Thesis

Delay modelling and state estimation techniques for robust telepresence robot navigation

Creator
Rights statement
Awarding institution
  • University of Strathclyde
Date of award
  • 2023
Thesis identifier
  • T16531
Person Identifier (Local)
  • 201569422
Qualification Level
Qualification Name
Department, School or Faculty
Abstract
  • A telepresence system allows to perform remote actions using a telepresence robot over a distance. The human operator controls the movements of the robot by sending control command signals over a communication channel and receives feedback to acknowledge if the telepresence robot has followed the instructions. Telepresence systems recently gained popularity due to their emerging usage in many applications including hospital consultations, remote co-working in offices, security and surveillance, factory inspections or instructor-led education. However, latency constraint introduces major challenges for precise and reliable robotic control in remote environment. Latency (i.e., time delay) can be caused by multiple factors including communication network issues, the physical distance between the human operator and the telepresence robot, processing data and system errors. Time delay also produces a visual mismatch between received navigation state feedback and the actual state of the robot in the remote environment, which negatively impacts the human operator’s performance. This thesis aims to address issues related to latency by proposing new state estimation techniques for robust navigation of telepresence robots and develops an associated framework and a simulation environment. The thesis can broadly be categorised into three main parts, 1) a telepresence framework consists of an off-the-shelf commercial (differential-drive) telepresence robot Beam plus, a multicamera motion tracking system (VICON) and Robot Operating System (ROS); 2) a new state estimation algorithm called Augmented State Extended Kalman Filter (AS-EKF) that compensates time delay; and 3) a simulation environment to reproduce the telepresence system with predictive technology using open-source software RViz and Gazebo. Time delay scenarios are considered for both certain and uncertain cases where the latter were modelled using probability density functions (PDF). The results show significant performance improvements compared to the standard Extended Kalman Filter (EKF) that does not consider delays. The simulation framework offers wider adaptability when a physical system is not plausible, and a controlled experimental environment is desired.
Advisor / supervisor
  • Dobie, Gordon
Resource Type
DOI
Funder

Relazioni

Articoli