道路交通场景事件的语义解释方法

Semantic comprehension method of road traffic events

针对传统道路交通事件语义解释方法依赖于定量数学建模方法,且底层语义概念与高级事件表达之间存在语义鸿沟的问题,提出了基于动态描述逻辑框架的事件语义定性表达和推理新方法.首先,对静态道路场景信息进行分类标记,构建交通领域本体模型,给出了本体知识库实时修改算法;其次,基于动态描述逻辑的动作公理,提出了描述车辆运动变化的运动模式集,实现了对一定约束条件下车辆机动能力范畴的刻画;最后,将事件语义解释过程抽象简化为目标的实现过程,提出了子目标生成以及实现规则,达到了依据交通事件语义解释交通态势变化过程的目的.实验结果表明:领域本体知识库的定义明确,拥有统一的框架结构,便于理解,且具有通用性;动态描述逻辑良好的表达和推理能力增强了事件语义描述的可靠性,较好地解决了语义鸿沟问题.

For the issues of existing semantic gap between low-level feature and high-level events and the event semantic representation and reasoning overly relying on quantitative mathematical modeling method ,in the paper ,a new approach for event representation and reasoning with dy‐namic description logic is proposed .In the approach ,firstly ,all types of the static road scene in‐formation are classified by employing the domain axioms of dynamic description logic for integra‐ting domain knowledge ontology .And a real-time update algorithm for the domain knowledge on‐tology is given .T hen the action formulary based on action axiom of dynamic description logic is proposed for describing the action categories of vehicle under certain constraints .A rule of sub-goal generation and implementation for the event semantic representation and reasoning are sug‐gested based on the approach to translate event semantic representation into different sub goals . T he application results show that the proposed domain ontology has advantages in clear defini‐tion ,unified framework ,easily understanding and general use .In addition ,the dynamic descrip‐tion logic language with better expressing and reasoning ability increase the reliability of the se‐mantic description of the event ,thus diminishing the semantic gap to a certain extent .