Enabling Safe Autonomous Driving in Real-World City Traffic Using Multiple Criteria Decision Making

Furda, Andrei; Vlacic, Ljubo · 2011 · OpenAlex-citations

DOI: 10.1109/mits.2011.940472

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 real-time decision-making for autonomous vehicles operating in complex urban traffic environments. The authors identify that while sensor and perception technologies are advancing, high-level vehicle control—specifically the ability to select safe and appropriate driving maneuvers—remains a critical bottleneck. The research focuses on enabling driverless vehicles to make decisions that comply with traffic rules, ensure safety, and optimize efficiency using Multiple Criteria Decision Making (MCDM) techniques. The proposed solution employs a modular control software architecture comprising a Perception Subsystem, a World Model, a Route Planner, and a Real-Time Decision Making subsystem. The World Model aggregates data from sensors, prior maps, and vehicle-to-infrastructure communications to maintain an updated representation of the traffic environment. The decision-making process is decomposed into two sequential stages. Stage 1 determines the set of feasible, safety-critical driving maneuvers. This stage utilizes Petri nets to model operational logic, filtering maneuvers based on World Model events (e.g., obstacle presence) and route planner indications (e.g., upcoming turns). This approach ensures that only maneuvers compliant with traffic rules and current environmental conditions are considered, allowing for formal verification of safety-critical logic. Stage 2 selects the single most appropriate maneuver from the feasible set identified in Stage 1. This selection employs MCDM to balance competing objectives such as safety, comfort, and efficiency. The authors define a hierarchy of objectives broken down into measurable attributes (e.g., distance to obstacles, speed, waiting time). Utility functions map each driving alternative to these attributes, and a scoring method, specifically the Simple Additive Weighting Method, calculates a value for each alternative based on weighted attribute importance. The maneuver with the highest score is executed. The paper illustrates this with an example of passing a stopped vehicle, comparing alternatives based on speed and lateral distance against attributes like collision avoidance and boundary adherence. The significance of this work lies in its structured approach to decomposing complex autonomous driving decisions into manageable, verifiable components. By separating safety-critical feasibility checks (modeled via Petri nets) from optimization-based selection (modeled via MCDM), the system ensures robustness and safety while allowing for nuanced decision-making. The authors conclude that MCDM is suitable for this application, providing a flexible framework that can handle the multi-objective nature of urban driving. The use of external XML files for Petri net definitions further enhances system maintainability by decoupling logic changes from source code modifications.

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.

• situational awareness

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