狭窄环境地形图建立的融合算法研究

针对移动机器人使用超声波传感器在已知狭窄环境和未知狭窄环境中的导航问题,提出一种多层融合算法,该算法通过在数据层融合中使用D-S证据理论,在决策层融合中使用意见一致性理论,建立已知狭窄环境的地形图;对于未知狭窄环境,使用D-S证据理论中的矛盾因子,建立了一个自适应超声波传感器模型.结果表明:使用上述方法有效地减少了超声波传感器镜面反射所引起的不确定性,提高了移动机器人使用超声波传感器在狭窄环境中建立地形图的准确性,证明本文所提出的方法是有效可行的.

Fusion algorithms for building narrow environment maps

Aiming at the navigation with ultrasonic sensors for mobile robots in known and unknown narrow environment, two fusion algorithms were proposed respectively. The known environment map was built by adopting a multi-layer fusion algorithm, in which D-S evidence theory was applied to data-level fusion and consensus theory was applied to decision-level fusion. A self-adaptive ultrasonic sensor model with the conflict factor of D-S evidence theory was proposed for navigating in unknown narrow environment for mobile robots. The uncertainty caused by specular reflections was effectively reduced by employing the fusion algorithms above. And the accuracy of building maps in narrow environments was improved. The fusion algorithms were proved to be effective and practical by experiments.