Autonomously driven vehicles (AVs) and humans are poised to share the roads for the foreseeable future, leading to complex interactions between AVs, human driven vehicles (HVs), pedestrians, etc. To ensure road safety and coordination, AVs and HVs must adhere to shared legal frameworks encompassing the rules of the road, the duty of care, and liability considerations. This shared framework requires autonomous and human agents to anticipate and cooperate with respect to one another’s actions and responsibilities, fostering an environment of mutual understanding and accountability.

The foundational assumption is that a high-level, universally applicable model of traffic rules should govern both humans and machines. This model would enable both parties to draw similar inferences, maintain consistent expectations, and behave in compatible ways on shared roads.

At this workshop, we will delve into the issue of representation of traffic rules and driving conduct in ways that enables machines to execute them autonomously, in a way that is understandable and explainable to humans, so that the authorities can verify compliance. A key tenet of this discussion is that liability and negligence should be applied uniformly to all road users. Human drivers should not be unfairly penalized due to an AV’s superior perceptual and computational capabilities.

This focus on traffic rules arises from their practical importance—human drivers are required to explicitly learn and demonstrate mastery of these rules. Representing these rules for machines, however, is fraught with challenges, including gaps in data, highly parameterized contexts, commonsense, dependencies on situational factors, and ambiguities inherent in legal language. For example, concepts like “safe gap” are inherently open-textured and context-dependent, while object modeling (e.g., understanding long trucks) and mental models of other road users add layers of complexity. Additionally, the prescriptive nature of traffic rules, combined with their interplay with broader legal principles like liability and negligence, further complicates the task. These issues open multiple, challenging, and intriguing lines of research.

The use of symbolic logic to formalize legal norms should be part of such a system, as such approaches hold promise for the automation of legal reasoning. Significant challenges are still open in the field of logic rule modelling, such as the acquisition bottleneck, but new technological advancements can assist in these tasks. One such advancement is the use of language models to automatically parse the natural language in a logic representation in a formal language and to validate the correctness of the resulting logic structure. Other developments should be brought to the fore.

The workshop aims to tackle these challenges by focusing on the use of formal languages and models to represent traffic rules in machine-readable formats as well as the integration of Large Language Models to aid the extraction of rules and regulations. Furthermore, the workshop aims to showcase the different approaches to the issue and their use in applying legal reasoning to the AV behaviour.

Submission Guidelines

All papers must be original and not simultaneously submitted to another journal or conference. The following paper categories are welcome:

• Full papers (10 pages)
• Short papers (6 pages)

• Format: CEUR-WS (two column)

  • Overleaf template available at: https://www.overleaf.com/latex/templates/template-for-submissions-to-ceur-workshop-proceedings-ceur-ws-dot-org/wqyfdgftmcfw

List of Topics

• Knowledge Representation Methods: Exploration of techniques suitable for representing legal norms (especially traffic rules), including deontic logic, first-order logic, and other comparable formalisms. 

• Rule-Exceptions, Conflicts, and Contrary-to-Duty Obligations: Formalization and reasoning approaches to address exceptions to rules, conflicting regulations, violations, and secondary obligations or prohibitions that arise when other deontic specifications are violated.

• Abstract Legal Concepts and Principles: Representation of foundational legal principles such as "human dignity," "mutual respect," "care," "trust," and "danger." 

• Practical Implementation of Formalisations of the Law: Development and application of tools such as legal knowledge bases, ontologies, reasoning engines, and SAT-solvers to operationalize formal representations. 

• Translation of Legal Texts to Formal Representations: Methods—both automated and manual—for converting natural-language legal provisions into formal languages, focusing on traffic rules. 

• Legal and Engineering Challenges: Examination of the technical and legal hurdles associated with applying formal representations to real-world systems, especially in the context of traffic laws. 

• AI Compliance and Ethical Reasoning: Techniques for enabling AI systems to reason ethically and comply with traffic rules through formalized legal knowledge. 

• Machine Learning and Hybrid Approaches: Integration of machine learning methods for knowledge extraction and hybrid symbolic/sub-symbolic approaches for reasoning with formalized traffic rules. 

• Legal and Liability Concepts: Discussion on responsibility allocation for rule violations and the implications for sanctions and liability in mixed-traffic environments. 

• Interface Between Code and Normative Rules: Exploration of how legal rules can be embedded into code and the challenges of ensuring accuracy and fairness. 

• Verification and Validation of Rule Bases: Ensuring that formalized rules are correct, complete, and aligned with legal norms. 

• Cognitive Models of Driving: Analysis of mental models that guide human drivers and how these can inform the design of AV systems. 

• Use Cases and Demos: Presentation of practical examples and demonstrations showcasing the application of formalized legal norms in automated systems. 

• Philosophical Considerations: Discussions on meaningful human control, particularly in terms of interaction between humans and autonomous agents.

• Corpora: textual corpora which has been analysed and evaluated using computational linguistic techniques.

Publication

Accepted papers will be presented in the workshop and will be included in the workshop proceedings.

Venue

The conference will be held in conjunction with ICAIL25 (https://sites.northwestern.edu/icail2025) in Chicago at Northwestern University, June 16 to 20, 2025.

Contact

All questions about submissions should be emailed to:

• A.Z.Wyner@Swansea.ac.uk
• galileo.sartor@swansea.ac.uk