Expert-LaSTS: Expert-Knowledge Guided Latent Space for Traffic Scenarios

TL;DR

This paper introduces Expert-LaSTS, a novel expert-knowledge guided representation learning method for traffic scenarios, enabling effective clustering and detection of new scenario types without manual labeling.

Contribution

It presents a new framework that incorporates expert knowledge into the latent space learning process for traffic scenario analysis, improving clustering and novelty detection.

Findings

Outperforms baseline methods in clustering accuracy

Enables automatic mining of traffic scenarios without manual labels

Provides extensive analysis of the learned latent space

Abstract

Clustering traffic scenarios and detecting novel scenario types are required for scenario-based testing of autonomous vehicles. These tasks benefit from either good similarity measures or good representations for the traffic scenarios. In this work, an expert-knowledge aided representation learning for traffic scenarios is presented. The latent space so formed is used for successful clustering and novel scenario type detection. Expert-knowledge is used to define objectives that the latent representations of traffic scenarios shall fulfill. It is presented, how the network architecture and loss is designed from these objectives, thereby incorporating expert-knowledge. An automatic mining strategy for traffic scenarios is presented, such that no manual labeling is required. Results show the performance advantage compared to baseline methods. Additionally, extensive analysis of the latent space is performed.