Sensing Best-Connected Future Path for Mobile Telerobot: A Radio-Source Location Agnostic Approach

Abstract

Maintaining good last-mile connectivity across the trajectory of a mobile telerobot is critical for ensuring quality of service for multimedia, as well as kinematic commands and sensory feedbacks exchanged between the robot and the remote operator. Hence, the remote operator needs to know the possible radio coverage quality in different future directions. Existing systems claim to achieve this by trying to predict the channel model. The proposed methods require rigorous learning phases and either require prior knowledge of the radio-source location or need to start from the vicinity of the radio-source or need additional sensors in the environment for localization. Given the time-varying nature of the radio-environment and the typical set up in an enterprise, the existing solutions are practically infeasible for live deployment. We propose an in-situ radio-sensing based practically deployable solution with zero-knowledge prediction of the radio-source location and unsupervised prediction of the future connectivity across the periphery of the field of view which does not require any training phase. Deviating from the conventional channel modelling approach, we introduce a novel concept of ‘virtual source’ based prediction error minimization. It is deployed on Double robot in a typical enterprise. The efficacy is proven through objective measures on available data sets and on real enterprise floor and, also, through subjective measures in a typical operational scenario.