Behavior Taxonomies and Rulebooks for Safety-Assured Autonomous Driving

Overview

Bridging Regulation, Safety, and Autonomy

Despite significant advances in planning, learning, and verification, behavioral requirements for autonomous vehicles—derived from traffic regulations, safety standards, and operational contexts—remain fragmented, difficult to formalize, and hard to certify. Developing structured, machine-interpretable rulebooks that bridge regulatory intent and algorithmic implementation remains an open and pressing problem for both research and deployment.

This full-day workshop brings together leading researchers and practitioners from academia and industry to address the challenge of defining, structuring, and enforcing behavioral specifications for autonomous driving systems. Through a combination of invited talks, a paper session, and a panel discussion, the workshop aims to advance a shared understanding of behavior taxonomies, rulebook formalisms, and their role in enabling safety-assured autonomy.

The workshop culminates in a forward-looking research roadmap and cross-disciplinary dialogue on the path from behavioral requirements to certifiable, deployable autonomous systems.

Topics of Interest

Key Themes

Rulebook Formalisms & Semantics Formal representations of driving rules, lexicographic prioritization, compositional properties, and machine-interpretable specifications.

Behavior Taxonomies & ODD Modeling Structured classification of driving behaviors, operational design domain scoping, and competency-based behavior specifications.

Rule-Informed Planning & Control Integration of behavioral rules into trajectory planning, MPC, reinforcement learning, and end-to-end architectures.

Safety Verification & Validation Scenario-based testing, safety case construction, formal verification of behavioral compliance, and continuous monitoring.

Regulatory & Policy Frameworks Traffic law formalization, liability-aware behavior specification, compliance assessment, and cross-jurisdictional harmonization.

Metrics & Evaluation Behavioral violation scoring, performance measurement, data-driven evaluation methodologies, and benchmarking.

Invited Speakers

Andrea Censi, Ph.D.

ETH Zürich · Institute for Dynamic Systems and Control

Invited Talk

Co-Design of Rulebooks and Autonomous Vehicles

Rulebooks are one way to encode behavior specifications in a scalable manner for autonomous vehicles without overconstraining agents. There is a tradeoff among the number and complexity of the rules, the implementation cost in terms of sensors, actuators, and computation, and the achievable system performance. This talk will describe how to characterize this trade-off and the problem of optimizing the rules to increase performance and reduce cost while maintaining the same safety guarantees.

Jerry Lopez

Torc Robotics · Safety Assurance

Invited Talk

A Framework for Incorporating Traffic Rules Compliance into a Machine-Learned Trajectory Planner

End-to-end machine learning is gaining traction in the autonomous vehicle space, and there is a need to ensure that ML models generating trajectories can incorporate traffic rules into their learning framework. This talk highlights methodologies that Torc is employing to incorporate traffic rules compliance into an end-to-end reinforcement learning trajectory planner model.

Henk Goossens

TNO · Innovation Partnerships, Safety Assessment of Automated Driving

Invited Talk

The Rulebook Flywheel for Safe Automated Driving

This presentation addresses the challenge of specifying and verifying competent driving behavior. It covers the TNO position on the design and use of behavior rulebooks for the definition, automated evaluation, and continuous improvement of competent driving behavior. Behavior rulebooks formalize both traffic and driving rules, offering consistency, transparency, and a base for measurable refinement towards safe and responsible autonomous driving.

Soheil Sohrabi, Ph.D.

General Motors · Risk Assessment Engineering

Invited Talk

Continuous Scoring of AV Behavioral Violation

Behavioral metrics for autonomous vehicles are often framed as binary indicators of rule violations, which limits their ability to distinguish between minor deviations and truly hazardous behavior. This talk proposes a continuous, non-linear severity scoring framework grounded in Extreme Value Theory to better capture the spectrum of abnormal AV behavior relative to human driving, yielding a statistically principled scoring scheme that aligns severity with behavioral rarity.

Chaitanya Shinde

Torc Robotics · Safety Architecture & Standards

Invited Talk

From ODD to Action: A Competency-Based Taxonomy for Systematic Autonomous Driving Behaviors

This talk presents a competency-based taxonomy that bridges the gap between Operational Design Domain definitions and actionable driving behaviors for autonomous vehicles. Drawing on expertise in functional safety (ISO 26262, ISO 21448) and AI safety standards development at SAE International and IEEE, the talk outlines a systematic approach for specifying, decomposing, and validating autonomous driving competencies across diverse operational scenarios.

Accepted Papers

Workshop Paper Presentations

Workshop papers are peer-reviewed in accordance with IEEE regulations and published in IEEE Xplore as part of the IEEE IV Workshops Proceedings.

The Rulebook Flywheel for Safe Automated Driving

Henk Goossens, Chris van der Ploeg, Michiel Braat, Jan-Pieter Paardekooper, Hari Hara Sharan Nagalur Subraveti, Arturo Tejada, Olaf Op den Camp

This paper covers a position statement on the use of behavior rulebooks for the definition, automated evaluation and continuous improvement of competent driving behavior of automated vehicles. Behavior rulebooks formalize the traffic and driving rules for assessing the automated vehicles' actions, offering consistency, transparency, and a base for measurable refinement. The authors assert that such structured evaluation supports reliable and accountable behavior, emphasizing how rule-driven assessment can guide automated vehicles toward safe and acceptable performance in real-world settings.

Program

Workshop Schedule

Draft schedule — final times and session order to be confirmed.

08:30	Registration & Coffee
09:00	Opening Remarks Workshop organizers outline goals and structure
09:15	Invited Talk: From ODD to Action: A Competency-Based Taxonomy for Systematic Autonomous Driving Behaviors Chaitanya Shinde · Torc Robotics
10:00	Invited Talk: The Rulebook Flywheel for Safe Automated Driving Henk Goossens · TNO
10:45	Coffee Break
11:00	Invited Talk: A Framework for Incorporating Traffic Rules Compliance into a Machine-Learned Trajectory Planner Jerry Lopez · Torc Robotics
11:45	Invited Talk: Co-Design of Rulebooks and Autonomous Vehicles Andrea Censi · ETH Zürich
12:30	Lunch Break
13:30	Invited Talk: Continuous Scoring of AV Behavioral Violation Soheil Sohrabi · General Motors
14:15	Invited Talk: To Be Announced Speaker to be confirmed
15:00	Invited Talk: To Be Announced Speaker to be confirmed
15:45	Coffee Break
16:00	Panel Discussion Cross-disciplinary dialogue on the future of behavioral rulebooks for autonomous driving
17:00	Research Roadmap & Closing Remarks Forward-looking discussion and workshop synthesis
17:30	Workshop Concludes

Submission Information

Call for Papers

We invite researchers and practitioners to submit workshop papers on topics aligned with the themes of this workshop. Submitted papers will be peer-reviewed in accordance with IEEE regulations and published in IEEE Xplore as part of the IEEE IV Workshops Proceedings (separate from the main IEEE IV Proceedings).

Papers must be formatted according to the IEEE IV 2026 formatting guidelines (IEEE double-column format, up to 6 pages + 2 paid pages). Submissions should be made via PaperCept using the workshop code below.

Paper Submission

January 30, 2026

Acceptance Notification

February 28, 2026

Final Submission

March 15, 2026

Workshop Date

June 22, 2026

Submit Your Paper

Submit via PaperCept at its.papercept.net. Select "Submit a contribution to IV 2026", then choose "Workshop paper" as the submission type.

Workshop Code: 2w765

View Full Call for Workshop Papers →

Bridging Regulation, Safety, and Autonomy

Key Themes

Workshop Paper Presentations

Workshop Schedule

Call for Papers

Submit Your Paper

Workshop Organizers