Compensating Delays for Precise and Real-time ROS Cloud Robotics Localization

Today high speed 5G networks offer almost negligible delay to reach the cloud infrastructure, however, in the field of cloud robotics, we can still find key components that suffer from latency. While in general, sensor signals benefit from fast transmission, image-based localization accuracy could still be limited due to delayed image processing. The solution presented in this paper uses the available real-time sensory data to improve the accuracy of the image-based localization by providing location predictions in moments when the data from the image is not yet available. We created a solution extending the localization capabilities of the Robot Operating System (ROS). We tested more types of sensory data, including feedback from the control messages, and used artificial intelligence algorithms to make the prediction.