基于双目视觉的移动机器人避障算法仿真研究

Based on Binocular Vision of Mobile Robot Obstacle Avoidance Algorithm Simulation

针对当前超声波定位的机器人避障技术中,只能感知外围的单一信息,在多机器人通信或者外围存在信号串扰的前提下,信息转化测距结果存在较大误差,存在避障盲区的问题,提出一种视觉多图像信息融合的机器人避障算法,通过双目立体视觉采集机器人的实际图像信息,运用像素自适应融合方法最大程度的采集视觉信息,归一化后的视觉信息带入A*算法进行最优路径的选择,克服单一信息选择中路径计算的弊端,建立多信息约束的路径搜索模型,最大化的解决视觉避障中的干扰问题。后期的计算机仿真结果表明,与传统的测距避障算法相比,有效提高了算法的抗噪性能和避障准确性,为移动机器人避障优化提供了依据。

In view of the current based on ultrasonic positioning of the robot obstacle avoidance method, can only perception peripheral single information, in multi - robot communication or peripheral existing signal crosstalk, under the precondition of information transformation ranging results there is A big error, there exists obstacle avoidance blind area of the problems, the article proposes A kind of multiple image based on visual information fusion robot obstacle avoidance algorithm, through the binocular stereo vision acquisition robot actual image information, the use of pixel adaptive fusion method maximum acquisition visual information, normalization of visual information into the A ＊ algorithm is optimal path choice, overcome A single information in the choice of the disadvantages of path is calculated and establish the multiple information constraint path search model, to solve the maximum visual obstacle avoidance of interference problem. Later the computer simulation exper/ments to prove and traditional rang/ng obstacle avoidance algorithm, the algorithm improve the antinoise performance and obstacle avoidance accuracy, improvement effect is obvious..