MDP及PROLOG在自动驾驶中的应用  

Application of MDP and PROLOG in autopilot

作　　者：班兵 杨志刚 杨航 Ban Bing;Yang Zhigang;Yang Hang(Shaanxi heavy truck Co.,Ltd.,Shaanxi Xi'an 710200)

机构地区：[1]陕西重型汽车有限公司

出　　处：《汽车实用技术》2019年第24期37-40,共4页Automobile Applied Technology

摘　　要：近些年来,自动驾驶迎来了新一轮研究热潮,相关领域技术发展迅速。其中行为决策系统在自动驾驶系统框架中占据重要地位。文章借鉴驾驶员行车过程中视觉行为的注意力分配机制,通过感兴趣区域推理和马尔科夫决策的有机协作生成驾驶动作。最后通过仿真,实现了简单场景下的决策指令生成。In recent years,a new wave of research upsurge on autonomous driving has emerged,and technology of related fields have developed rapidly.The behavior decision-making system plays an important role in the framework of autonomous driving system.In this paper,the attention distribution mechanism of drivers’visual behavior during driving is used for reference,and driving actions are generated through the organic cooperation of region of interest reasoning and markov decision making.Finally,the decision-making instruction generation under simple scenes is realized through simulation.

关 键 词：行为决策 感兴趣区域 马尔科夫决策过程 

分 类 号：U469.7[机械工程—车辆工程]