Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Knowledge Graphs of Driving Scenes to Empower the Emerging Capabilities of Neurosymbolic AI

Ruwan Wickramarachchi, Cory Henson, Amit Sheth

TL;DR

DSceneKG (Driving Scenes Knowledge Graph), a suite of KGs of driving scenes that is built from real-world, high-quality scenes from multiple open autonomous driving datasets is introduced and its application in seven different tasks is highlighted.

Abstract

In the era of Generative AI, Neurosymbolic AI is emerging as a powerful approach for tasks spanning from perception to cognition. The use of Neurosymbolic AI has been shown to achieve enhanced capabilities, including improved grounding, alignment, explainability, and reliability. However, due to its nascent stage, there is a lack of widely available real-world benchmark datasets tailored to Neurosymbolic AI tasks. To address this gap and support the evaluation of current and future methods, we introduce DSceneKG -- a suite of knowledge graphs of driving scenes built from real-world, high-quality scenes from multiple open autonomous driving datasets. In this article, we detail the construction process of DSceneKG and highlight its application in seven different tasks. DSceneKG is publicly accessible at: https://github.com/ruwantw/DSceneKG

Knowledge Graphs of Driving Scenes to Empower the Emerging Capabilities of Neurosymbolic AI

TL;DR

DSceneKG (Driving Scenes Knowledge Graph), a suite of KGs of driving scenes that is built from real-world, high-quality scenes from multiple open autonomous driving datasets is introduced and its application in seven different tasks is highlighted.

Abstract

In the era of Generative AI, Neurosymbolic AI is emerging as a powerful approach for tasks spanning from perception to cognition. The use of Neurosymbolic AI has been shown to achieve enhanced capabilities, including improved grounding, alignment, explainability, and reliability. However, due to its nascent stage, there is a lack of widely available real-world benchmark datasets tailored to Neurosymbolic AI tasks. To address this gap and support the evaluation of current and future methods, we introduce DSceneKG -- a suite of knowledge graphs of driving scenes built from real-world, high-quality scenes from multiple open autonomous driving datasets. In this article, we detail the construction process of DSceneKG and highlight its application in seven different tasks. DSceneKG is publicly accessible at: https://github.com/ruwantw/DSceneKG

Paper Structure

This paper contains 16 sections, 2 figures, 1 table.

Table of Contents

  1. Introduction
  2. DSceneKG: Driving Scenes Knowledge Graph
  3. The Development of DSceneKG
  4. Driving Scenes Ontology
  5. Instantiating a Driving Scene KG from a Public Dataset
  6. Applications of DSceneKG: Emerging Neurosymbolic AI Capabilities
  7. Machine Perception
  8. Knowledge-based Entity Prediction
  9. Explainable Scene Clustering/ Typing
  10. Computing Semantic Similarity
  11. Knowledge Completion and Augmentation
  12. Semantic Search
  13. Causality
  14. Cross-modal Retrieval of Complex Data
  15. Conclusions
  16. ...and 1 more sections

Figures (2)

  • Figure 1: A driving scene from an AD dataset is instantiated as a subgraph in DSceneKG.
  • Figure 2: Applications of DSceneKG across seven different problems in AI