基于AmigoBot的群控制问题研究

发布时间：2018-05-31 15:52

本文选题：群控制 + 栅格地图　；参考：《济南大学》2017年硕士论文

【摘要】：伴随着德国工业4.0时代的到来,机器人产业进入了井喷式的发展时期。一方面,生产制造的领域的主力局变成了工业机器人,现如今的汽车产业、电子制造产业都在大规模的使用各种机器人。另一方面,谷歌、百度等互联网公司依靠先进的互联网技术开始涉足机器人行业,谷歌公司加大投入对无人车编队的研发,大疆公司对无人机的群控制研究投入也日渐增加。大数据的使用使机器人之间实现了网络化和智能化,智能机器人已经成为了机器人新的代名词。本文以美国ActivMedia Robotics公司AmigoBot机器人作为研究平台,研究了多个机器人在无障碍环境下群集控制策略以及在有障碍的环境中的群集控制策略。论文的主要研究工作包括以下几个方面。首先,采用栅格地图构建了机器人的环境模型,并且对机器人声纳传感器的置信度进行了研究。设计了置信度的蒸发和增强机制,以此模拟群集机器人对环境的认知,从而改善了机器人在有障碍的环境中的群控制策略的性能。接着,研究了多机器人的群集控制策略。设计了一个人工势能函数来评价相邻的两个机器人的相对位置关系,对于某一时刻,每个机器人的总势能即描述了当前群集形态的特性。然后利用Newton法来搜索机器人的最优运动方向,使机器人都能够形成期望的群集形态。设计了无领航者和有领航者的群集控制模型,通过仿真结果来验证设计的群集控制策略的有效性。最后,研究了多机器人在有障碍的环境中的群控制策略。机器人利用声纳传感器感知障碍物信息,并采用有置信度的栅格地图来建立环境模型。文中设计了机器人与障碍物之间的排斥力模型,用以控制机器人的运动方向来避开障碍物。将Newton方法与排斥力相结合,实现了机器人在有障碍的环境中的群集控制策略。
[Abstract]:With the arrival of German industry 4.0, the robot industry has entered the period of blowout development. On the one hand, the main field of manufacturing has become industrial robots, today's automobile industry, electronic manufacturing industry are using a variety of robots on a large scale. On the other hand, Google, Baidu and other Internet companies rely on advanced Internet technology to start to dabble in the robotics industry. Google has increased its investment in research and development of unmanned vehicle formations, and Dajiang has increasingly invested in research on group control of drones. With the use of big data, the network and intelligence of robots have been realized, and intelligent robots have become a new synonym for robots. In this paper, the cluster control strategy of multiple robots in a barrier-free environment and a cluster control strategy in an obstacle environment are studied on the platform of AmigoBot robot of ActivMedia Robotics Corporation of the United States. The main research work includes the following aspects. Firstly, the environmental model of robot is constructed by grid map, and the confidence degree of robot sonar sensor is studied. The evaporation and enhancement mechanism of confidence degree is designed to simulate the cognition of swarm robot to environment and improve the performance of swarm control strategy in obstacle environment. Then, the cluster control strategy of multi-robot is studied. An artificial potential energy function is designed to evaluate the relative position of two adjacent robots. For a given time, the total potential energy of each robot describes the characteristics of the current cluster configuration. Then the Newton method is used to search the optimal motion direction of the robot, so that the robot can form the desired cluster shape. A cluster control model without and with a pilot is designed, and the effectiveness of the designed cluster control strategy is verified by simulation results. Finally, the swarm control strategy of multi-robot in obstacle environment is studied. The robot uses sonar sensor to perceive obstacle information and uses grid map with confidence to build environment model. The repulsive force model between robot and obstacle is designed to control the motion direction of robot to avoid obstacles. The Newton method is combined with the repulsive force to realize the cluster control strategy of the robot in an obstacle environment.
【学位授予单位】：济南大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【参考文献】