Question: Do you think that robots should have installed restrictions regarding the freedom they have over their control of tasks that could have dire consequences if done wrong?

Giuseppe Cotugno answered on 3 Jan 2020:

In my personal opinion, it is a bit premature to consider this case yet. Robots are machines which are built and optimised to perform a specific task, at present assumptions should be made on the nature of the task and the environment the robot operates in since a lot of issues in robotics are still unsolved. Practically, in a real world application, I think it is unlikely that such limitations have to be built in the robot itself.

However, if we extend the horizon to research prototypes, there are cases when the robot has to have limitations over its behaviour. Typically, this is the case when robots are interacting with human beings directly, for example when handing over tools.

A robot which is interacting with humans must have some sort of self-imposed limitations in order to be safe to work with. For example, it could instantaneously stop if an accidental contact is made with a person. There is a whole discipline (human-robot interactions) which studies how a robot can be made safe to work with a human and I would expect this to become more and more of a mandatory feature in future, as robots became more capable of doing things.

In general, I would expect to see some self-imposed limitations in the physical behaviour of the robot, because it might be moving so quickly that there might not be time for a human supervisor to react and stop it. In my personal opinion, for anything else related to decision making that might have serious consequences if performed incorrectly, then probably it is best to have a human involved.