Improving Explainable Object-induced Model through Uncertainty for Automated Vehicles
Open Access
- 11 March 2024
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 443-451
Abstract
The rapid evolution of automated vehicles (AVs) has the potential to provide safer, more efficient, and comfortable travel options. However, these systems face challenges regarding reliability in complex driving scenarios. Recent explainable AV architectures neglect crucial information related to inherent uncertainties while providing explanations for actions. To overcome such challenges, our study builds upon the "object-induced" model approach that prioritizes the role of objects in scenes for decision-making and integrates uncertainty assessment into the decision-making process using an evidential deep learning paradigm with a Beta prior. Additionally, we explore several advanced training strategies guided by uncertainty, including uncertainty-guided data reweighting and augmentation. Leveraging the BDD-OIA dataset, our findings underscore that the model, through these enhancements, not only offers a clearer comprehension of AV decisions and their underlying reasoning but also surpasses existing baselines across a broad range of scenarios.Keywords
All Related Versions
This publication has 19 references indexed in Scilit:
- Explainability of Deep Vision-Based Autonomous Driving Systems: Review and ChallengesInternational Journal of Computer Vision, 2022
- Driving behavior explanation with multi-level fusionPattern Recognition, 2022
- Why Not Explain? Effects of Explanations on Human Perceptions of Autonomous Driving2021
- Information Aware max-norm Dirichlet networks for predictive uncertainty estimationNeural Networks, 2021
- BDD100K: A Diverse Driving Dataset for Heterogeneous Multitask Learning2020
- Explaining Autonomous Driving by Learning End-to-End Visual Attention2020
- Look who’s talking now: Implications of AV’s explanations on driver’s trust, AV preference, anxiety and mental workloadTransportation Research Part C: Emerging Technologies, 2019
- Stop explaining black box machine learning models for high stakes decisions and use interpretable models insteadNature Machine Intelligence, 2019
- Deep Object-Centric Policies for Autonomous Driving2019
- End-to-End Learning of Driving Models from Large-Scale Video Datasets2017