中文摘要：

针对未知环境下的移动机器人实时避障问题，设计了一种基于模板更新和模板匹配的避障算法．利用视觉传感器获取的场景信息与存储于机器人内部的模板图像进行匹配，确定机器人的可行区域和障碍区域．根据虚拟引力方法计算出机器人下一时刻的行进方向和行驶速度．鉴于场景光照变化情况，采用了一种基于粒子滤波的模板更新方法．为了验证算法的正确性和有效性，在室内不同的场景下做了大量的实验．实验结果表明：该方法能够实现可靠的障碍物检测，并引导机器人有效地躲避各种静态和动态障碍物，而且算法具有很好的实时性和鲁棒性．

英文摘要：

To improve the obstacle avoidance of mobile robots in uncertain environments, a real-time obstacle avoidance algorithm based on template matching and updating was proposed. First, scene information was obtained by a vision sensor, and compared with template images stored in the robot. The scenes free space and obstructed space were then confirmed. Secondly, the robot＇s running orientation and speed were determined using the visual attractive force （VAF） method. To deal with variations in the scene＇s illumination, a template updating algorithm based on a particle filter was adopted. In order to test the correctness and validity of the method, experiments were conducted using many types of scenes. Experimental results showed that the method detects and reliably avoids both stationary and moving obstacles. This validates the real-time property and robustness of the algorithm.