基于kinect传感器进行手势控制X80机器人移动

发布时间：2018-07-16 08:31

【摘要】：人机交互(HCI)技术是研究用户与计算机之间的通信交互技术,是一个决定系统是否“友好”的重要因素。人机交互的原理是利用人体自身的各种动作及细微变化,与计算机系统进行交互。这种崭新的互动形式,极大地提高了交互的识别精准度和人性化。手势作为一种常见的交流方式,在人与机器的交互领域里扮演着越来越被广泛使用的角色,如Xbox、智能手机以及广告板等等。利用手势识别控制机器人不仅允许用户通过自然的手势动作操纵远程机器人,也保证了用户可以“友好”的与发展中的网络机器人系统进行交互。现如今,随着机器人科学的不断进步,机器人拥有越来越强大的生命力,而移动机器人的智能运动技术是研究机器人的一个重要课题。如何在一个自然环境中以非接触式的手势交互,与机器人之间进行指挥运动是本文研究的主要重点。本文利用kinect传感器和X80移动机器人进行手势控制机器人移动的研究。具体包括:(1)基于kinect传感器的工作机制和设备的操作原理,设计了5种自然交互的手势模板,包括前进、后退、向右转动、向左转动和停止。(2)通过对比隐马尔科夫模型算法和动态时间规整算法的优缺点,选出一种适合本文研究需求的相对高效的算法——动态时间规整算法,并通过限制全局路径和失真度阈值进行优化。利用优化后的动态时间规整算法验证上述5种手势识别的准确性。(3)基于PID原理以及无线机器人软件开发工具包,依据手势识别的结果,解码器输出对应的变量,从而控制无线机器人电机的运转。(4)利用kinect识别出的手势控制X80机器人的前后左右运动。
[Abstract]:Human computer interaction (HCI) technology is an important factor to study the communication and interaction between users and computers, which determines whether the system is "friendly" or not. The principle of human-computer interaction is to interact with computer system by using various actions and subtle changes of human body itself. This new form of interaction greatly improves the recognition accuracy and humanization of interaction. As a common way of communication, gestures play more and more popular roles in the field of interaction between human and machine, such as Xbox, smart phone, billboard and so on. The gesture recognition control robot not only allows the user to manipulate the remote robot through natural gesture action, but also ensures that the user can interact with the developing network robot system in a "friendly" way. Nowadays, with the development of robot science, robot has more and more powerful vitality, and the intelligent motion technology of mobile robot is an important subject to study robot. How to communicate with robots by non-contact gestures in a natural environment is the main focus of this paper. In this paper, kinect sensor and X 80 mobile robot are used to control the movement of robot. The main contents are as follows: (1) based on the working mechanism of the kinect sensor and the operation principle of the device, five kinds of natural interactive gesture templates are designed, including forward, backward, right turn. Turn to the left and stop. (2) by comparing the merits and demerits of the hidden Markov model algorithm and the dynamic time warping algorithm, a relatively efficient algorithm named dynamic time warping algorithm, which is suitable for the research of this paper, is selected. The optimization is carried out by limiting the global path and the distortion threshold. Using the optimized dynamic time warping algorithm to verify the accuracy of the above five gesture recognition. (3) based on pid principle and wireless robot software development kit, according to the result of gesture recognition, the decoder outputs the corresponding variables. In order to control the operation of the wireless robot motor. (4) the movement of the X80 robot is controlled by the gesture recognized by kinect.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP391.41;TP242

【相似文献】