多自由度实验台架的AUV避障模拟研究  

Obstacle avoidance simulation of multi-DOF test platform research for AUV

作　　者：刘和祥[1] 边信黔[1] 范士波[1] 

机构地区：[1]哈尔滨工程大学自动化学院,黑龙江哈尔滨150001

出　　处：《电机与控制学报》2009年第1期143-148,共6页Electric Machines and Control

基　　金：哈尔滨科技创新人才研究专项资金(159040220003)

摘　　要：为了模拟自治水下机器人(AUV)对未知水下环境的感知以及避障过程,采用双目立体视觉系统模拟AUV的前视声呐,并利用自制24 cm×24 cm标定板,对双目立体视觉系统进行了标定。对标定后摄像机的视角进行限制,使之能够鲁棒性地模拟前视声呐的视域。为了提高运算的效率,对获取的图像对进行了校正,建立了校正前后图像的映射关系,利用这种映射关系对校正后的图像对进行立体匹配,并利用匹配结果所得到的目标位置刷新障碍物信息。通过模拟动物的反应式行为,建立了基于规则的AUV避障算法。为了在实验室检验AUV的避障算法,搭建了AUV空间避障模拟平台进行模拟实验。实验结果显示了AUV对未知环境的感知和避障过程,同时验证了算法的有效性及可行性。To simulate perception and obstacle avoidance in unknown underwater environment for autonomous underwater vehicle （AUV） , double-stereo vision system was used to simulate Forward Looking Sonar, and calibrated with 24cm×24cm calibration board. Through limiting calibrating camerag view angle, the system can simulate the view scope of the Forward Looking Sonar robustly. In order to improve calculation efficiently, the images obtained was calibrated, and set mapping relations between the previous and after adjusting images, carried out the stereo match for the after adjusting images by using this mapping relations, and then refurbished obstacle information through the target position which was attained by the matching result. Through simulating animal＇ s reactive behavior, the obstacle avoidance algorithm based on the rules was introduced. To verify the obstacle avoidance algorithm in the laboratory, built the AUV spatial obstacle-avoiding simulating platform, and carried out the modeling test. The simu- lating results show the perception and obstacle-avoiding process of AUV in the unknown underwater environment, and verify that the obstacle avoidance algorithm is practical and effective.

关 键 词：自治式水下机器人 双目视觉 规则法 避障模拟 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]