This repository is a curated list of research papers and open-source resources focused on knowledge-driven autonomous driving (AD). It serves as a comprehensive, continually updated reference for researchers and practitioners in the field, aiming to track the frontier of knowledge-driven AD advancements.

How It Works

The project categorizes resources by key components of knowledge-driven AD, including papers, datasets, benchmarks, environments, and driver agents. It highlights various knowledge forms used, such as object referral, gaze heatmaps, advice, and question answering, across different tasks like vehicle control, motion planning, and scene captioning. The list also includes links to relevant simulators and tutorials.

Quick Start & Requirements

This repository is a curated list and does not have a direct installation or execution command. Users are expected to follow the links provided for individual papers, datasets, and code repositories to set up and run specific projects. Prerequisites will vary significantly depending on the linked resources.

Highlighted Details

• Extensive collection of papers and projects on knowledge-driven autonomous driving.
• Detailed tables categorizing datasets by sensors, knowledge forms, tasks, and metrics.
• Includes a comprehensive list of simulators and driver agent research.
• Features a dedicated section for survey papers and tutorials on the topic.

Maintenance & Community

The repository is actively updated, with recent additions in late 2023 and early 2024. It welcomes contributions via Pull Requests. A citation is provided for the project's survey paper.

Licensing & Compatibility

The "Awesome Knowledge-driven Autonomous Driving" repository itself is released under the Apache 2.0 license. Individual linked resources may have different licenses, requiring users to check each project's specific terms.

Limitations & Caveats

As a curated list, the repository does not provide direct functionality. Users must navigate to external links for code, datasets, and simulators, each with its own setup requirements and potential limitations. The rapidly evolving nature of the field means resources may become outdated.