1st International Workshop on Software Engineering for Autonomous Driving Systems (SE4ADS 2025)SE4ADS 2025

About SE4ADS

Software Engineering research in Autonomous Driving Systems (ADSes) faces some research community-oriented challenges because of the need to use specific hardware to install a complex system (i.e., the ADS) and run necessary tools (e.g., a simulator). Due to such challenges, reproducing research results becomes especially difficult, and newer approaches often either only compare against a random baseline as opposed to the original implementations of state-of-the-art approaches, or must compare against a re-implementation that introduces threats to validity regarding the fidelity of the implementation with respect to the original approach’s design. Moreover, re-implementing prior research work wastes time and resources because it requires researchers to repeat each other’s efforts and, in turn, hinders new opportunities for potential breakthroughs.

The main theme of this workshop, which we refer to as SE4ADS, is the exchange of ideas regarding the establishment of a community-wide infrastructure for facilitating research in the area of software engineering for autonomous driving systems. In support of that theme, SE4ADS provides a forum for practitioners and researchers to (1) share ideas and potential solutions regarding tools, libraries, benchmarks, and datasets that should belong to the infrastructure; (2) explore issues and challenges related to such research; (3) discuss mechanisms for enabling and facilitating tool availability, reusability, and interoperability in that research area; and (4) determine solutions for replicating or reproducing experiments and analyses in the workshop’s target research area. SE4ADS will help forge a new research community and create new collaborations.

The goals of the workshop are as follows:

1. Converge on requirements and challenges for a self-sustainable, community-based research infrastructure to support software engineering for ADS. An increasingly growing number of tools and methodologies have been produced for designing, implementing, testing, analyzing, and maintaining ADSes. These tools and methodologies may have mismatched assumptions; be inaccessible, unsupported, or unmaintained while still implementing ideas useful for researchers or practitioners; or may even support different perspectives on how to construct and maintain ADSes. One of the major goals of SE4ADS is to gather researchers and practitioners together to obtain different views as to how to manage and unify these tools—and determine the best means for converging the most useful tools and datasets among them to produce a community-wide infrastructure to facilitate reusable, replicable, and reproducible software engineering research for ADS.

2. Uniting the software engineering for ADS community. The workshop will provide the opportunity for software engineering researchers who study ADSes, as well as educators and industrial practitioners, to build a community that leverages both novel and previously existing software engineering tools, techniques, and datasets in order to tackle problems that are slowing or blocking progress for the software engineering for ADS community.

3. Construct a repository of ADS-specific baselines, benchmarks, and datasets. As part of a shared community-wide infrastructure, workshop participants will discuss issues regarding the need to construct and maintain baselines, benchmarks, and datasets containing bug-fix pairs, reusable test cases, and reusable driving scenarios. Through the workshop, participants will aid in the generation and subsequent launch of a community-wide standard of ADS artifacts that enable meaningful, objective comparison of research techniques.

4. Determine mechanisms needed to support ADS construction and maintenance for industrial researchers and practitioners. The immediate needs of academic researchers and educators do not necessarily coincide with or support the needs of industry practitioners and researchers. To address this issue, a major goal of SE4ADS is to solicit feedback from industrial practitioners and researchers to determine the mechanisms, features, and desirable properties of a community-wide infrastructure for ADS-oriented software engineering.

Call for Papers

The International Workshop on Software Engineering for Autonomous Driving Systems invites submissions for 5 types of papers: (1) position papers, (2) industrial papers, (3) experience reports, (4) education and training papers, and (5) artifact papers. Position papers will be up to 4 pages in length, while the other types of paper will be up to 7 pages in length. Position papers will include opinions and ideas about the infrastructure—including its designs or requirements—and how its goals can be achieved. Artifact papers can describe any artifacts that could be included in the infrastructure’s repository of baselines, benchmarks, or datasets—or tools that can serve as an initial set of tools incorporated into the infrastructure. The other types of papers may include experiences using a similar infrastructure as the one proposed, industrial-oriented experiences regarding architecture-based maintenance, and education and training challenges that may be addressed by the proposed infrastructure

Submissions must conform to the IEEE conference proceedings template, specified in the IEEE Conference Proceedings Formatting Guidelines (title in 24pt font and full text in 10pt type, LaTeX users must use \documentclass[10pt,conference]{IEEEtran} without including the compsoc or compsocconf options). Your submission does not need to be anonymous.

Primarily, we expect submitted papers to cover thoughts and opinions on how to design, construct, and maintain ADS testing approaches to enable empirical research in this area. A workshop overview (2 pages) and accepted papers will be published in the ICSE companion.

Topics

This workshop accepts (not limited to) the submissions of the following topics:

• Reusable and Extendable Simulation Environments
• Reproducible ADS Testing Approaches
• Reusable ADS Test Oracles
• Benchmark Datasets for ADS
• Mining ADS repositories
• Requirements engineering for ADS
• Software Architectures of ADS