Many complex tasks such as search and rescue, explosive ordinance disposal, etc. that eventually will be performed by autonomous robots are still performed by human operators via teleoperation. Since the requirements of teleoperation are different from those of autonomous operation, design decisions of teleoperation platforms can make it difficult to convert such platforms for autonomous use. In this paper, the authors discuss the differences and the potential difficulties involved in this conversion, as well as strategies for overcoming them. They demonstrate these strategies by "autonomizing" a fully teleoperated robot designed for tasks using an autonomous architecture that has the requisite capabilities.
@inproceedings{kievitkylaretal12icai,
title={From Teleoperation to Autonomy: "Autonomizing" Non-Autonomous Robots},
author={Brent Kievit-Kylar and Paul Schermerhorn and Matthias Scheutz},
year={2012},
booktitle={Proceedings of the 12th International Conference on Artificial Intelligence},
url={https://hrilab.tufts.edu/publications/kievitkylaretal12icai.pdf}
}