.

Figures (8)

• Figure 1: Methodology for scenario setup
• Figure 2: Flowchart of parameter handling
• Figure 3: Exemplary intersections from inD dataset inDdataset
• Figure 4: Complete causal Bayesian network. The darker the node, the less equally distributed are the occurring parameter values.
• Figure 5: Causal effect of Conflict on deviation from lane centerline