Decision-Making for Automated Vehicles Using a Hierarchical Behavior-Based Arbitration Scheme

Orzechowski, Piotr F.; Burger, Christoph; Lauer, Martin · 2020 · OpenAlex-citations

DOI: 10.1109/iv47402.2020.9304723

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the challenge of behavior planning and decision-making for highly automated vehicles (AVs), which often rely on finite state machines (FSMs) that suffer from poor explainability, maintainability, and scalability. To overcome these limitations, the authors propose a hierarchical behavior-based arbitration scheme. This architecture combines modular, atomic behavior blocks with generic arbitrators to compose complex driving behaviors in a bottom-up manner. The framework is designed to be generalizing and scalable, capable of integrating various scenario-specific solutions, such as POMDPs or learning-based methods, into a single traceable system. The methodology defines behavior blocks as modular units containing invocation conditions (applicability), commitment conditions (continuation), and command functions (output). These blocks are combined using arbitrators that select behaviors based on abstract information without needing knowledge of the underlying implementation. The authors introduce several arbitration schemes, including priority-based, sequence-based, and a novel cost-based arbitrator that selects the option with the lowest expected cost, such as travel velocity. The system utilizes an abstract environment model and a twofold maneuver representation for structured (corridor-based) and unstructured (trajectory-based) environments. Specific behavior blocks are defined for urban driving (e.g., FollowEgoLane, ChangeLane), highway driving, parking, and emergency scenarios. The proposed architecture was evaluated using the CoInCar-Sim simulation framework on a 5.7 km test track in Karlsruhe, Germany, featuring urban segments, intersections, and a parking lot. The experimental setup employed a priority-based top-level arbitrator managing urban driving, parking, and safe stop fallbacks. Urban driving behaviors were selected via a cost-based arbitrator optimizing for expected velocity and routing costs. The simulation results demonstrated that the vehicle successfully executed the full route, including lane changes and parking, within 9 minutes and 40 seconds. The behavior selection log confirmed that the arbitrators correctly switched between behaviors based on invocation conditions and cost estimates, with the safe stop fallback activating appropriately when the vehicle left the route. The significance of this work lies in providing a flexible, modular framework for AV decision-making that improves upon traditional FSMs. By decoupling behavior logic from arbitration logic, the system enhances maintainability and allows for independent development and testing of behavior blocks. The hierarchical structure supports functional safety through formal verification of simple arbitrators and error confinement. The introduction of cost-based arbitration enables dynamic preference handling, while the modular design facilitates the integration of diverse planning algorithms, offering a robust foundation for scalable automated driving systems.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• driverless ads

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Theoretical Contribution: computational model