Utilizing Background Knowledge for Robust Reasoning over Traffic Situations

TL;DR

This paper explores text-based methods leveraging commonsense knowledge and knowledge graphs for zero-shot reasoning in traffic situations, introducing new datasets and evaluating multiple knowledge-driven approaches.

Contribution

It presents three novel knowledge-driven approaches for zero-shot traffic reasoning and introduces two new datasets for causal and domain knowledge evaluation.

Findings

Unified-QA achieves top performance on causal reasoning dataset

Language models with inference and commonsense knowledge excel at cause-effect prediction

DPR+Unified-QA performs best on human-driving QA sets

Abstract

Understanding novel situations in the traffic domain requires an intricate combination of domain-specific and causal commonsense knowledge. Prior work has provided sufficient perception-based modalities for traffic monitoring, in this paper, we focus on a complementary research aspect of Intelligent Transportation: traffic understanding. We scope our study to text-based methods and datasets given the abundant commonsense knowledge that can be extracted using language models from large corpus and knowledge graphs. We adopt three knowledge-driven approaches for zero-shot QA over traffic situations, based on prior natural language inference methods, commonsense models with knowledge graph self-supervision, and dense retriever-based models. We constructed two text-based multiple-choice question answering sets: BDD-QA for evaluating causal reasoning in the traffic domain and HDT-QA for measuring the possession of domain knowledge akin to human driving license tests. Among the methods, Unified-QA reaches the best performance on the BDD-QA dataset with the adaptation of multiple formats of question answers. Language models trained with inference information and commonsense knowledge are also good at predicting the cause and effect in the traffic domain but perform badly at answering human-driving QA sets. For such sets, DPR+Unified-QA performs the best due to its efficient knowledge extraction.