The relevant technologies of fixed-point object transfer task for robot in an indoor environment were studied based on robot technology middleware ( RTM) to improve the quality of service and levels of intelligence for intelligent service robot system. Firstly, to improve the stability and development efficiency of system, a method of granularity partition for functional modules was proposed by fuzzy dendrogram clustering algorithm and DS ( Dempster-Shafer ) evidence theory based on the established functional modules set. The dynamic compact hybrid map was built based on the bottom grid map established by Rao-Blackwellized particle filter algorithm to solve the problem of simultaneous localization and mapping in the process of robot fixed point object transfer. Then the memory tracking rules were used to implement fixed-point navigation of mobile robot. To tackle the problem of object transfer, the improved saliency region extraction algorithm was used to extract the coordinates of spatial objects. Then the task of grasping objects were completed by intelligent robot arm platform. The designed system was tested by robot mobile platform and UR5 robot arm which was verified the effectiveness and feasibility by the proposed distributed control strategy.%为提高智能服务机器人的服务质量及智能化水平,基于机器人技术中间件(robot technology middleware,RTM),对室内环境下的机器人定点物体传送任务及其相关技术进行了深入研究.首先,为了提高系统的稳定性及开发效率,建立功能模块集,并结合模糊树图与DS(Dempster-Shafer)证据理论实现模块的粒度划分;对于机器人定点物体传送过程中所面临的即时定位与地图创建问题,采用Rao-Blackwellized粒子滤波算法完成底层栅格地图的创建,进而构造动态精简式混合地图,并利用记忆循迹规则实现机器人定点导航;通过改进的显著性区域提取算法实现对空间物体坐标的提取,并利用智能机械臂平台完成对物体的抓取及传送任务.以机器人移动平台和UR5机械臂为基础对设计系统进行实际测试,结果验证了所提分散控制方法的有效性和可行性.
展开▼
