A survey on Trustworthy AI in Robotic Systems with different levels of autonomy.

What is it about?

This article surveys the Variable Autonomy (VA) robotics literature from the perspective of Trustworthy AI, specifically from the perspective of transparency and explainability. Variable Autonomy robotic systems can change their level of autonomy (from pure teleoperation to semi-autonomy and shared control, all the way to full autonomy). This enables robots to team up with humans and adapt to a variety of scenarios.

Featured Image

Why is it important?

When robots team up with humans or are deployed in demanding applications such as disaster response, remote inspection and maintenance of critical infrastructure, healthcare, dangerous environments, etc, you want them to be trustworthy. Robotic systems that are transparent and explainable to human teammates (e.g. end users and stakeholders) are not only more efficient but they can also be deployed more responsibly. In the context of Variable Autonomy, transparency and expandability allow for meaningful human control over the robots and meaningful collaboration with the robots. For example, clearly knowing who is in control of the robotic system (human or AI) in a given moment.