Legally-Guided Automated Decision-Making System Using Language Model Agents for Autonomous Driving

被引:0
作者
Wang, Ya [1 ,2 ]
Barta, Dainel [1 ]
Hesse, Julian [1 ]
Buchwald, Philip [1 ]
Paschke, Adrian [1 ,2 ]
机构
[1] Fraunhofer Inst Open Commun Syst, Kaiserin Augusta Allee 31, D-10589 Berlin, Germany
[2] Free Univ Berlin, Kaiserswerther Str 16-18, D-14195 Berlin, Germany
来源
RULES AND REASONING, RULEML+RR 2024 | 2024年 / 15183卷
关键词
Neurosymbolic System; Large Language Model Agent; Ontological Reasoning; Rule Compliance; Autonomous Driving; ONTOLOGY-BASED FRAMEWORK; COMPLIANCE CHECKING;
D O I
10.1007/978-3-031-72407-7_17
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Recent advances in language models have facilitated the development of agent-based systems. Despite their encouraging results in various reasoning tasks, these systems often operate as "black boxes", raising concerns about potential illegal behavior due to opaque decision-making processes. This concern is particularly critical in autonomous driving, where precise decision-making requires a thorough understanding of traffic scenes and strict adherence to established norms. In this paper, we propose a legally-guided automated decision making system (LAD) that employs language models to dynamically retrieve facts for related rules through context-based query generation while delegating decisionmaking to a symbolic solver. In our experiments, we demonstrate that this neuro-symbolic system, with a limited number of formalized traffic rules, provides a more accurate, interpretable, and traceable solution for rule-compliant decision-making compared to pure language models.
引用
收藏
页码:234 / 248
页数:15
相关论文
共 50 条
[31]  
Qian C, 2024, Arxiv, DOI arXiv:2307.07924
[32]  
Raz Joseph., 1980, CONCEPT LEGAL SYSTEM, V2d
[33]  
Ribeiro D., 2023, arXiv
[34]   Normative Diagrams as a Tool for Representing Legal Systems [J].
Sasdelli, Diogo ;
Trivisonno, Alexandre Travessoni Gomes .
REVIEW OF SOCIONETWORK STRATEGIES, 2023, 17 (02) :217-231
[35]  
Sun RX, 2024, Arxiv, DOI arXiv:2306.00739
[36]  
Tao J., 2021, ASAM OpenX ontology user guide
[37]   Solving olympiad geometry without human demonstrations [J].
Trinh, Trieu H. ;
Wu, Yuhuai ;
Le, Quoc V. ;
He, He ;
Luong, Thang .
NATURE, 2024, 625 (7995) :476-482
[38]  
w3, 2024, W3C: A semantic web rule language combining OWL and RuleML (SWRL)
[39]  
w3, 2024, W3C: Resource description framework (RDF)
[40]   A survey on large language model based autonomous agents [J].
Wang, Lei ;
Ma, Chen ;
Feng, Xueyang ;
Zhang, Zeyu ;
Yang, Hao ;
Zhang, Jingsen ;
Chen, Zhiyuan ;
Tang, Jiakai ;
Chen, Xu ;
Lin, Yankai ;
Zhao, Wayne Xin ;
Wei, Zhewei ;
Wen, Jirong .
FRONTIERS OF COMPUTER SCIENCE, 2024, 18 (06)