基于光流法的移动机器人避障研究

发布时间：2018-07-12 17:07

本文选题：机器视觉 + 机器人避障　；参考：《华北理工大学》2017年硕士论文

【摘要】：可移动机器人的自主避障功能是其在未知环境中顺利行驶的必备功能,视觉是科学研究中获取场景信息的主要方法。基于视觉传感器的避障方法,可在机器人避障过程中获取比其他避障方法更多的避障信息。在视觉的避障方法中,基于光流法的机器人避障系统是近些年研究的新热点。为了实现移动机器人的自我避障功能,同时提高机器人的避障准确度,实验中将以轮式机器人作为实验平台,并基于HS(Horn-Schunck)光流法设计机器人的自主避障系统,并在真实环境下进行避障实验且验证避障系统的有效性。首先,通过对摄像头拍摄的图像序列进行图像处理,使用的图像处理方法有:图像灰度化、图像滤波和图像缩放。分别使用HS光流算法、LK光流算法和改进的HS算法计算图像序列中的光流场,并在光流场中计算出FOE点和机器人与障碍物的碰撞时间TTC,根据TTC的变化设计机器人的行驶策略。接下来设计机器人的避障系统,建立机器人的避障实验平台,通过串行通信方式实现笔记本电脑到机器人上Sabertooth电机驱动板的指令传输。之后,在真实环境中设计并进行机器人的避障实验,通过Opti Track系统定位并记录机器人的避障数据,检验避障系统的有效性,同时,对传统HS算法与改进HS算法的避障效果进行对比,验证改进算法的有效性。课题中通过实验提出将机器人的旋转角度与光流矢量模ΔF建立关系,使机器人可在不同的ΔF范围内旋转相应的角度,从而优化机器人的避障策略。将权重系数l的取值与机器人的避障效果建立关系,从而提出在一定条件下机器人避障效果最优所对应的权重系数取值范围。
[Abstract]:The autonomous obstacle avoidance function of mobile robot is a necessary function for mobile robot to run smoothly in unknown environment. Vision is the main method to obtain scene information in scientific research. The obstacle avoidance method based on vision sensor can obtain more obstacle avoidance information than other obstacle avoidance methods. Among the obstacle avoidance methods of vision, optical flow based obstacle avoidance system for robot is a new research hotspot in recent years. In order to realize the self-obstacle avoidance function of mobile robot and improve the accuracy of obstacle avoidance, the wheeled robot will be used as the experimental platform in the experiment, and the autonomous obstacle avoidance system will be designed based on HS (Horn-Schunck) optical flow method. An obstacle avoidance experiment was carried out in real environment to verify the effectiveness of the obstacle avoidance system. Firstly, the image sequences taken by the camera are processed, and the image processing methods are as follows: image graying, image filtering and image scaling. The LK optical flow algorithm and the improved HS algorithm are used to calculate the optical flow field in the image sequence respectively. The FOE points and the collision time between robot and obstacles are calculated in the optical flow field, and the driving strategy of the robot is designed according to the variation of TTC. Then the obstacle avoidance system of the robot is designed and the robot obstacle avoidance experiment platform is established. The instruction transmission from notebook computer to the Sabertooth motor driving board on the robot is realized by serial communication. After that, the obstacle avoidance experiment of robot is designed and carried out in real environment. The obstacle avoidance data of robot are located and recorded by Opti track system, and the effectiveness of obstacle avoidance system is tested, at the same time, The effectiveness of the improved HS algorithm is verified by comparing the traditional HS algorithm with the improved HS algorithm. In this paper, the relationship between the rotation angle of the robot and the optical flow vector mode 螖 F is put forward, so that the robot can rotate the corresponding angle in different 螖 F range, so as to optimize the obstacle avoidance strategy of the robot. The relationship between the weight coefficient l and the obstacle avoidance effect of the robot is established, and the range of weight coefficients corresponding to the optimal obstacle avoidance effect of the robot under certain conditions is put forward.
【学位授予单位】：华北理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【参考文献】