Cloud based implementation of ROS through VPN

摘要

In this study, we designed a simulation using Robot Operating System (ROS) to offload one resource intense process - simultaneous localization and mapping (SLAM). A local client located in Malaysia fed laser sensor information to the cloud server hosting the SLAM process located in Singapore. The client and server were placed in the same network using VPN for ROS to operate normally. The average round trip time measured is 18.6540ms with a standard deviation of 23.7870ms. It is feasible to offload SLAM to a cloud server based on the result that we obtained. Extending on the result of our simulation, we conducted a study using a real robot with similar implementation. Based on our result, it is feasible to implement ROS on cloud for SLAM.