Research On Ontology-based Situation Assessment And Decision-making Approach For Autonomous Vehicles

Autonomous vehicle can effectively improve traffic safety,reduce pollutant emissions and relieve the pressure of urban traffic congestion,which will have a profound impact on improving the operation efficiency of transpotation system and changing the mode of human travel.The behavioral decision-making system is an important component of autonomous vehicle.As the "brain" of the autonomous vehicle,the advantages and disadvantages of behavioral decision-makinga method will directly affect its driving speed,safety,adaptability and intelligence.It is of great significance to study efficient and reliable behavioral decision-making methods for improving the safety and autonomy of autonomous vehicles,promoting their early industrialization.At present,the following problems exist in the behavioral decision-making method of autonomous vehicle:first,the multi-source heterogeneous information(such as road network elements,traffic participants,traffic rules and driving experience,etc.)existing in the driving scene cannot be effectively expressed and fully utilized.Second,the behavioral decision-making method is not adaptive enough to the driving scene.When faced with a new driving scene,a lot of adjustments need to be made to the control logic.To solve these problems,ontology-based approach for behavioral decision-making is proposed.The detailed research content is as follows:(1)Aiming at the problem that behavioral decision-making knowledge cannot be effectively expressed and fully utilized in driving scenes,this thesis adopts an ontology-based approach to modeling driving scenes.By analyzing the relationship between"human-vehicle-road" in the driving scene and extracting the knowledge of road network,traffic participants and road traffic facilities,the ontology conceptual model of the driving scene is established.After the ontology conceptual model are instantiated by prior road information and environment perception information,the ontology knowledge base described by OWL2 is obtained.(2)Aiming at the problem of rule combination explosion and real time caused by too many conditional attributes,this thesis proposes a deterministic situation assessment method.The main conditional attributes that affect behavioral decision-making are classified and analyzed from the perspectives of safety,legitimacy and reasonableness.In safety assessment,risk measurement indicators TTC and TIV are used to evaluate the influence of surrounding obstacles on the behavior decision-making of Autonomous vehicle.The legitimacy assessment takes into account the speed limit,the traffic light state and the legitimacy of lane changing.The reasonableness assessment mainly considers the global target.The situation assessment generates situation parameters in the horizontal and vertical directions respectively.Situation parameters combined with traffic rules and driving experience,behavioral decision-making rule base is constructed.Based on the prolog online reasoning system,the behavioral decision-making rules are matched with the factual knowledge from the ontology conceptual model,and the decision results are inferred.Finally,the validness and realtime of the method described in this paper is proved by the real-world experiments in different driving scenarios,including straight road driving scenario,lane changing scenario,intersection passing scenario.