摘要
同步地图构建与定位(SLAM)是机器人领域经典问题,也是典型的计算密集型任务,传统实现方法完全依靠机器人本地的计算资源进行实时求解,具有精度低、执行速度慢、计算开销很大缺点。本文针对这一现状,基于云机器人架构和"软件即服务"(Saa S)的思想,提出了基于云计算的SLAM服务框架。该框架能够以黑盒形式为机器人提供SLAM问题求解服务,从而有效降低机器人在同步地图构建与定位过程中的计算开销。本文进一步提出了了这一框架下云端和机器人端之间的数据传输优化解决方案。原型实现和实验表明,通过引入后端云计算基础设施,该框架能够有效提升机器人SLAM问题的求解速度,降低机器人端的能耗,并且将数据传输量控制在合理范围之内。
Simultaneous localization and mapping(SLAM)is a classic problem of robotics,which is also a compute-intensive problem.The implementation of traditional method depends entirely on the robot with its computing resources to get real-time results,which has low accuracy,slow speed of execution and computation overhead.In this paper,based on Cloud Robotics architecture and 'Software As A Service'(Saa S),a SLAM service framework based on cloud computing has been put forward.The framework can serve the robots for SLAM in the form of black box,so as to effectively reduce the computational overhead in the process of SLAM.Furthermore,this paper gives an optimization solution for data transmission between the cloud and robot under this framework.Implementation and experiment shows that with the introduction of the back-end cloud computing infrastructure,the framework can effectively improve the speed of robot SLAM problem,reduce the energy consumption,and control the data transfer volume in a reasonable range.
出处
《软件》
2015年第10期17-20 25,25,共5页
Software
基金
国防科学技术大学校重大应用基础研究项目(ZDYYJCYJ20140601)
国家自然科学基金项目(61201117)
关键词
云机器人
SLAM
云计算
通信优化
Cloud robotics
SLAM
Cloud computing
Communication optimization
