中文摘要：

面向在地面搜索地下气味源的任务,针对传统六边形路径算法步长固定且无法对气味源定位的不足,提出移动机器人依靠气体传感器的基于自身行为特征的变步长气味源搜索算法。机器人根据其自身在搜索过程中所表现出的行走特征,采用行为控制模型,选择与具有不同特征的行为相匹配的步长调整策略。步长调整策略包括：根据实测所得即时气味浓度变化率的动态步长改变策略、根据徘徊程度确定的动态步长缩短策略、固定步长策略。非均匀土壤且实际扩散情况下的计算机仿真证实,改进算法具有气味发现、气味跟踪和气味源定位三种功能。

英文摘要：

To perform the task of searching underground odor source, on the basis of traditional hex-path algorithm, the behavior-feature-based variant step-size algorithm for mobile robot was provided. According to its behavior feature in search process, a robot used behavior control model to determine what step size strategy should be selected. The step size strategies include three kinds： dynamic-varying based on instant odor density, dynamic-decreasing based on wandering extent, and invariant. Computer simulation work proves the new algorithm has all three functionalities of complete odor source search work： odor finding, odor tracing and odor source localizing.