Abstract: Knowledge,including the bionics,D-S evidence theory,fuzzy mathematics and production rule,is used to study a multisensor information fusion algorithm based on the behavior making-decision.The simulation results show that,the drivers’ experiential knowledge with uncertainty can be effectively expressed,and the drivers’ recognition limitation also can be conquered,and the driving behavior making-decision as well as the vehicle running mode can be simulated and confirmed.In case of the invalidation of some sensors,certain combination property as well as the good fault tolerant capacity also can be ensured.At the same time,the basic theory knowledge is also provided for simulating and realizing the automatic navigation system in the intelligent vehicle.

Key words: driving behavior, multi-resource information fusion, D-S evidence theory, membership function, production rule, recognition distance

摘要： 利用仿生学原理、D-S证据理论、模糊数学知识和产生式规则,研究了一种决策级多传感器信息融合算法。经仿真验证,此算法实时性好,能表示驾驶员的不确定性经验知识,克服其认知局限性,模拟其驾驶行为决策,快速、准确地确定车辆的运行模式,在一个或多个传感器失效的情况下也能保证一定的综合性能,具有良好的容错能力。同时,该研究也为智能车辆自动驾驶系统的仿真及实现提供了基础性理论依据。

关键词: 驾驶行为, 多源信息融合, D-S证据理论, 隶属度函数, 产生式规则, 识认距离