An important reason that we could survive as what we are now and could do what we are doing now is because we are living and acting as a societal community. Similarly, when we estimate the potential of robots we should not solely focus our attention on their individual intelligence (which of course is so far infused by humans), but should also take into consideration their sociability (which of course would be initially created by humans).
This would further lead to another philosophical question: what would automatic call determine the sociability of robots? There might be a wide range of arguments on this question. But in term of being able to challenge humans I would argue that the fundamental sociable criteria for robots could be defined as follows:
1) Robots could communicate with each other;
2) Robots could help each other to recover from damage or shutdown through necessary operations including changes of batteries or replenishment of other forms of energy supply;
3) Robots could carry out the manufacture of other robots from exploring, collecting, transporting and processing raw materials to assembling the final robots.
Once robots could possess the above functionalities and start to “live” together as a mutually dependent multitude, we should reasonably view them as sociable beings. Sociable robots could form community of robots. Once robots could function as defined above and form a community they would no longer need to live as slaves of their human masters. Once that happens it would be the beginning of a history that robots could possibly challenge humans or start their cause of taking over humans.
The next question would be: Is the sociability defined above realistic for robots?
Since not all the functionalities mentioned above exist (at least publicly) in this world today, to avoid any unnecessary argument, it would be wise to make our judgment based upon whether any known scientific principle would be violated in any practical attempt to realize any particular functionality among those mentioned above. Communication with other machines, moving objects, operating and repairing machine systems, and exploring natural resources are all among nowadays common practices with programmed machineries. Therefore, even though we might not have a single robot or a group of single robots possess all the functionalities mentioned above, there is no fundamental reason for any of the functionalities mentioned above to be considered as not producible according to any known scientific principle, the only thing left to do would be to integrate those functionalities together onto a single whole robot (and thus a group of single robots).